GEOSPATIAL MODELING OF URBAN BUILDINGS AND LAND USE FOR

CLIMATE CHANGE IMPACTS AND RESOURCE PRODUCTIVITY

A Dissertation Presented

By

Mithun Saha

To

The Department of Civil and Environmental Engineering

In partial fulfillment requirement

for the degree of

Doctor of Philosophy

In the field of

Environmental Engineering

Northeastern University

Boston, Massachusetts

August 2016

ii

ACKNOWLEDGEMENT

First of all, I cannot express enough thanks to my committee for their continued support

and encouragement: Dr. Auroop Ganguly, Dr. Matthias Ruth, and Dr. James Connolly.

My completion of this dissertation could not have been accomplished without the support

of my academic advisor and mentor. I would like, to thank Dr. Matt Eckelman, for being a

great mentor and colleague throughout every step of the course of my doctoral journey. I

am indebted for his help and support, and for sharing his extensive knowledge and very

contagious passion for our research.

I am also grateful to Professor David Fannon, for his invaluable help and advice.

Appreciations to my former colleagues in NU sustainability research engineering team, Dr.

Pei Zhai, Dr. Leila Pourzahedi, and soon to be Dr. Mahdokht Montazeri for all their helpful

discussions, suggestions, and friendship. Also, thanks to all my friends for encouragement.

In addition, I would like to thank departmental staff member Mr. Michael Macneil for

providing continuous logistical and technical support for conducting doctoral research.

Finally, thanks to the most important people in my life, my family for all their love and

support. My parents who are living 12,500 km far from me and have been waiting

relentlessly to see their elder son be a holder of doctorate someday. Last but not the least,

there is one person whom I can’t thank enough and therefore just want to dedicate this

work, the love of my life, my beautiful wife, Susmita Biswas.

iii

ABSTRACT

Urbanization is expected to continue, with more than two-thirds of the world’s population

likely to live in urban areas by 2050, leading to a net urban influx of approximately 2.5

billion people. Existing infrastructure must be equipped to address this dramatic urban

growth while also adapting to potential adverse impacts of climate change and other natural

hazards. To be sustainable, cities must themselves, become efficient users of materials and

energy as well as respond to future climatic conditions. Two main urban engineering

strategies are to map how current stocks may respond to climate change and to identify

resources that could be used to improve local productive capacity and reduce dependencies

on distant resources. The dissertation herein addresses these two overarching strategies

through a series of specific case studies for the Boston area using GIS based urban stock

assessment as a framework.

GIS is used extensively in urban system modeling and resource assessment. The geospatial

modeling presented in this dissertation involves the combination of spatial and remote

sensing techniques in a way that multi-dimensional, location-based data can be analyzed

and visualized to assess urban resources at building and sub-parcel-level resolution.

Corrosion models are coupled to climate change projections to estimate the durability of

all concrete buildings throughout the city subject to enhanced carbonation and chlorination

processes. Climate change will decrease the time to corrosion initiation in new concrete

buildings by 10-26 years. Scalable spatial frameworks and models for assessing urban

biomass potential are presented and applied to across urban land-use and building types, to

iv

address questions of self-sufficiency for local food and bioenergy systems. Through a

series of studies, up to 24% of Boston’s land area and up to 20% of the entire metropolitan

area could be plausibly utilized for bioenergy crops, while 17% of Boston’s area could be

used for urban agriculture, potentially supplying up to 30% of all fruits and vegetables

consumed in the city. Results at the sub-parcel-level have direct utility for recent

government initiatives. This work advances the field of urban engineering through

application of novel coupled geospatial-biophysical analyses to entire urban regions at high

resolution.

v

RESEARCH PRODUCTS

Journal publications:

Saha, M., Eckelman, M. (2015) Geospatial assessment of potential bioenergy crop

production on urban marginal land. Applied Energy, 159, 540-547

Saha, M., Eckelman, M. (2014) Urban scale mapping of concrete degradation

from projected climate change. Urban Climate, 9, 101-114

Publications in preparation:

Saha, M., Eckelman, M. Geospatial assessment of urban agriculture potential in

Boston

Saha, M., Eckelman, M. A GIS-based Assessment of Regional Scale Bioenergy

Production Potential on Marginal and Degraded Land

Conference proceedings:

Saha, M., Martin, L., Amidon, J., Ruth, M., Eckelman, M. (2015) 4d-Mapping of

urban biomass production for food and fuel, ISIE Conference, July 07-10,

Guildford, Surrey, UK

Saha, M., Martin, L., Amidon, J., Ruth, M., Eckelman, M. (2015) Mapping urban

biomass production of food and fuel, ISSST Symposium, May 18-20, Dearborn,

Michigan

vi

TABLE OF CONTENTS

LIST OF FIGURES ..................................................................................................... ix

LIST OF TABLES ....................................................................................................... xi

NOMENCLATURE ................................................................................................... xii

Chapter 1 Introduction to Urban Metabolism and GIS Application .......................1

1.1 Background ...............................................................................................................1

1.1.1 Urban metabolism ...............................................................................................1

1.1.2 Urban stock analysis ...........................................................................................4

1.2. Introduction to geospatial assessment ........................................................................5

1.2.1 Geographic information system (GIS) .................................................................6

1.2.2 GIS applications ..................................................................................................9

1.3. GIS-based urban stock assessment .......................................................................... 11

1.4. Motivation and research objectives ......................................................................... 14

1.5. Dissertation Structure.............................................................................................. 16

Chapter 2 Urban Scale Mapping of Concrete Degradation from Projected Climate

Change ......................................................................................................................... 20

2.1 Introduction ............................................................................................................. 21

2.2 Climate Change Scenarios ....................................................................................... 25

2.2.1 Atmospheric CO2 Concentrations ...................................................................... 25

2.2.2 Temperature Predictions.................................................................................... 27

2.3 Corrosion ................................................................................................................. 28

2.3.1 Concrete Carbonation Modeling ........................................................................ 28

2.3.2 Concrete Chlorination Modeling ....................................................................... 31

2.4 Spatial Analysis ....................................................................................................... 33

2.5 Results ..................................................................................................................... 35

2.5.1 Carbonation and Chlorination Depth. ................................................................ 35

2.5.2 Geospatial Results ............................................................................................. 38

2.6 Discussion ............................................................................................................... 40

2.6.1 Implications for Current Code Requirements for Concrete Cover ...................... 40

2.6.2 Concrete Technologies for Climate Change Adaptation..................................... 42

2.6.3 Future Research Needs ...................................................................................... 43

2.7 Conclusion ............................................................................................................... 44

vii

Chapter 3 Geospatial Assessment of Potential Bioenergy Crop Production on

Urban Marginal Land ................................................................................................. 46

3.1 Introduction ............................................................................................................. 47

3.2 Methods ................................................................................................................... 52

3.2.1 Land use type screening .................................................................................... 53

3.2.2 Bio-geophysical screening................................................................................. 54

3.2.2.1 Exclusion of parcels by soil quality and slope ................................................. 54

3.2.2.2 Exclusion of areas by shadow analysis ........................................................... 55

3.2.3 Biomass and bioenergy yield ............................................................................. 57

3.3 Results ..................................................................................................................... 58

3.3.1 Marginal land resources in Boston .................................................................... 58

3.3.2 Biomass and bioenergy potential ....................................................................... 61

3.3.3 Spatial Validation.............................................................................................. 63

3.4 Discussion and Implications..................................................................................... 63

3.5 Conclusions ............................................................................................................. 66

Chapter 4 A GIS-based Assessment of Regional Scale Bioenergy Production

Potential on Marginal and Degraded Land ................................................................ 68

4.1 Introduction ............................................................................................................. 69

4.1.1 Regional assessment of marginal land ............................................................... 69

4.2 Methods ................................................................................................................... 71

4.2.1 Marginal Land Assessments .............................................................................. 71

4.2.2 Biomass and bioenergy yield ............................................................................. 74

4.3 Results and Discussion ............................................................................................ 75

4.3.1 MAPC Marginal land ........................................................................................ 75

4.3.2 Biomass and Bioenergy Yield ........................................................................... 78

4.4 Conclusion ............................................................................................................... 80

Chapter 5 Geospatial Assessment of Urban Agriculture Potential in Boston ........ 82

5.1 Introduction ............................................................................................................. 83

5.1.1 Urban agriculture .............................................................................................. 83

5.1.2 Spatial analysis ................................................................................................. 85

5.1.3 Connections to self-sufficiency and resilience ................................................... 88

5.2 Methods ................................................................................................................... 91

5.2.1 Study area and datasets ..................................................................................... 91

viii

5.2.2 Mapping flat rooftops ........................................................................................ 92

5.2.3 Mapping ground level parcels ........................................................................... 96

5.2.4 Estimating food yields ..................................................................................... 100

5.2.5 Validation ....................................................................................................... 101

5.3 Results ................................................................................................................... 102

5.3.1 Rooftop area mapping ..................................................................................... 102

5.3.2 Ground level farmland mapping ...................................................................... 104

5.3.4 Food yield potential ........................................................................................ 106

5.3.5 Validation ....................................................................................................... 107

5.4 Discussion and Implications................................................................................... 108

5.5 Conclusions ........................................................................................................... 111

Chapter 6 Conclusion and Future Works .............................................................. 113

REFERENCES .......................................................................................................... 118

APPENDIX ................................................................................................................ 138

ix

LIST OF FIGURES

Fig. 1-1 The urban metabolism of Toronto Port Land………………………… 3

Fig. 1-2 Feature type and their representations in two spatial data models…… 7

Fig. 1-3 A spatial database for urban mining studies…………………………... 8

Fig. 2-1 Predicted estimates of CO2 concentrations…………………………… 27

Fig. 2-2 Predicted mean temperature increase for Boston’s Logan Airport

station for A1FI and B1 scenarios ……………………………………

28

Fig. 2-3 Concrete structures (buildings, in red) within Boston (outlined in gray) 34

Fig. 2-4 Estimated carbonation depth (mm) in BMA for a building constructed

in 2000………………………………………………………………...

37

Fig. 2-5 Estimated chlorination depth (mm) in BMA for building constructed

in 2000………………………………………………………………...

37

Fig. 2-6 a) Location of gridded climate data, b) Climatically vulnerable zones

within Boston………………………………………………………….

39

Fig. 2-7 a) Concrete Structures classified according to different age, b) % of

buildings with compromised cover thickness over the service life……

40

Fig. 3-1 Flowchart of modeling processes used for biomass mapping and

bioenergy assessment…………………………………………………

53

Fig. 3-2 Shadow analysis examples: a) input extruded-2D building footprint

and b) shadow map……………………………………………………

56

Fig. 3-3 Marginal land estimation in Boston…………………………………... 60

Fig. 3-4 Available marginal land in Boston…………………………………… 61

Fig. 3-5 Randomly selected urban marginal land parcels (Scale 1: 600) …….... 63

Fig. 4-1 Study area (MAPC region) …………………………………………… 71

Fig. 4-2 Urban marginal land estimation………………………………………. 74

Fig. 4-3 Available marginal land in MAPC cities……………………………... 77

Fig. 4-4 Available marginal land in four municipalities………………………. 78

Fig. 5-1 Boston rooftops and ground level parcels extraction steps…………… 92

x

Fig. 5-2 a) Delineated buildings, b) The division of the Boston area for LiDAR

data……………………………………………………………………

94

Fig. 5-3 a) Boston downtown geographic area of interest; b) High resolution

LiDAR data; c) Extracted building footprint layer……………………

96

Fig. 5-4 Shadow distribution on Boston Downtown at a) 10 AM and b) 4 PM

on July 21st…………………………………………………………….

100

Fig. 5-5 Flat rooftop mapping steps…………………………………………… 102

Fig. 5-6 Flat Roofs in Boston………………………………………………….. 103

Fig. 5-7 a) Flat roof distribution in Boston’s neighborhoods, by; b) overlaid

aerial image of flat roofs in the Dorchester neighborhood ……………

103

Fig. 5-8 Ground level farmland estimation in Boston…………………………. 105

Fig. 5-9 Available ground level areas in Boston………………………………. 106

Fig. 5-10 Randomly selected rooftop (top row) and ground level (bottom row)

parcels…………………………………………………………………

108

xi

LIST OF TABLES

Table 2-1 Structural durability conditions for concrete used in this study ……... 30

Table 2-2 Recommended concrete properties for corrosion protection ………… 32

Table 2-3 Minimum cover (mm) required to counteract the impact of climate

change on carbonation-induced corrosion damage risks by 2100…….

38

Table 2-4 Minimum cover (mm) required to counteract the impact of climate

change on chlorination-induced corrosion damage risks by 2100…….

38

Table 3-1 Review of studies on marginal land assessment in USA ……………. 49

Table 3-2 Description of parameters used to estimate urban marginal land……. 54

Table 3-3 USDA-NRCS Marginal soils classification………………………...... 55

Table 3-4 Energy-crop yield and heat content information …………………...... 58

Table 3-5 Biomass and bioenergy yield ………………………………………... 62

Table 4-1 Description of parameters used to estimate marginal land area……… 73

Table 4-2 Energy-crop yield and heat content information ……………………. 75

Table 4-3 MAPC marginal land ………………………………………………... 78

Table 4-4 Biomass and bioenergy yield ………………………………………... 79

Table 5-1 Review of studies on urban agriculture potential in North America…. 85

Table 5-2 Description of parameters used to estimate rooftop area……………. 93

Table 5-3 Description of parameters used to estimate ground level area ………. 98

Table 5-4 USDA-NRCS soils classification ……………………………………. 98

Table 5-5 Energy-crop yield and heat content information …………………...... 101

Table 5-6 Potential food production in Boston …………………………………. 107

xii

NOMENCLATURE

ARRM Asynchronous Regional Regression Model

AOGCM Atmospheric-Oceanographic Global Climate Models

ACI American Concrete Institute

BAU Business as Usual

BETYdb Biofuel Ecophysiological Traits and Yields database

CCSM Community Climate System Model

CHP Combined Heat and Power

DSM Digital Surface Model

EBI Energy Bioscience Institute

GFDI Geophysical Fluid Dynamics Institute Model

GHG Greenhouse Gas

GIS Geographic Information System

HADCM2 Hadley Climate model

HHV High Heating Value

IPCC Intergovernmental Panel on Climate Change

LHV Low Heating Value

LiDAR Light Detection and Ranging Data

Mi Marginal land parcel area (ha)

MAGICC Model for Assessment of Greenhouse-gas induced Climate Change

MAPC Metropolitan Area Planning Council

MATEP Medical Area Total Energy Plant

xiii

NAD North American Datum

NCAR National Center for Atmospheric Research

PCM Parallel Climate Model

SFA Stock and Flow Analysis

SRDBMS Spatial Relational Database Management System

SRES Special Report on Emissions Scenarios

SSURGO Soil Survey Geographic Database

TMDL Total Maximum Daily Load

USDA US Department of Agriculture

UTM Universal Transverse Mercator

WGS World Geodetic System

xc Carbonation depth (mm)

xcl Chlorination depth (mm)

1

Chapter 1 Introduction to Urban Metabolism and GIS Application

1.1 Background

1.1.1 Urban metabolism

The global footprint of cities is increasing, with more than 50% of world’s population

currently living in cities and peri-urban areas [1]. Urban populations consume the majority

of the world’s resources and contribute towards environmental deterioration both locally

and globally [2]. Modern cities are the epicenter of resource consumption [3]. While they

aim to improve resource efficiency and reduce (particularly local) environmental burdens,

cities must also adapt to global climate change and be prepared for natural calamities.

Several examples portend the challenges ahead. In 2012, in the aftershock of Hurricane

Sandy, nearly three-quarters of gas stations in New York City remained inoperable, leading

to gasoline rationing [4]. During winter 2015, areas of New England received record

snowfall of nearly 110 inches that left thousands of people without power for multiple days

[5].

Extremes events can severely curtail the functioning or urban systems, with significant

implications for shelter, energy, and food security, three essential pillars of human survival.

In a globalized world, cities are increasingly reliant on global supply chains for energy and

food, while local capacity to provide material resources to meet the needs of residents has

dwindled (with the exception of water) [6]. Many cities are reassessing their self-

sufficiency as one strategy to prepare for the challenges that climate change will bring.

2

While developed as a method of quantifying urban systems processes of different

dimension and spatiotemporal variability, the field of urban metabolism has gained

importance both among scientific and urban planning and management communities.

Urban metabolism considers all physical and techno-economical processes that originate

from or flow within an urban ecosystem [3, 4]. Practically, urban metabolism investigates

both material and energy flows across the urban boundary and stocks accumulation within

cities [4]. In general, more research has focused on flow analysis with the ideal of

integrating flows among different actors, so-called industrial symbiosis. For example,

Kennedy et al. examined the Port Lands area of Toronto, mapping out both resource inputs

and emissions, and waste streams that were or could be reused (Fig. 1-1).

Previous studies have also discussed possible synergy of urban metabolism in assessing

urban resource self-sufficiency. According to Sharifi and Yamagata. 2016 [6], “a

sustainable and self-reliant city should be capable of sustained disruptions by leveraging

its resources for material, energy and food production”. This will help cities thriving while

minimizing environmental impact and avoiding great harm to life and devastation to

property. In this way, cities are often comparable to an ecosystem [7]. Many studies

conducted on urban metabolism have been sector specific and focused on issues such as

material [8] and energy [9]. At the same time, a number of studies have focused on

developing criteria and indicators for assessing urban stocks and flows in different domains

[8, 10-13].

3

In summary, the studies described above indicate that it may not be possible for cities to

be self-sufficient regarding developing its resources and infrastructure to produce all

energy and materials locally, but knowing the urban productive capacity provides an

opportunity to plan for various future scenarios [7]. Every region has a unique set of

resource demands, stocks, and climate vulnerabilities based on geographic location [14].

Past research has called for site and context specific assessment as imperative to understand

which stocks are most vulnerable to change and which resources can potentially be used to

reduce cities’ reliance on global resources [8, 14, 15].

Fig. 1-1: The urban metabolism of Toronto Port Lands [16]

4

1.1.2 Urban stock analysis

As a sub-field of urban metabolism, urban stock analysis is the assessment of stored and

internal urban resources [4, 7, 17]. Stocks can materials and energy that are readily

accessible for use, such as diesel fuel, or ‘hibernating’ stocks that are not in use but still

exist within the city boundary, such as steel in abandoned railroad lines. A bottom-up

approach is common for stock estimation [18], where individual materials are located and

accounted for essentially as an inventory [19]. Stocks can also be assessed using a top-

down approach by conducting system material and energy balances [19]. This approach

includes an indirect measurement of stock’s size by observing a time-dependent flows

through the stock system. With the top-down approach, the estimated material stocks are

measured at the national or regional level and then scaled down to the local level [17],

which may introduce a large uncertainty bound for the spatial distribution [18].

A city’s stocks of buildings and land make up its local resource base. From a self-

sufficiency perspective, it is crucial to develop a framework to quantify these resources,

utilize them efficiency, and minimize any adverse local impacts on the urban environment

[20]. For this purpose, the amounts of resources available for urban activities is necessary

to quantify. In this context, SFA (Stock and Flow Analysis) was developed to quantify the

amount of material or energy that is harvested from nature and its life cycle [16].

So far, numerous studies investigated material stock and flow analysis (MSFA) of different

materials. Gordon et al. [21], Hashimoto et al.[22], and Schandl and West [23] are

5

representative studies that leveraged MFSA at national, regional, and local scale. MFSA is

useful for conducting a high-resolution analysis in the space-time scale. For example, a

historical visualization of the cityscape is performed by building a 4d-GIS model where

spatial data contain information about height (3d-GIS) and are arranged in a chronological

order over a defined time period [7].

On the basis of discussion above, it is evident that bottom-up stock and flow analyses can

be conducted with high-resolution spatiotemporal data. Quantifying existing urban stocks

and combining with historical data or future scenarios can provide estimates of the current

physical quality of existing stocks, their capacity for productive activities (to produce

resource flows), or how they may degrade over time, all of which can help to support local

policy making [19].

1.2. Introduction to geospatial assessment

The geospatial assessment involves the combination of geographical information system

(GIS) and remote sensing technique in a way that multi-dimensional location-based data

can be entered, checked, analyzed and visualized as data referenced to the earth [24].

Location-based data are also called georeferenced data or, more commonly, as spatial data.

GIS are scale-independent and can be used for examining, exploring and analyzing spatial data

at global, continental, regional, and, local scales [25]. Therefore, GIS can be a useful tool for

characterizing and visualizing spatial distributions of infrastructures and resources (energy and

food) stocks and flows across the urban environments. Therefore, GIS-based location

intelligence can be extremely effective for informing the policy makers and the broader

6

community with accurate and comprehensive information. The following section reviews the

theoretical concept and functionality of GIS and studies the possible role of GIS in urban stocks

assessment. Additionally, it describes how GIS can be used to measure, analyze, and visualize

the geographical characteristics of stored and internal stocks. Finally, the section discusses the

issues and challenges in the use of GIS for urban designers and engineers.

1.2.1 Geographic information system (GIS)

In general terms, GIS is a computer system that can be considered as “a special case of

information systems where the database consists of observations on spatially distributed

features, activities or events, which are defined in space as points, lines, or areas” [25]. A GIS

manipulates data about these points, lines, and areas to retrieve data for ad hoc queries and

analyses” [26]. GIS data are characterized using either a vector or raster data model. The vector

model consists of points, lines, and polygon features plotted as coordinates in space. Whereas,

a raster data model is conceptualized as a continuous gridded surface of equal cell size (Fig. 1-

2). Virtually all current GIS-based software packages are capable of handling both vector and

raster layers. The data layer preference is based on the model appropriateness for specific

phenomena. Using a spatial tool involves capturing the spatial distribution and pattern of

features through the measurement of cartographic representation. Urban stocks and other

related infrastructures are all spatially distributed and, so, can be studied using GIS.

The GIS system performs three specific types of functionality with increasing convolution [27].

The first category is spatial visualization or cartographic representation of location data. It is

an essential function of GIS. A map is a simple form of spatial visualization and representation

of spatial data. GIS maps are in digital form and named map layer which is a set of spatial data

7

containing the attributes of the features (e.g., material and energy stocks) and their geographical

locations. The geographic locations are recorded as x, y coordinates with a specific coordinate

system or as latitude and longitude. A GIS system provides a comprehensive set of symbols

and colors for users to choose to symbolize layer attributes. Cartography preparation and

geoprocessing analysis are not new, but GIS system performs this more precisely than other

methods. Current GIS technology provides more user interface with 3D spatial analysis and

automatic model builder capability.

Fig. 1-2: Feature type and their representations in two spatial data models [18]

The second type of GIS feature is spatial data management. In GIS, spatially referenced data

are typically arranged in the form of a layer. For example, a census data layer for the U.S. is a

collection of demography characteristics of the country with related tables of attributes

associated with each census. A collection of data layers forms a spatial database. For example,

8

Fig. 1-3 shows a geodatabase for urban stock studies. GIS manages a spatial relational database

management system (SRDBMS) with a structured query that extracts features based on their

locations and spatial relationships. For example, spatial query functions can be used to find the

point locations of cyanobacteria bloom in the Charles River with more than 0.05 mg/L of

nitrate concentration. GIS integrates general database maneuvers, such as a spatial query, and

several geostatistical analyses with the unique spatial visualization and geographical analysis

benefits offered by maps.

Fig. 1-3: A spatial database for urban mining studies [18]

Finally, GIS-based tools are used for spatial analysis and modeling. The analysis and modeling

steps are based on features absolute or relative locations, and the outcome depends on the

locations of the features being analyzed. GIS spatial analysis and modeling functions allow

9

users to define and execute spatial and attribute procedures to perform analysis in space. This

is the most important part of GIS functionality. GIS provides a large range of computational

tools that operate on the spatial aspects of the data, on the non-spatial attributes of these data,

or both. These tools range from simple geometric measurements such as area, perimeter and

length, to more complex functions such as spatial interpolation, network analysis, and spatial

statistics. These capabilities separate GIS packages from other information system based

modeling methods and make them valuable for a broad range of engineering and environmental

applications, including explaining natural events, simulating complex processes, predicting

outcomes, and supporting decision making. GIS is supported by substantial research

literature where the growing trends in the design of algorithms and novel computing

techniques for visualization and analysis of georeferenced data are evident [18].

1.2.2 GIS applications

GIS software can address a broad range of spatially related questions or procedures. Below

are some examples of the diverse capabilities of a GIS-based software package [25].

Spatial measurement: “The information of distance or the spatial extent or volume

of a feature or incident will be necessary but essential and, using proximity analysis,

GIS can establish the distance of objects about a theme or other objects. Any units

of measures can be deployed, including statistical measurements such as sum,

mean, mode and standard deviation” [25].

Spatial distributions: “Spatial distributions of features may be either random and

regular or clustered, and GIS have the functionality, usually via the use of the

nearest neighbor analysis, to describe distributions in these terms. Using contiguity

10

analysis, they can also calculate the relationship between differing distributions

across a surface, i.e., spatial autocorrelation” [25].

Network analysis: “This analysis applies to linear features such as transport routes,

rivers, pipelines and cable networks. Analyses can establish least costs routes,

shortest path routes, the degree of connectivity, etc. Measurement in network

analyses can be regarding monetary units, distance, time, etc.” [25].

Temporal analysis: “Spatial changes can be in absolute terms or defined over time.

Thus, it is valuable to know, for instance, the varying rates of growth in an urban

area over equal consecutive time periods, or to identify the proportional changes in

land use for a given area over time. The collection of long-term remotely sensed

data has significantly expedited time series analyses” [25, 28].

Modeling: “This is a broad heading that frequently includes “what if” scenarios or

models are constructed to show what a probable spatial distribution of an object

might be, given its known distribution in a sample area, and this can be done for

different temporal scenarios. Optimum location analysis is a modeling step that

optimizes the location of any activity based on known inputs of the principal

production functions. Digital terrain modeling allows for the inclusion of the height

dimension for GIS analyses of slopes, aspect, contours and volumes” [25].

Interpolation: “This is simply the generation of missing values based on a set of

known values within a study area. For instance, if a series of spot heights (altitudes)

are known, then it is possible to interpolate contour lines for the same area.

Interpolation can be applied to a wide range of measured values” [25].

11

1.3. GIS-based urban stock assessment

Modeling of urban stocks has been a useful tool in order to assess a city’s response to future

conditions and its resource self-sufficiency in the context of urban metabolism [7]. GIS can

be used to determine where the stocks are located, how much they are and how they are

distributed. Two of the most fundamental categories of urban stocks are buildings and land

parcels. GIS can inform evaluation of a city’s degree of self-reliance, or its ability to fulfill

human needs (shelter, energy, and food) [29]. Also, GIS-based approaches can assess the

environmental and socio-economic impacts or trade-offs among different patterns of urban

stocks and scenarios of use [30].

GIS-based tools are particularly useful for analyzing different urban stocks using a bottom-

up approach [18]. Typically, the process starts with extracting local stock information as a

geodatabase or database with appropriate geolocation or standard Cartesian projection

systems like Universal Transverse Mercator (UTM), or a national, regional or locally-

defined Euclidean grid system. The information stored in a spatial database may include

detailed map data layers describing the spatial distribution, configuration and properties of

urban land, structures, and infrastructures (such as buildings, roads, and bridges), the major

uses of interest (such as driveways, lawns, and food gardens) and other environmental,

demographic and economic data (such as population density, soil quality, or parcel

ownership). Once a comprehensive spatial database is built, geoprocessing and statistical

tools can be used to delineate or summarize spatially and allocate stocks. Spatial

visualization functions are then used to map the spatial distribution of the stocks [7].

12

Several studies have been used GIS as a tool for stock assessment, primarily focused on

mapping the distribution of specific materials such as concrete, steel, and other metals.

Tanikawa and Hashimoto [7] applied the GIS-based technology to estimate construction

material stocks over time with spatiotemporal data. Their study involved the use of a

geodatabase containing urban area buildings, roadways or railways, and sewer networks.

Two spatial databases of two urban study areas were constructed. The construction material

stocks were estimated by spatial database of buildings, roadways and railways stocks.

Based on this analyses, they estimated material accumulation both above and below

ground.

Sugimoto et al. [31] developed a 4d-GIS model which is a database of spatial 3D GIS data

with a time scale, to estimate building material stock and flow and visualize the transition

of buildings in urban districts for contribution to spatial designs. By utilizing 4d-GIS, they

observed that steel buildings initially accounted for 60%, but the number of reinforced

concrete buildings increased over the 50-year study period. Building material stock grew

from 0.35 million tons in 1961 to 97.5 million tons in 2010, largely because of the increase

of RC buildings. They also predicted the building renovation for Business as Usual (BAU)

and District Renewal Plan (DRP) scenarios up to 2050. The results show that building

material stock does not differ much between BAU and DRP scenarios in 2050, but it was

observed in the DRP scenario, building material stock increased by 0.2 million tons

because buildings will be rebuilt collectively in units.

13

Wallsten et al. [17] also used GIS to spatial location and distribution of hibernating metal

stocks of copper, aluminum and, iron in the infrastructure systems as a raw material for

local power, telecommunication, gas and district heating in the city of Norrköping,

Sweden. Here, they used a spatial database containing data layers of cables and pipes, and

buildings. The city area was divided into multiple zoning based on building types and local

zoning code. Per housing unit metal concentrations were estimated for each building-type.

This was followed by the differentiation of the active and inactive infrastructure systems.

Then they calculated active and inactive copper and aluminum stocks (concentrations) with

different cable and pipe types and size. Finally, the metal stocks were cataloged for each

urban districts and mapped considering urban district as area features.

Krook et al. [20] used a similar approach. They used spatial data to characterize the power

grids of Gothenburg and Linköping cities in Sweden. The power grid location was

quantified with regards to city’s total cable length, voltage levels, locations and operational

status. The stocks of copper, in-use and hibernating were estimated for local power

networks by multiplying the cable length with an average copper concentration. The

parameter quantified indicates the economic conditions for the recovery of cables in

hibernation located in the urban environment.

Van Beers and Graedel [19] also quantified and mapped end-of-life stocks and flows of

copper and zinc in Australia at local scale area. The study was based on existing stocks in

use, residence times, and historical and projected future evolution. The research showed

that the integration of GIS with stock analysis facilitated the end-of-life copper and zinc

14

comparison for different demography and industrial characteristics. The outcome also

provided useful insights for the optimization of copper and zinc recycling.

Based on the review above it can be summarized that GIS-based tools have been

successfully applied to study urban stocks of materials at high resolution. These studies

are primarily focused on assessing the sustainability of resources from the context of

recovering and reusing materials once they reach end of life. However, there has been

relatively little work on analyzing the degradation of urban material stocks over time.

There is also potential to utilize apply GIS and remote sensing in bottom-up assessment of

existing urban stocks in order to determine the resulting resource flows (e.g., energy, food)

that these stocks could support. These research questions of urban self-sufficiency are of

particular interest when considering possible supply chain disruptions due to extreme

events and global climate change.

1.4. Motivation and research objectives

To, Assessment of potential future climate impacts and self-sufficiency need to be

incorporated into urban planning in order to sustain future resourceful and low impact cities

with increasing populations. To adequately address these modeling challenges requires

state-of-the-art tools and city-specific case studies. Many recent studies have discussed the

importance of GIS and remote sensing based tools as a tool for assessing infrastructure

response to future conditions and urban self-reliance. Stewart et al. [32] conducted one

motivational study that looked at the temporal extent of concrete degradation under climate

change in the main Australian cities. They have pointed out the importance of mapping the

15

vulnerable concrete structures in climatically sensitive and disaster-prone urban regions.

This type of research will be useful in devising appropriate mitigation techniques and

possible concrete design code modification. Another useful study conducted by Niblick et

al. [33] assessed Pittsburgh’s energy potential in producing biofuel from local scale

sunflower agriculture. They showed how GIS-based tools could be used in estimating the

city’s alternative energy resources and contribute to the state’s renewable portfolio

standard. These studies sparked the following questions that motivated the dissertation

research.

1. What is the present and future extent of the urban built environment’s susceptibility

to climate change?

2. How self-sufficient can a city be in producing energy or food from agricultural

activities by leveraging local land and building resources? Which crop species are

most promising? Are there advantages to taking a regional rather than a strictly

urban approach to assessing resource self-sufficiency?

These overarching research questions were explored through literature review to analyze

conducted as part of the dissertation were motivated by the aforementioned articles, to

expand them and fit them to other less research areas in urban stock analysis. The present

work considers the following specific research objectives:

1. Urban scale mapping of both new and existing concrete buildings under projected

climate change.

16

2. GIS-based modeling to assess biomass-based bioenergy production potential on

urban marginal land using Boston as a case study.

3. Application of the geospatial model developed in objective 2 in assessing regional

scale bioenergy production potential using marginal and degraded lands.

4. GIS and remote sensing application in assessing urban food production potential

on underutilized land parcels and building rooftops.

1.5. Dissertation Structure

The work shown in this dissertation was conducted under the scope of a Northeastern

University Tier 1 grant on “Mapping coastal urban production of food and fuel: What do

know and where can we grow”, as well as Adviser’s startup funds. This research connects

urban metabolism and self-sufficiency and presents both novel methods as well as

engineering analyses that are relevant to current policies and the incorporation of urban

resource sustainability metrics in urban design and engineering, as described in each

chapter.

Chapter 2 describes a study on 4D-GIS based assessment of mapping vulnerable building

stocks in Boston under different projected climate change scenarios. As, Anthropogenic

increases in atmospheric greenhouse gas (GHG) concentrations and resultant changes in

climate will have significant detrimental effects to urban infrastructure, from both extreme

events and longer-term processes. Here we investigate impacts to concrete structures due

to enhanced corrosion through increases in carbonation and chlorination rates. High and

low emission scenarios (IPCC A1FI and B1) are used in combination with downscaled

temperature projections and code-recommended material specifications are used to model

17

carbonation and chloride-induced corrosion of concrete structures in the Northeast United

States. Geospatial modeling in the Boston metropolitan area is used to project building and

block-level vulnerability of urban concrete structures to future corrosion, and related

maintenance needs, and to project cover thickness degradation for the existing building

stock. The methods described here can be used for city-specific modeling of long-term

climate impacts on concrete infrastructure and provide a scientific basis for future-oriented

construction codes. This study was published in 2014 [34] and featured in Boston Globe.

Chapter 3 describes another GIS-based study for assessing a city’s energy self-reliance

from leveraging local resources. Urban marginal land can be used for growing high yield

bioenergy crops such as miscanthus and poplar. In this study, a GIS-based modeling

framework was created to assess potential urban marginal lands in Boston that include

vacant lands and under-utilized public and private areas with adequate soil quality and

sunlight. Using ArcGIS model builder as a spatial analysis tool, land parcels were screened

for typical urban features such as buildings, driveways, parking lots, water and protected

areas. The resultant layer was subjected to bio-geophysical modeling that includes soil

quality, slope analysis and, finally shadow analysis. The next study will explore other urban

regions of Massachusetts that might be able to fulfill part of their energy demand locally

while providing benefits in environmental quality, economic development, and urban

resilience. This study was published in 2015 [35].

Chapter 4 describes a study in continuation of the work shown in chapter 3. This is an

application of GIS-based model developed in chapter 3, for regional scale bioenergy

18

production potential on marginal and degraded lands. Eastern Massachusetts was

considered for this study. The ‘marginal’ land use category includes vacant and abandoned

lands, inner city underutilized areas, and degraded lands. Three different energy crops;

miscanthus, poplar and willow were considered as biomass feedstocks for bioenergy

production estimation. Fast-growing poplar was found as most suitable high yield

bioenergy crops for this region. Bioenergy potential calculation revealed that there exists

significant bioenergy potential for this region that can be used for heat, conversion to fuels,

or electricity production, particularly for microgrid or district heating applications. The

outcomes of this study are in line with previous work that evaluated marginal land

availability in Boston and other major Eastern U.S. cities, and confirm the accuracy of the

spatial model constructed in chapter 3. The study outcome is currently in preparation for

journal submission.

Chapter 5 describes another GIS-based assessment of urban scale food system and

production potential. Urban farmlands can be leveraged for developing a sustainable food

system by growing high yield food plants. Here, the GIS and remotely sensed data were

used to develop an automated model to assess Boston’s available area for urban farming,

including both rooftop and ground level areas. Geoprocessing and spatial analysis tools

were used to process geographic data layers for zoning, ownership, slope, soil quality, and

adequate light availability. Surface slope (roof pitch) was determined for all buildings in

the city through creation of a digital surface map from remotely sensed LiDAR data. Also,

ground level public and private vacant lands and underutilized residential and commercial

areas were mapped. Finally, food yield data for typical urban crops were used to estimate

19

the city’s food production potential. The study outcome was compared with other regions

in Eastern United States that might be able to fulfill part of their food demand locally while

providing benefits in local environmental quality and economic development. The study

outcome is in preparation for journal submission.

As the final chapter, Chapter 6 presents critical summary of all the research completed in

this dissertation and explains potential future work to follow these studies.

20

Chapter 2 Urban Scale Mapping of Concrete Degradation from Projected

Climate Change

This study has been published

Saha, M., & Eckelman, M. J. (2014). Urban scale mapping of concrete degradation from projected climate

change, 9, 101-114.

Anthropogenic increases in atmospheric greenhouse gas (GHG) concentrations and

resultant changes in climate will have significant detrimental effects to urban

infrastructure, from both extreme events and longer-term processes. Here we investigate

impacts to concrete structures due to enhanced corrosion through increases in carbonation

and chlorination rates. High and low emission scenarios (IPCC A1FI and B1) are used in

combination with downscaled temperature projections and code-recommended material

specifications are used to model carbonation and chloride-induced corrosion of concrete

structures in the Northeast United States. The results suggest that current concrete

construction projects will experience carbonation and chlorination depths that exceed the

current code-recommended cover thickness by 2077 and 2055, respectively, well within

the lifetimes of these buildings, potentially requiring extensive repairs. Geospatial

modeling in the Boston metropolitan area is used to project building and block-level

vulnerability of urban concrete structures to future corrosion, and related maintenance

needs, and to project cover thickness degradation for the existing building stock. The

methods described here can be used for city-specific modeling of long-term climate

impacts on concrete infrastructure and provide a scientific basis for future-oriented

construction codes.

21

2.1 Introduction

Understanding the implications of climatic variation has become a critical issue for

infrastructure maintenance planning. The Earth’s average temperature has been increased

by 0.6 oC since the 1900s and is expected to increase by approximately 1.4-5.8 oC by the

end of this century [36]. Many of the effects of climate change, including changes in

temperature, pollutant concentrations, relative humidity, precipitation, and wind patterns,

as well as increased frequency of severe events could have significant impacts on the

operations and lifespan of critical and non-critical infrastructure [37]. Infrastructure

capacity could be acutely overwhelmed (e.g., sea walls failing due to storm surge) or

degraded gradually. Assessing the potential impacts of climate change on the built

environment is difficult, as the relationship between material degradation and climate is

complex [38]. The Northeastern United States is likely to see an increase in extreme

precipitation events as well as overall increases in temperature and relative humidity [39].

Climate-induced damages to urban infrastructures are of particular concern. Urban areas

in the United States currently include approximately 250 million residents, projected to

grow to ~365 million by 2050 [40]. While the urban share of population and economic

output in the US has grown in the past decades, much of the existing urban infrastructure

has become increasingly susceptible to failures [41, 42]. Aging buildings and

transportation, energy, water, and sanitation infrastructure are all expected to become more

stressed in their ability to support existing services for urban residents in the coming

decades, especially when the impacts of climate change are added as stressors [43]. Climate

22

change will also contribute directly to physical degradation of infrastructure and building

materials [44].

While much research on climate change impacts has focused on infrastructure

susceptibility to extreme events and flooding from long-term sea level rise [45], relatively

few studies have been carried out on the direct effects of climate change on the structural

deterioration of infrastructure. One direct mechanism is acidic attack of cementitious

materials. Concrete degradation due to acid rain has been extensively studied [46], and

elevated levels of atmospheric CO2 will increase the formation of carbonic acid in

precipitation. Similarly, uptake of CO2 by the oceans and the resulting decrease in pH will

amplify degradation of structures in urban coastal areas that are exposed to seawater [47].

Another mechanism for climate-induced concrete degradation is through early failure of

the protective concrete cover over reinforcing steel, leading to corrosion and spalling, due

to changes in CO2 and temperature [48, 49], which has only recently been analyzed. Yoon

et al. [50] was among the first to consider the effects of climate change on concrete

performance and lifetime, in particular the effect on carbonation rates; however, this model

does not account for the influence of temperature change, which can significantly affect

the diffusion coefficient of CO2 into concrete, the rate of reaction between CO2 and

Ca(OH)2, and the rate of dissolution of CO2 and Ca(OH)2 in pore water. The model is also

a time-independent predictive model that assumes CO2 concentrations to be constant up to

a given time, thereby underestimating carbonation depths under changing atmospheric

conditions [51]. Stewart et al. [52] built on the work by Yoon et al. [50] by taking into

23

account the effect of temperature on the diffusion coefficient, but they did not consider the

influence of temperature on the other aforementioned parameters. Their work looked not

only at carbonation and chlorination, but also at the time to crack initiation, crack

propagation, and failure due to reinforcement corrosion. Similar carbonation and

chlorination models were used by Stewart et al. [52] in their work, who noted that there is

a need for an improved model that considers the time-dependent effect of CO2

concentration and other parameters such as temperature and relative humidity.

Recently, Talukdar et al. [48] estimated carbonation (but not chlorination) penetration

depths in concrete due to projected climate change. Several deterministic model parameters

were experimentally verified using unloaded/undamaged concrete. They reported 25-35

mm increase in penetration depth due to carbonation alone. Separately, Bastidas-Arteaga

et al. [53] investigated the influence of global warming on chloride ingress into concrete

using a stochastic model of chloride penetration and corrosion initiation. Their particular

approach was to model future weather conditions, recognizing that temperatures will

fluctuate not only over the century, but also during a given year, and that the duration of

the hot season throughout most of the world is expected to lengthen over this century. They

found significant correlation between chloride ingress over time associated with projected

global warming. Talukdar et al. [54] then improved their carbonation model and coupled

it to the climate model proposed by Bastidas-Arteaga et al. [53] to project concrete

infrastructure degradation and to consider the suitability of current code requirements.

24

The current study builds on these previous reports by estimating climate-induced changes

in corrosion depths for both carbonation and chloride induced corrosion for multiple

climate scenarios and at a high level of geospatial resolution. A scenario modeling

approach is used to estimate the carbonation and chlorination depth in cement concrete

under both high and low greenhouse gas (GHG) emission scenarios (IPCC A1FI and B1).

The results are then compared to the minimum American Concrete Institute (ACI) code-

recommended concrete cover thickness, which is meant to prevent corrosion of reinforced

concrete structures but does not currently reflect future anticipated changes in climate.

Geographic information system (GIS)-based spatial analysis identifies the location of

vulnerable concrete structures within metropolitan Boston, a densely populated area home

to nearly 4.5 million people and representative of urban coastal areas on the Atlantic

seaboard [40]. The results from this study are both temporally and geographically specific

and relevant to infrastructure policy and construction codes, maintenance planning, and

technology adoption to mitigate corrosion-induced damage of the built environment.

25

2.2 Climate Change Scenarios

Projections of atmospheric and climatic conditions are required at the urban scale for this

study. Earlier work on climate change effects on Boston by Ruth [55] projected increases

in annual temperatures of 2-5 oC (4-9oF) and in annual precipitation up to 25%, with the

frequency and intensity of extreme precipitation events and droughts increasing as well.

Temperature changes will likely be accompanied by continual rise of sea levels with

seasonal and periodic fluctuations, and may exacerbate already existing natural hazards in

the region, including floods, heavy rainstorms, and hurricanes [56]. Temperature changes

and CO2 concentrations were accounted for in this study, as these are likely to have the

most direct effect on concrete deterioration. Rainfall and relative humidity also affect

erosion of concrete surfaces as well as diffusion of pollutants into the subsurface but were

not included for the following reasons. The effects of rainfall on concrete structures are

highly site specific [48], and both precipitation and relative humidity projections carry high

uncertainty in long-term climate models [57].

2.2.1 Atmospheric CO2 Concentrations

Carbonation of concrete is a function of ambient concentrations of CO2. Future

atmospheric CO2 concentrations were projected based on the IPCC A1 and B1 special

report on emission (SRES) scenarios [58]. The A1 scenario indicates very rapid economic

growth, a global population that peaks in mid-century and declines thereafter, and the rapid

introduction of new and more efficient technologies, as well as substantial reduction in

regional differences in per capita income. A1 scenario has two sub-categories that include

A1FI and A1B. A1FI (fossil intensive) scenario considers the rapid introduction of new

26

and more efficient technologies, substantial reduction in regional differences in per capita

income, and intensive fossil energy consumption [58]. (A1B is a moderate scenario that

balances fossil intensive and non-fossil intensive energy sources.) The B1 scenario

assumes the same population trend as A1, but with rapid changes in economic structure

towards a service and information economy, with reduction in material intensity and

introduction of clean and resource-efficient technologies. A control emissions scenario

based on year 2000 ambient CO2 concentrations is also considered to provide a reference

for other emission scenarios.

Fig. 2-1 depicts projected CO2 concentrations from 2000-2099 based on the Model for

Assessment of Greenhouse-gas induced Climate Change (MAGICC) [59]. Boston is an

urban center with a high density of fossil fuel combustion activities, particularly from the

transportation sector. In such urban settings, CO2 concentrations can be 5-30% higher than

nearby rural environments [60]. These elevated urban CO2 concentrations are reflected in

an urban environment term ke, which is introduced as a correction factor in the methods

section.

27

Fig. 2-1: Predicted estimates of CO2 concentrations. Actual atmospheric CO2

concentration has been plotted side by side along with predicted concentrations for 2000-

2013 and shown in the inset panel

2.2.2 Temperature Predictions

Future temperatures for Boston were predicted using data generated from four different

atmospheric-oceanographic global climate models (AOGCM). These include the

Community Climate System Model (CCSM), Hadley Climate model (HADCM2),

Geophysical Fluid Dynamics Institute Model (GFDI) and Parallel Climate Model (PCM).

Each of these were downscaled to give station-based predictions for Boston’s Logan

Airport. Daily average temperature data were extracted for each of the A1FI and B1

emissions scenarios respectively.

28

Downscaling was performed using the asynchronous regional regression model (ARRM)

developed by [61]. This technique was established to bridge the gap between large-scale

outputs from AOGCMs and the fine-scale output required for urban climate impact

assessments. ARRM uses piecewise regression to quantify the relationship between

observed and modeled quantiles and then downscale future projections [62]. Fig. 2-2

presents the change in average annual temperature in next 100 years for different

AOGCMs.

Fig. 2-2: Predicted mean temperature increase for Boston’s Logan Airport station for

A1FI and B1 scenarios; results downscaled from CCSM, HADCM3, GFDI, PCM models

2.3 Corrosion

2.3.1 Concrete Carbonation Modeling

Carbonation depth depends on several parameters such as concrete quality, protective

cover thickness, temperature, and ambient CO2 concentration. Carbonation of concrete has

been studied extensively and various models have been developed for the purpose of

predicting carbonation depths [48, 52]. It has been observed that corrosion may occur when

0

1

2

3

4

5

6

7

2000 2020 2040 2060 2080 2100

Tem

pera

ture

In

cre

ase (

oC

)

Year

CCSM-A1FI

HADCM3-A1FI

GFDI-A1FI

PCM-A1FI

CCSM-B1

HADCM3-B1

GFDI-B1

PCM-B1

29

the distance between the carbonation front and the rebar surface is less than 1-5 mm [50],

although probabilistic analyses for assessing durability design specifications tend to ignore

this effect. Here we assume that corrosion initiation occurs when carbonation front depth

equals the concrete cover thickness.

In the carbonation depth model recommended by DuraCrete [63] and Yoon et al., [50] and

others, carbonation depth (xc in cm) in year t is predicted as a diffusion process:

𝑥𝑐(𝑡) = √(2𝐷𝐶𝑂2

(𝑡)

𝑎) ∙ 𝑘𝑒,𝐶𝑂2 ∙ 𝐶𝐶𝑂2

(𝑡) ∙ (𝑡 − 𝑡0) ∙ (𝑡𝑖,𝐶𝑂2

𝑡−𝑡0)

𝑛𝑚

; 𝑡 ≥ 2000 (i)

The time and temperature-dependent diffusion coefficient DCO2(t) is given by:

𝐷𝐶𝑂2(𝑡) = 𝑓𝑇(𝑡) ∙ 𝐷0,𝐶𝑂2(𝑡 − 𝑡0)−𝑛𝑑,𝐶𝑂2 (ii)

where D0,CO2 is the initial CO2 diffusion coefficient, with a recommended range of 0.5-50

× 10-4 cm2/s ; nd,CO2 is the ageing factor for the CO2 diffusion coefficient determined

empirically; and t0 is the initial year, 1999. The time-dependent temperature factor is

described as:

𝑓𝑇(𝑡) ≈ 𝑒𝑥𝑝 (𝐸

𝑅 −

1

273+𝑇𝑎𝑣𝑔(𝑡)) (iii)

where Tavg(t) is the running average temperature (oC) over the interval t–t0, E is the

activation energy of the diffusion process (40 kJ/mol) [64]; and R is the universal gas

constant (8.314×10−3 kJ/mol-K).

Cement characteristics are combined in Eq. (i) in a single factor a:

𝑎 = 0.75𝐶𝑒 ∙ 𝐶𝐶𝑎𝑂 ∙ 𝛼𝐻 ∙𝑀𝐶𝑂2

𝑀𝐶𝑎𝑂 (iv)

30

where Ce is the cement content (kg/m3); CCaO is the calcium oxide content in cement; MCaO

is the molar mass of CaO; MCO2 is the molar mass of CO2, and αH is the degree of hydration,

given by:

𝛼𝐻 ≈ 1 − 𝑒−3.38(𝑤/𝑐) (v)

The water-cement ratio w/c is a design parameter for concrete and is codified according to

the intended exposure condition (Table 2-1).

The remaining terms in Eq. (i) are as follows: ke,CO2 is a correction factor to account for

increased CO2 levels in urban environments; CCO2(t) is the time-dependent concentration

of ambient CO2 (10−3 kg/m3) in Section 2.1 (using the conversion factor 1 ppm =

0.0019×10−3 kg/m3); and (𝑡𝑖,𝐶𝑂2

𝑡−𝑡0)

𝑛𝑚

reflects the aging of the concrete matrix, including

ti,CO2 a reference time, 1yr; and an ageing factor for microclimatic conditions (nm)

associated with the annual frequency of wetting and drying cycles. nm is 0 for sheltered

outdoor surfaces and 0.12 for unsheltered outdoor surfaces. All parameter values and

supporting references are given in Table 2-1.

Table 2-1. Structural durability conditions for concrete used in this study

Parameters Unit Value Reference

General

f/c MPa 34.5 ACI Code, [65]

w/c dimensionless 0.4 ACI Code, [65]

Ce kg/m3 450 Stewart et al. [51]

CCaO dimensionless 0.65 Stewart et al. [52]

MCO2 g/mol 44 Stewart et al. [52]

MCaO g/mol 56 Stewart et al. [52]

nm dimensionless 0.12 Yoon et al. [50]

Carbonation

D0,CO2 cm2/s 4×10-4 Yoon et al. [50]

ti year 1.0 Yoon et al. [50]

nd,CO2 dimensionless 0.24 Yoon et al. [50]

31

ke,CO2 dimensionless 1.2 Yoon et al. [50]

Chlorination

D0,Cl m2/s 7.0×10-12 Stewart et al. [52]

ti year 0.07 Stewart et al. [52]

nd,Cl dimensionless 0.35 Stewart et al. [52]

Co kg/m3 8.0 Val and Stewart. [66]

Ccr kg/m3 3.35 Val and Stewart. [66]

ke,T dimensionless 0.92 Stewart et al. [52]

kt dimensionless 1.0 Stewart et al. [52]

kc dimensionless 1.0 Stewart et al. [52]

The years 2000 to 2099 were considered. Corrosion depths were also estimated without

any climate change effects by keeping CO2 and temperature constant over the time period

in the corrosion models. This was termed as control condition and will be used in later

sections.

2.3.2 Concrete Chlorination Modeling

Chloride ions can penetrate concrete by capillary absorption, hydrostatic pressure, and

diffusion [67]. The most recognized method is diffusion, governed by Fick’s second law

[68], although chloride penetration processes and field conditions can deviate from those

assumed for Fickian diffusion [66]. Surface chloride concentration, Co and diffusion

coefficients, Dcl are easily calculated by fitting Fick’s law to measured chloride profiles.

An improved model utilizing a time-dependent chloride diffusion coefficient proposed by

DuraCrete [69] is used here to calculate chloride concentration. Corrosion initiation occurs

when the chloride concentration at the level of reinforcement exceeds the critical chloride

concentration, typically 3.35 kg/m3, which is independent of concrete quality [69].

32

The chloride concentration at depth xcl mm in year t has the standard solution:

𝐶(𝑥, 𝑡) = 𝐶0

[

1 − 𝑒𝑟𝑓

(

𝑥𝑐𝑙

2√𝐷𝐶𝑙(𝑡).(𝑡𝑖,𝐶𝑙𝑡−𝑡0

)𝑛𝑚

.(𝑡−𝑡0)

)

]

; t ≥ 2000 (vi)

DCl(t) is the time and temperature-dependent apparent chloride diffusion coefficient:

𝐷𝐶𝑙(𝑡) = 𝑘𝑒,𝑇 ∙ 𝑘𝑡 ∙ 𝑘𝑐 ∙ 𝑓𝑇(𝑡) ∙ 𝐷0,𝐶𝑙(𝑡 − 𝑡0)−𝑛𝑑,𝐶𝑙 (vii)

where ke,T is a correction factor to account for increased temperatures due to urban heat

island effects; kt is the test method factor (1.0); kc is the curing factor (1.0); ti,Cl is the

reference time in years (28 days or 0.077 years); fT(t) is the temperature effect on diffusion

coefficient given by Eq. (iii); and nd,Cl is the chloride-specific ageing factor determined

empirically.

Table 2-2. Recommended concrete properties for corrosion protection (ACI Code, [65])

No. Exposure Condition Maximum

w/c ratio by

weight

Minimum

f/c (MPa)

1 Concrete intended to have low permeability when

exposed to water 0.50 27.5

2 Concrete exposed to freezing and thawing in a

moist condition or to deicing chemicals 0.45 31.0

3 Corrosion protection of reinforcement in concrete

exposed to chlorides from deicing chemicals, salt,

salt water, brackish water, seawater or spray from

these sources

0.40 32.5

The surface chloride concentration Co is considered as a time-independent variable as

exposure to chlorides for a specific structural member would not change significantly from

year to year [52]. However due to climate change, fluctuation in wetting/drying cycles,

rainfall and wind patterns could vary but there is no data to support how this might affect

33

Co. The surface chloride concentration is reported as 7-8 kg/m3 for splash/tidal zones [66].

In this study Co = 8 kg/m3 was assumed. Structural exposure conditions were chosen from

Table 2-2 as condition 3, which references “concrete exposed to chlorides from deicing

chemicals, salt, salt water, brackish water, seawater or spray from these sources” [65]. The

necessary data for estimating DCl(t) were derived from [52]. All parameter values used in

the chlorination model are shown in Table 2-1.

2.4 Spatial Analysis

The analysis described in Sections 2-3 apply generically to new buildings constructed in

the Boston area. In order to identify and evaluate the spatial distribution of climatically

vulnerability for actual concrete structures within the city, geospatial analysis was

performed at higher resolution using ArcGIS 10.2 (ESRI). Gridded temperature data for

the year 2000-2099 were collected from the National Center for Atmospheric Research

(NCAR) climate database [70]. The NCAR database has gridded downscaled data available

from the community climate system model (CCSM) only. Therefore, for spatial analysis

purposes, CCSM is chosen among four different AOGCMs discussed in Section 2.2.

Raster-based geo-statistical analysis has been performed to identify the climatically

vulnerable region within the city according to high (A1FI) and low emission (B1)

scenarios. The resolution of downscaled climate data was 3 miles × 3 miles (0.33o). The

raster analysis of gridded climate data was performed using the ‘geospatial analysis’ tool

in ArcGIS. The raster analysis provided contour plots of temperature within the city.

Ordinary kriging with no transformation was used for spatial interpolation.

34

The goal of the spatial analysis is a building-by-building assessment of concrete

vulnerability to enhanced carbonation and chlorination. Rather than using a baseline

construction year of 2000, here we identified and cataloged concrete structures according

to their actual year of construction. Building age and composition data were obtained from

GIS clearing houses including the city of Boston GIS data hub [71], MassGIS data layers

[72] and Boston Redevelopment Authority planimetric dataset [73]. Using these data, an

atlas of concrete buildings by decadal cohort was created and analyzed to identify

vulnerable building structures. The scope of climate vulnerability was limited to concrete

buildings but could also be extended to roadways and bridges. Fig. 2-3 presents the

locations of all 1,700 concrete buildings in Boston. Concrete buildings represent 3% of

total available land area within the city and 46% of total building area. Carbonation and

chlorination models were run for each building cohort beginning in 1951 to estimate the

year at which the concrete cover for each building would be compromised.

Fig. 2-3: Concrete structures (buildings, in red) within Boston (outlined in gray)

35

2.5 Results

2.5.1 Carbonation and Chlorination Depth.

Results from the corrosion depth analysis indicate significant and prolonged climate-

induced effects on both carbonation and chlorination of concrete structures (Fig. 2-4 and

2-5). Results suggest that for Boston, for the HADCM3 model under the worst case

scenario (A1FI), carbonation depths would equal the protective cover thickness by the year

2077 and would exceed the cover thickness by approximately 15 mm by the end of the

century. This represents an increase in depth of nearly 40% compared to that expected

assuming current temperatures and CO2 concentrations (Table 2-4) for which the ACI code

was designed. Penetration rates for chlorination exceed those for carbonation. For chloride-

induced corrosion under the same HADCM3–A1FI scenario, the protective layer is

exhausted even earlier, by the year 2055, with penetration depths 12% higher than in the

control scenario (Table 2-4).

By 2030, the maximum chloride penetration depth was calculated to be 27 mm, which is

11-12 mm higher than the predicted carbonation depth (Fig. 2-4). After 2030, accelerated

carbonation due to climate change effects noticeably diverges from the control case (Fig.

2-4). This is generally expected as, over time, the temperature differential becomes greater

and hot seasons become extended. Averaging across climate models and emission

scenarios, corrosion damage will be noticeable after 2083 (Fig. 2-4) for carbonation and

by 2062 due to chloride ingress (Fig. 2-5). As, after that timeframe both the carbonation

and chlorination depth will exceed the code recommended protected cover (38-50 mm)

installed in the majority of concrete structures.

36

Predicted carbonation and chlorination depths were compared with results from previous

studies, and were found to be higher than the previous studies for both carbonation and

chloride induced corrosion. Talukdar et al. [54] reported 23 mm for Vancouver, while [32]

found 28 mm for Sydney. Talukdar and Banthia [74] later compared different cities

(London, Mumbai, New York, Toronto and Vancouver) across the globe and estimated

carbonation depths that ranged between 15-35 mm by the end of the century. For New

York City, the estimated depth was 35 mm for a high emission scenario, which is 25% less

than the carbonation depth estimated for Boston in the present work. The higher result here

is due in part to a different assumed initial diffusion coefficient. Here we assumed a value

of D0,CO2 of 4×10-4 cm2/s, following previous work [50], whereas Talukdar and Banthia

assumed a range of 0.5-1.6×10-4 cm2/s. Secondly, this study considered predicted rise in

GHG emissions for SRES scenarios according to IPCC 4th assessment report, which

described higher values than the 3rd assessment report used by Talukdar and Banthia [74].

Chlorination results have been compared with Stewart et al. [32], who investigated major

coastal urban areas in Australia. The results suggest a 9 mm increase in design concrete

cover for Boston by 2100 for the A1FI SRES emission scenario, which is slightly higher

than previous results reported for Australian cities (5-7 mm increase). Differences in

structural exposure conditions that include surface chloride concentration, Co concrete

strength, f/c and regionally typical water cement ratio (w/c) contributed to the higher results

found for Boston.

37

Fig. 2-4: Estimated carbonation depth (mm) in BMA for a building constructed in 2000

Fig. 2-5: Estimated chlorination depth (mm) in BMA for building constructed in 2000

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100

Carb

on

ati

on

Dep

th (

mm

)

Year

A1FI-CCSM A1FI-HADCM3 A1FI-GFDI

A1FI-PCM B1-CCSM B1-HADCM3

B1-GFDI B1-PCM Control

0

10

20

30

40

50

60

2000 2020 2040 2060 2080 2100

Ch

lori

nati

on

Dep

th (

mm

)

Year

A1FI-CCSM A1FI-HADCM3 A1FI-GFDI

A1FI-PCM B1-CCSM B1-HADCM3

B1-GFDI B1-PCM Control

38 mm

∆x = 26 yr

∆x = 10 years

38 mm

38

Table 2-3. Minimum cover (mm) required to counteract the impact of climate change on

carbonation-induced corrosion damage risks by 2100.

Emission

scenarios

CCSM HADCM3 GFDI PCM

A1FI 51.1 52.6 51.0 49.8

B1 43.0 43.7 43.1 42.9

Control 37.6 37.6 37.6 37.6

ACI Code 38 38 38 38

Table 2-4. Minimum cover (mm) required to counteract the impact of climate change on

chlorination-induced corrosion damage risks by 2100.

Emission

scenarios

CCSM HADCM3 GFDI PCM

A1FI 47.9 49.6 47.8 46.4

B1 44.7 45.5 44.8 44.5

Control 44.1 44.1 44.1 44.1

ACI Code 38 38 38 38

2.5.2 Geospatial Results

The results above suggest that new concrete structures constructed in urban coastal regions

using current code requirements may degrade prematurely and require costly maintenance

over their service life. Spatially resolved analysis at the urban scale will help to identify

vulnerable zones of existing buildings within the city to anticipate monitoring, testing, and

potential maintenance for specific concrete structures that are part of the current building

stock. The results are summarized in Fig. 2-6 for the high (A1FI) emission scenario.

39

Fig. 2-6: a) Location of gridded climate data, b) Climatically vulnerable zones within

Boston (A1FI emission scenario, predicted mean temperature at 2099, CCSM model)

The kriging analysis identifies the areas within the city those are vulnerable due to increase

in temperature over time (Fig. 2-6b). The average temperature in Boston may increase up

to 4-5oC by the year 2100. Results show that high-density areas of downtown Boston, East

Boston and Charlestown and part of South Boston will be subjected to higher temperature

increases from climate change. These areas are located within a four-mile radius of the

high-temperature increase (red) zones identified by kriging analysis (Fig 2-6b). An earlier

study revealed that these same areas will also be highly vulnerable to storm surge [55].

Consequently, the concrete structures situated in these areas will be vulnerable to higher

degrees of structural deterioration associated with corrosion.

Among the 1,700 concrete buildings identified within the city, approximately 60% are

located within the identified high risk zones for corrosion (Fig. 2-7a). In addition, nearly

45% of the concrete buildings in Boston are more than 35 years old, so that significant

carbonation and chlorination may have already occurred. There are roughly 10-12%

concrete buildings located especially in downtown region that will exceed the typical

a) b)

40

design service life for concrete structures (~ 60-75 years) by 2030 (Fig 2-7b). Their

extended service age and location on the waterfront increases the overall extent of

carbonation and chlorination and puts them at high risk for corrosion initiation over the

next 20-30 years. By 2050, penetration depths in nearly 60% of existing buildings will

exceed the code-recommended cover thickness (Fig 2-7b). Degradation rates will increase

over time with the projected change in climate and eventually affect 100% of the existing

building stock before 2080. Consequently, these buildings may require significant

inspection and maintenance. These spatial and temporal results can be used to formulate

appropriate testing protocols and maintenance schedules.

Fig 2-7: a) Concrete Structures classified according to different age, b) % of buildings

with compromised cover thickness over the service life

2.6 Discussion

2.6.1 Implications for Current Code Requirements for Concrete Cover

The results presented are based on structural exposure condition of ACI Code, 2011 which

only pertains to minimum concrete cover (38.5 mm) and concrete compressive strength

(32.5 MPa). It is important to note that, for a specific structure, more precise results in

a) b)

50%

41

terms of climate change impacts on corrosion can be obtained only by specifying material

and construction specifications and structural detailing. This increase in corrosion

phenomena is driven more by increases in atmospheric CO2 concentration in which there

is high confidence, and less by the less accurate projections in temperature and relative

humidity. Even under the control emission scenario with stable CO2 concentrations and

temperature, the increase in damage risks over the 21st century will still be significant and

cannot be ignored, particularly for chlorination-induced corrosion of structures located on

the coast.

The results also suggest that if concrete buildings in coastal locations are designed with up

to 10-12 mm extra concrete cover or use higher grade concrete or low carbon steel then

this will reduce the effects of climate change even if the climate change trends occur

according to the projected highest emission scenario (A1FI). A less severe emission

scenario (B1) would require less additional cover, perhaps as low as 3-5 mm. Where

stability is governed by chlorides, a 5-10 mm increase in design concrete cover is suitable

for the A1FI emission scenario. Increasing design cover by up to 8 mm or increasing

concrete compressive strength by one grade would increase construction costs by

approximately 2-4% [32], but has the potential to save billions of dollars per year in repairs.

Relevant ACI technical committees have recently discussed the effect of climate change

on concrete durability [75] and emphasized the need to formulate necessary code to

mitigate carbonation ingress in concrete structures. But to our knowledge, no U.S. state or

municipal authority has yet altered concrete cover requirements in response to future

climate projections.

42

2.6.2 Concrete Technologies for Climate Change Adaptation

Several anti-corrosion technologies are available that reduce the vulnerability of reinforced

concrete structures and thus improve their adaptive capacity to changing climatic

conditions. These include protective surface coatings, which are easy to apply to existing

buildings but can carry high costs. Acrylic-based surface coatings can reduce carbonation

depths by 10-65% [32]. Stainless steel, galvanized steel or other methods of cathodic

protection, or glass fiber reinforced polymer rebar can be used instead of low carbon steel

in new buildings. Chlorination can also be reversed through electro-chemical chloride

extraction, though the high cost of this technology means that it would be reserved for

critical structures [76]. These adaptation steps have the benefit of reducing or reversing

diffusion would improve the performance of concrete infrastructures under a changing

climate.

Many existing concrete structures are likely to suffer from decreased durability due to

climate change. As this risk varies widely with location, environmental exposure and

material design is therefore critical to predict for every individual structure. A protective

approach would suggest that increased monitoring and maintenance of concrete structures

would be preferred. Clearly, the costs and benefits of such an approach will vary widely

by location, and a life cycle assessment approach may be useful for efficient practical

implementation [51]. The interaction of atmospheric CO2 with concrete has implications

not just for concrete performance but also for climate change mitigation and GHG

accounting. Carbonate weathering of concrete and natural minerals acts as a carbon sink.

43

Yearly CO2 uptake by existing concrete structures can be estimated and included in

accounts of urban GHG emissions [77].

2.6.3 Future Research Needs

As chlorination induced corrosion is typically more widespread and severe than

carbonation-induced corrosion, especially in coastal regions, more research and field data

is necessary in this regard. One specific need is that the mechanistic relationship between

chlorination and carbonation processes has yet to be properly modeled. Chloride ingression

occurs through the release of bound chloride in the hardened concrete, leading to reduction

in alkalinity and potentially increasing the risk of carbonation-induced corrosion [78].

Therefore, the synergistic interaction between these two processes may result in corrosion

rates higher than those predicted here. Finally, it should also be noted that the results based

on these models depend on downscaled climate data, but that there is significant

uncertainty associated with downscaling and with long-term climate projections generally.

Every model has its associated uncertainty in parameter and data related uncertainty which

may affect the overall modeling performance as well as outcome. Therefore, more

elaborate, site-specific experimental carbonation and chlorination depths estimation are

necessary to compare experimental results with model performance for validation

purposes. The outcome of city-specific studies can also be extended to assess regional or

national level vulnerability assessments and policy modeling of code requirements. This

assessment can be used to anticipate climate change implications in the planning, design

and maintenance of concrete structures.

44

2.7 Conclusion

Relatively little research has been directed to long-term susceptibility of urban

infrastructure and materials to climate change outside of sea-level rise and flooding.

Critical questions remain about how and where climate change will affect the underlying

materials and structures in different types of urban settlement zones. The present study

details both a temporal and geospatial framework for analyzing climate change impact on

urban concrete structural deterioration, using corrosion depth as the primary metric. The

major findings of this study are:

Climate change will accelerate corrosion and degradation of concrete structures in

Boston. By the year 2055, the chlorination-induced corrosion depth in concrete

structures built in year 2000 may exceed the code recommended protected cover

thickness of 38 mm (1.5 in). For carbonation-induced corrosion the threshold year is

2077. Concrete code modification may be required in light of regional climate

projections.

Chlorination-induced corrosion is more prevalent in Boston compared to carbonation.

Carbonation induced corrosion primarily depends on atmospheric CO2 concentration

while chlorination depends on surface chloride concentration.

From the spatial analysis, it can be inferred that nearly 60% of the existing concrete

structures in Boston will be more prone to structural deterioration associated with

corrosion by 2050 compared to control scenario.

In conclusion, accelerated corrosion in concrete will be a long-term, globally prevalent

effect of climate change, particularly in coastal cities. Corrosion is also preventable with

45

appropriate changes to concrete codes and the use of protective coatings or alternative

building materials. Code adjustment can be made based on regional assessments, and finer-

scale modeling of concrete mechanics can extend this work to account for the effect of

loading and cracking. Additional modeling can also be applied under different future

climate scenarios, length scales, and geographic regions to anticipate the pervasive effects

of regional or global climate change on both unstressed and stressed concrete structures.

46

Chapter 3 Geospatial Assessment of Potential Bioenergy Crop Production

on Urban Marginal Land

This study has been published

Saha, M., & Eckelman, M. J. (2015). Geospatial assessment of potential bioenergy crop production on urban marginal land, 159, 540-547

Urban marginal land can be used for growing high yield bioenergy crops such as

miscanthus and poplar. Here, a GIS-based modeling framework was created to assess

potential urban marginal lands in Boston that include vacant lands and under-utilized

public and private areas with adequate soil quality and sunlight. Using ArcGIS model

builder as a spatial analysis tool, land parcels were screened for typical urban features such

as buildings, driveways, parking lots, water and protected areas. The resultant layer was

subjected to bio-geophysical modeling that includes soil quality, slope analysis and finally

shadow analysis. Approximately, 2660 hectares of potential marginal land was identified

as suitable, representing 24% of total land area in Boston. Using crop yield information, it

was estimated that this marginal land could be used to produce up to a total of nearly 42,130

tons of high yield short rotation poplar biomass in a regular growing season. Also,

bioenergy potential calculation revealed that for short rotation poplar, this amount of

biomass can potentially yield up to 745 TJ (LHV) to 830 (HHV) TJ of yearly primary

energy for the city of Boston that can be used for heat or electricity production, particularly

for microgrid or district heating applications. This is equivalent to ~0.6% of Massachusetts

primary energy demand for 2012. Ongoing work will explore other urban regions of

Massachusetts and the Eastern US that might be able to fulfill part of their energy demand

locally while providing benefits in environmental quality, economic development, and

urban resilience.

47

3.1 Introduction

Urban inhabitants represent the majority of global energy demand (75%), with more than

50% of the population currently residing and working in cities [79]. Although densely

settled cities cannot be self-sufficient in food or energy production, many communities are

considering growing dedicated energy crops on under-utilized land to produce food and

fuel for district heating and small-scale electricity production [80]. Such schemes provide

opportunities for public and private actors in municipalities to fulfill part of their energy

demand locally while providing potential benefits to residents in the form of improved

landscapes, economic development, and modulation of urban heat islands [81]. Local

sources of energy may provide a temporary buffer to communities when a power grid

failure or heating fuel supply disruption occurs due to natural catastrophes such as

hurricanes and floods [73]. However, it is important that urban bioenergy production

address community concerns such as odor, noise, or increased traffic [82], and not impede

other beneficial uses of valuable urban land.

A study conducted by U.S. Department of Energy reported an increase of bioenergy

production by more than 300% in the past decade, with potential 1.9 PWh available

annually in the contiguous United States [83, 84]. While much of the recent increase is due

to corn ethanol, woody biomass has seen growth as a primary or secondary fuel in electric

generating units and in residential high-efficiency pellet stoves. In addition to large-scale

agricultural and forestry operations to supply this bioenergy, urban marginal land can also

play a part in larger bioenergy production systems. Previous studies have characterized

urban marginal lands as land parcels that have limited economic value and not suitable for

48

agricultural or residential purposes [82, 85]. A central challenge of this work is to

determine the extent of marginal land for urban regions [86]. Recent urban energy and

geography research has focused on the development of computational tools to acquire data

and estimate urban marginal land for different cities [87]. One study estimated that

approximately 15% of U.S. urban land on average can be considered marginal [88].

Recently, many cities have been reclaiming their marginal or under-utilized land parcels

for use as recreational parks, playgrounds, and community gardens. However, growing

dedicated energy crops on these urban lands is a fairly new concept and needs further

investigation, even just to understand the scale of potential energy benefits.

Marginal land estimations for the U.S. have been conducted nationally, regionally and at

the city scale (Table 3-1). Scale and location are critical issues, as it is not cost-effective to

transport biomass resources over long distances [87]. Limited agreement exists among

techniques used to estimate potential for bioenergy schemes in previous studies [71, 79,

89, 90], but common approaches make use of Geographic Information System (GIS) and

remote sensing based tools [86]. GIS-based tools assess spatial patterns of biomass based

bioenergy production on marginal land for both urban and non-urban areas as well as

availability of suitable land.

49

Table 3-1. Review of studies on marginal land assessment in USA

Year Author Scale (USA) Crop Marginal land

definition

Percentage

of total

area

2011 Gopalakrishnan

et al. [91]

Regional

(Northeast)

Lignocellulosic Contaminated

brownfield

8.0

2013 Gelfand et al. [92]

Regional (Midwest)

Cellulosic Crop land with low soil quality

10

2013 Grewal and

Grewal [73]

Local

(Cleveland, OH)

Algae Vacant lands 40

2013 Niblick

et al. [33]

Local

(Pittsburgh, PA)

Sunflower Vacant and abundant

lands

35

2014 Milbrandt

et al. [93]

National Lignocellulosic Abandoned lands,

brownfield, right-of-ways

8.5

Milbrandt et al. [93] looked at national level for estimating biomass based bioenergy

resources The study reported 8.6 million km2 of marginal land availability in contiguous

U.S., which is equivalent to 8.5% of total land area. They considered non-urban abandoned

lands, brownfield and transportation right-of-ways for marginal land estimation and

herbaceous crops (switch grass and miscanthus) as biomass feedstock. At the regional

level, Gelfand et al. and Gopalakrishnan et al. [91, 92] conducted assessment for the US

Midwest and Northeast, respectively, considering lignocellulosic biofuel based bio-energy

production system in primarily non-urban regions. At these scales, an important benefit of

using marginal land is that it does not diminish agricultural production and use of prime

farmlands and therefore can avoid ensuing impacts due to indirect land use change [94].

At local or municipal scales, several studies have conducted detailed mapping of marginal

land availability through parcel-level screening of land use and soil quality [33, 73, 95-98].

Typically, these local scale studies assessed bioenergy availability in meeting urban or

50

regional renewable portfolio standards, or as a question of urban self-reliance. Proper

quantification and geolocation of practically usable marginal land is critical to the

successful planning of urban bioenergy systems. Unlike regional or national level studies,

local scale studies are capable of incorporating parcel-level ownership and assessment

records, road and riparian boundaries, and socio-economic considerations that are relevant

for municipal authorities. Niblick et al. [33] incorporated several of these aspects in an

urban land study for Pittsburgh, PA, finding 35% of the city as marginal lands of limited

economic value that could be sustainably cultivated for sunflower based biofuel

production. Metal uptake was also considered in a subsequent study, as urban vacant lots

frequently have contaminated soils [99]. Finally, Grewal and Grewal [73] investigated

Cleveland, OH as a test case and assessed vacant lands equivalent to 40% of that city’s

total land area that could be suitably used to develop high-yield algae based biodiesel

production scheme.

Here we built on previous work for the assessment the marginal land and bioenergy

potential at urban scales, using Boston, MA USA as a case city. Additional attribute-based

and geospatial modeling tools were employed, and, to our knowledge, this is the first urban-

scale study to conduct a detailed parcel-level screening of layers for public and private

ownership, zoning, parcel size, slope, soil quality and shadow analysis. Both herbaceous

(miscanthus) and woody based (poplar and willow) energy crops were considered for

biomass and bioenergy yield estimation. These are believed to be the best yielding fast-

growing species in the Northeast U.S. [95]. Each has been shown to have a positive net

energy balance and can be effective in both fulfilling energy demand and mitigating climate

51

change [89]. The outcome of the study looks to provide policy makers and bio-energy

developers with a better understanding of the scale of urban bioenergy opportunities, while

also contributing to the larger research question of urban energy self-sufficiency.

52

3.2 Methods

Estimation of potential bioenergy yield from urban marginal land for Boston was

performed in several steps, represented in Fig. 3-1. A GIS-based site suitability analysis

was performed using ArcMap 10.3 (ESRI, Redlands, CA). Land parcel layer was used as

an input into GIS model and series of linear combinations of spatially referenced layers

were used as screening layer with some boundary conditions to identify land parcels that

can be suitably used. Here urban marginal land was defined as land that is not suitable for

primary agriculture, has a soil slope < 15% and has a minimum parcel size. Land features

that fit these criteria can be diverse especially within built up areas. For Boston, these

included (but may not be limited to) public and private vacant lands, residential and

commercial under-utilized areas, and degraded lands and fill. Finally, total biomass and

bioenergy potential was calculated using estimated urban marginal land, energy crop yield

information, and heat content. The input data sources, estimation approaches, and

validation techniques are explained in detail below for each step.

53

Fig. 3-1: Flowchart of modeling processes used for biomass mapping and bioenergy

assessment

3.2.1 Land use type screening

First, potential marginal land areas were identified using GIS site suitability model

developed for screening purposes (Fig. 3-1). This model consists of an input layer, erase

layers and the output suitable parcel layer (Table 3-2). The input layer was a 2013 record

of all City of Boston parcels. Several screening layers were joined and overlaid with the

input layer to exclude areas unsuitable for biomass cultivation because of existing

improvements or zoning restrictions. These screening layers included area occupied by

buildings, driving lots, parkways, protected areas (parks and recreational areas). These

layers were obtained from City of Boston Department of Innovation and Technology [12],

54

Department of Neighborhood Development [100], MassGIS [101] and Tufts University

GIS database [9]. The scale of all the layers was 1:1,000,000. Streets and highways were

excluded from the input parcel layer and so did not require subsequent exclusion.

3.2.2 Bio-geophysical screening

In this second step, the previously estimated suitable parcel layer was the input into a

separate GIS site suitability model developed for bio-geophysical analysis, consisting of

an input layer, three screening layers and an output layer. The bio-geophysical model

performed minimum area screening, soil quality, slope and shadow analysis. Based on

literature, a minimum parcel size for marginal land was determined of at least 93 m2 (1000

ft2) with a marginal slope of 15% or less to minimize runoff and erosion and for ease of

cultivation and harvesting [102].

Table 3-2. Description of parameters used to estimate urban marginal land

Spatial parameters Scale Projection

Land parcels 1: 1,00,000

North American Datum (NAD) 1983,

State plane Massachusetts, ft

Buildings 1: 1,00,000

Drive ways and

parking lots

1: 1,00,000

Water and

protected areas

1: 1,00,000

Soil types 1: 1,00,000

World Geodetic System (WGS) 1984

Soil slopes 1: 1,00,000

Extruded 2D

buildings

1: 1,00,000

3.2.2.1 Exclusion of parcels by soil quality and slope

Soil quality information for Boston was obtained from the United States Department of

Agricultural Natural Resources Conversation Services (USDA-NRCS) soil survey

55

geographic data base (SSURGO) [103]. The SSURGO database contains information

about soil type and slope information for the entire country. Based on SSURGO data, there

are four distinct classes of soil types (A, B, C and D) exist for Boston. USDA-NRCS has

three major farmland classifications: prime farmland, farmland of statewide importance

and not prime farmland [104]. Areas with soil listed as ‘not prime farmland’ are classified

as marginal (Table 3-3). The bio-geophysical model used spatial selection criteria to select

and erase parcels that are classified as ‘not prime farmland’ and soil slope <= 15% from

the input ‘suitable parcel’ layer.

Table 3-3. USDA-NRCS Marginal soils classification [104]

USDA NRCS classification Soil type Slope (%) Soil type

Prime farmland Silt loam/Shaly silt/loam/clay loam 0-8% A

Farmland of statewide

importance

Silt loam/Shaly silt/loam/clay loam 2-15% B

Not prime farmland Urban/industrial dump/gravel pits 0-35% C/D

3.2.2.2 Exclusion of areas by shadow analysis

Shadow analysis can be used to estimate the distribution and intensity of sunlight over a

geographic area for specific time duration. This tool accounts for how daily and seasonal

shifts of the sun angle, along with variations in elevation, orientation (slope and aspect),

and shadows cast by topographic features (buildings), affect the amount of incoming solar

radiation [105], which is of obvious importance for assessing site suitability for biomass

growth. Here, sun-shadow analysis was performed to identify the suitable parcels that

receive minimum six hours of sunlight during mid-crop season (Fig. 3-2a). Land parcels

with full sun exposure on the south, southeast and southwest with no buildings on those

sides were considered optimal, while parcels with shade during any part of the day were

considered a shadow area (Fig. 3-2b). ArcMap 3D sun-shadow analysis tools was used for

56

performing shadow analysis which requires extruded 2D buildings as an input features.

The analysis was conducted on extruded 2D buildings for July 21st between 8AM – 4PM.

July 21st was considered as the representative day for analysis that lies at the midpoint of

the seven-month growing season (April – October). The shadow map obtained was input

as an erase layer in bio-geophysical model along with soil quality and slope layers. The

marginal land layer was compared by laying with 15 cm high-resolution aerial imagery

obtained from MassGIS data layers. The purpose of the aerial analysis was to ground truth

the mapped parcels by considering criteria that could not be easily applied through GIS

analysis. The two major criteria examined through the use of aerial imagery were light

exposure and vegetation density. The parcel-by-parcel crisscross was performed to

determine the extent of similarities and additional parcels not identified in spatial analysis.

The resultant urban marginal land layer obtained from these models was further compared

with city of Boston’s new zoning code to ensure compliance with the approved zoning

requirements for urban agriculture [26].

Fig. 3-2: Shadow analysis examples: a) input extruded-2D building footprint and b)

shadow map

a)

b)

57

3.2.3 Biomass and bioenergy yield

As an urban coastal area Boston has a subtropical climate that could be favorable for

energy-crop production especially from late spring to early fall between April – October

[106]. A recent study conducted by the Massachusetts Department of Environmental

Resources and the Massachusetts Clean Energy Center reported a list of suitable bioenergy

crops with yield information for different areas within the states [107]. These included

herbaceous crops such as perennial switch grass, miscanthus, and woody biomass such as

short-rotation poplar and willow. Among these, willow, poplar and miscanthus have been

found most suitable for high yield and survival under changing climate scenarios [108],

including resilience to environmental conditions such as flash flood, low temperatures, salt

and alkali [107]. The crop yield information were obtained from Biofuel Ecophysiological

Traits and Yields database (BETYdb) [21]. BETYdb is a national database developed by

Energy Bioscience Institute (EBI) that maintains cellulosic biofuel crop species and yield

information for research, forecasting, and decision making support. For each crop type,

BETYdb has gridded crop yield data availability with 0.1o × 0.3o resolution that also uses

the USDA SSURGO soil quality database in estimating yields [109, 110].

For this study, crop yield values of three different energy crops for Boston area were

spatially interpolated from the BETYbd gridded dataset. Ordinary kriging method with no

transformation was chosen as method of interpolation. Later crop yields for each mapped

parcel were estimated by overlaying centroids of each parcel with the interpolated raster

pixels. The average yield values for each crop types are listed in Table 3-4. Total biomass

yield Mbio (t yr-1) was estimated by multiplying each individual marginal land parcel area

58

Ai (ha) with associated yield data Yi (t ha-1 yr-1) (eq.i) and summing. Bioenergy yield Ebio

was calculated using biomass yield (t yr-1) and heat content H of energy crops (eq.ii).

Higher and lower heating values were obtained for each crop (dry wt) [111]. Table 3-4 lists

the heat content values used.

𝑀𝑏𝑖𝑜 = ∑ 𝐴𝑖 ∙ 𝑌𝑖𝑖 (i)

𝐸𝑏𝑖𝑜 = 𝑀𝑏𝑖𝑜 ∙ 𝐻𝐿𝐻𝑉/𝐻𝐻𝑉 (ii)

Table 3-4. Energy-crop yield and heat content information

Energy crops Crop yield (dry t ha-1) Heating value (dry wt, GJ ha-1)

Lower (LHV) Higher (HHV)

Miscanthus 7.6 18.1 19.6

Poplar 15.8 17.7 19.7

Willow 11.2 18.4 19.7

3.3 Results

3.3.1 Marginal land resources in Boston

The marginal area estimation steps are displayed in panel format in Fig. 3-3. This shows

the a) input parcels, all the exclusion layers including b) buildings, c) water and protected

areas, d) driveways and parking lots e) parcels that are not suitable due to inferior soil

quality and inadequate sunlight and, f) final marginal land layers. As previously described

in Section 2, marginal land estimation was conducted in several steps using GIS based

screening model, bio-geophysical model and shadow analysis. Fig. 3-4 visualizes the total

marginal land availability in the city. Approximately, 2660 ha (26.6 km2) of potential urban

marginal land was mapped out representing 24% of total land area in Boston and have a

minimum of 93 m2 (1000 ft2) parcel size (Fig. 3-4). Of this urban marginal land, 15%

consists of vacant lands and the remainder is mostly under-utilized residential and

59

commercial area. There are approximately 283 ha (0.28 km2) of public and 116 ha (0.12

km2) of private marginal vacant lands available for immediate ground-level crop

production. Fully 70% of under-utilized land use mapped are residential neighborhood

areas that include house backyards excluding paved parking lots. The remaining 640 ha

(0.64 km2) are commercial under-utilized areas that could be categorized as marginal and

used for agricultural purposes. These findings are in the same range as estimated by Niblick

et al. for Pittsburgh, PA [33] and Grewal and Grewal for Cleveland, OH [73].

Marginal land resources are concentrated in southern part of the city. Due to proximity to

water resources, sufficient sunshine and low population density relative to the rest of the

city, the southern neighborhoods of Dorchester, Hyde Park, Mattapan, Roslindale, and

West Roxbury are promising areas for growing dedicated energy crops. These areas have

also been identified by the City of Boston as potential location for developing food-based

urban agriculture [94]. On the other hand, the highly urbanized and densely populated core

of Boston was not found suitable due to the high proportion of impervious areas, limited

parcel sizes, large shadow areas associated with buildings, and proximity to the coast;

however, this part of the city has some potential for rooftop urban agriculture.

60

Land parcels (Input layer)

Buildings

Water and protected areas

Driveways and parking lots

Unsuitable parcels from slope and shadow analysis

Urban marginal land (Output layer)

Fig. 3-3: Marginal land estimation in Boston

e)

b)

a)

c)

d)

f)

Erased

layers

61

Fig. 3-4: Available marginal land in Boston

3.3.2 Biomass and bioenergy potential

Evaluation of bioenergy potential of energy crops in Boston are shown in Table 3-5. These

values should be seen as an upper bound, as not all marginal land available in the city may

be practically be used to cultivate energy crops. Community acceptance, soil

contamination, safety, and traffic considerations could reduce the amount of marginal land

that is practically usable [86].

Boston

62

Table 3-5. Biomass and bioenergy yield

Energy crops Biomass

yield (t yr-1)

Primary

Bioenergy yield

(TJ yr-1, HHV)

Energy

District

heating

(TJ yr-1)

Energy

CHP

(TJ yr-1)

Miscanthus 20,300 398 338 240

Poplar 42,100 830 705 498

Willow 29,800 586 500 352

It was estimated that Boston’s marginal land could be used to produce nearly 42,130 tons

of high yield short rotation poplar biomass in a regular growing season. Potential biomass

yield for willow and miscanthus were found to be nearly equal. Short rotation poplar was

found to be the most promising energy crops based on heating value calculation. This

amount of biomass can potentially yield up to 830 TJ yr-1 (230 GWh yr-1) of primary energy

on average (using high heating values). For context, this is equivalent to ~0.6% of State of

Massachusetts’ primary energy demand for 2012 [97]. This energy could be used directly

in Boston itself for heat and power production, particularly in high-density commercial or

industrial areas. For example, the existing Medical Area Total Energy Plant (MATEP) is a

natural gas fired district heating and power generation facility that serves the Longwood

Medical Area and supplies 2.1 GWh annually, only 2.5% of the potential total heat

available in the city from biomass [70]. With an average 85% conversion efficiency for a

biomass district heating plant that used poplar, this amount of primary energy could yield

up to 705 TJ yr-1 (195 GWh yr-1) of useful energy for heating (Table 3-5) [112]. Also, for

a biomass driven decentralized combined heat and power (CHP) facility with 60% energy

efficiency that used miscanthus as a feedstock could yield up to 240 TJ yr-1 (66 GWh yr-1)

of energy [112].

63

3.3.3 Spatial Validation

Validation of geospatial model was conducted to ensure accuracy of identified parcels and

check robustness of the system. To accomplish this, eight suitable parcels were randomly

selected from the 60,000 identified marginal land parcels using ‘Create random point’ tools

of ArcMap. The marginal land layer (Fig. 3-4) was used as an input and mask layer in the

validation model and the python based geo-processing tool generated 20 random points. 8

out of 20 points coincide with centroids of the parcels identified (Fig. 3-5). The resultant

parcels were further layered with high resolution (15 cm) ortho imagery for comparison

and validation.

Fig. 3-5: Randomly selected urban marginal land parcels (Scale 1: 600)

3.4 Discussion and Implications

The preceding analysis shows that urban production and use of bioenergy crops have the

potential to contribute to a city's clean energy economy and supplement existing energy

infrastructure. Multiple conversion pathways are possible. Bioenergy crops can be

64

combusted directly in a dedicated biomass power plant; the city of Vienna has recently

installed Europe’s largest such plant with a capacity of 53 MW of combined heat and power

[3]. MATEP and similar plants could also take advantage of local biomass feedstock by

adding a biomass boiler or co-firing. An urban bioenergy production facility can be

operated as an independent energy generation facility, or as part of an integrated district

heating network [113]. Plants operating continuously are adaptable to change in biomass

resources, thus providing predictable and reliable base power [114]. Bioenergy crops could

also be pyrolized to syngas, converted to liquid fuels, or directly combusted in small-scale

commercial or residential heating systems [115-117].

Total bioenergy production could be increased if the definition of marginal land were

expanded to include additional types of under-utilized urban land, particularly on

commercial properties. Marginal land categories of saline lands, abandoned or degraded

forests were excluded from the present study scope, but may be relevant for other urban

areas. Increasing the area of cultivated marginal land or improving crop yields will increase

the potential bioenergy yield. Marginal land yields are by definition lower than for prime

agricultural lands, but may be improved by optimizing planting and maintenance processes

and climatically appropriate crop choices [92].

In addition to considerations of urban energy supply, there are also non-energy benefits of

urban bioenergy schemes. Growing bioenergy crops can promote awareness of

environmental stewardship, improve soil quality and increase soil carbon, reduce

stormwater pollution, support urban biodiversity and habitat, and reduce urban heat island

65

effect [81, 107]. Urban biomass and bioenergy can improve self-sufficiency and physical

resiliency in urban areas, to help protect from risks caused by natural hazards and

infrastructure failures [73]. Local bioenergy production can be a means of community-

based economic development, by creating new jobs and promote innovative clusters of

related businesses [94].

There are obvious challenges to implementing an urban bioenergy scheme on a large scale.

While outside the scope of the present work, the logistics of cultivating, harvesting,

transporting, and storing biomass combined from potentially hundreds of urban plots will

be complex. Previous works describing modeling biomass supply chains and logistics may

be applied, particularly optimization models constructed for urban systems [79, 104].

Previous research has found that to implement a robust local urban bioenergy production

it is critical to locate district heating networks and biomass cogeneration facilities in

proximity to feedstock collection facilities [95]. Distributed bioenergy production may

also suffer from diseconomies of scale. Boston’s land values and labor costs are high,

which will impact capital and operating costs and project financing. A simple cost

estimation by Chin et al. [94] revealed that a half-acre (2000 m2) urban farming parcel

requires capital cost of approximately $10,000 that includes equipment purchase, sales, and

marketing. Operating costs for an urban farm of the same size are estimated to be $5,000-

$10,000 per year with gross revenue of $60,000. Present-day costs may well be much

higher in Boston. Land access and ownership are important considerations, especially

considering that 85% of the marginal lands identified are in private hands and must be

either cultivated by residents or leased to urban farmers [106]. It may also be difficult to

66

obtain permits for cultivation or conversion facilities, and there are currently no zoning

ordinances specific to bioenergy cropping.

3.5 Conclusions

Bioenergy production on urban marginal lands have several benefits, such as fulfilling

partial energy demand, increasing resilience and mitigate climate change impact by

lowering GHG emission. Based on reasonable assumptions, precise data acquisition and

analysis methods, the following outcomes have been summarized:

For Boston, total area of marginal land was estimated up to 2660 ha (26.6 km2) that can be

suitably used for bioenergy production and represents 24% of total land area. This area

consists of 85% of residential and commercial unutilized areas and 15% vacant lands.

South and Southwestern part of the city were found most suitable for energy crop

harvesting.

Three species of bioenergy crops were found suitable namely herbaceous miscanthus, short

rotation poplar and willow that could be planted in different experimental areas throughout

the city and have great potential for bioenergy development. However site-specific

experimental crop harvesting is necessary to validate the suitability.

The maximum potential bioenergy yield is approximately 830 TJ for short rotation poplar

if 100% of the marginal lands identified are used for growing the energy crops. Energy

yield obtained from both biomass district heating plant and CHP plant indicate promising

67

outcome. However even partial development of bioenergy crop production on urban

marginal land could fulfill significant amount of the city’s heating demand during winter

season.

The present study confirms that the total potential supply of urban bioenergy is significant

even for a densely settled city such as Boston. While Boston was the target of the present

study, these methods can in principle be applied to any city or metropolitan area. Urban

bioenergy schemes may be more appropriate for cities undergoing depopulation and with

a higher proportion of vacant marginal lands and lower economic rents than Boston. In

practice, there exist numerous logistical and financial obstacles to using urban bioenergy

at large scales, but also many potential co-benefits to communities, including measures of

energy security, economic development, and environmental quality that could potentially

be realized.

68

Chapter 4 A GIS-based Assessment of Regional Scale Bioenergy

Production Potential on Marginal and Degraded Land

This study presents an application of the GIS-based model developed in Chapter 3, for site-

specific bioenergy production potential on urban marginal lands available in the Eastern

Massachusetts region. The ‘marginal’ land use category includes vacant and abandoned

lands, inner city underutilized areas, and degraded lands. Three different energy crops;

miscanthus, poplar and willow were considered as biomass feedstocks for bioenergy

production. A total area of marginal land was estimated up to 71,200 ha (712 km2) that can

be suitably used for bioenergy production and represents 20% of the MAPC land area.

Among 102 cities assessed in the study region, Boston, Marshfield, Franklin, and Concord

were found to have the greatest bioenergy potential in terms of total marginal land area

availability. Fast-growing poplar was found as most suitable high yield bioenergy crops

for this region. Bioenergy potential calculation revealed that for short rotation poplar, this

amount of biomass can potentially yield up to 22 PJ (HHV) of yearly primary energy for

this region that can be used for heat, conversion to fuels, and/or electricity production. This

is equivalent to ~15% of Massachusetts primary energy demand for 2012. The outcomes

of this study are in line with previous work that evaluated marginal land availability in

Boston and other major Eastern U.S. cities, and confirm the accuracy of the spatial model

constructed in Chapter 3.

69

4.1 Introduction

4.1.1 Regional assessment of marginal land

The United States has approximately 120 million ha of marginal land area available that

includes federally funded brownfields, closed landfills and abandoned lands [118].

Approximately 67% of this land area is within the administrative boundary of urban

metropolitan regions. A typical urban community’s reliance on conventional fossil energy

is not only carbon intensive but also vulnerable to supply disruptions due to environmental,

economic and geopolitical factors. In the Northeastern United States, heating demand has

increased significantly in last decade [107]. To reduce supply related uncertainty and

increase energy self-reliance, many cities have started to focus on the sourcing part of its

energy from leveraging local renewable energy resources [119]. However, inner cities are

often densely populated with limited land resources suitable for large-scale renewable

energy production. Research suggests growing energy crops on marginal or degraded land

close to larger metropolis can be a viable solution [33, 35]. Suitable suburban land can be

repurposed to grow high-yield crops that can be used as biomass feedstock for bioenergy

production. Successful implementation however, warrants more site-specific investigation.

The precise definition of marginal land may vary depending on prior uses and geographical

considerations such as location and scale. While different definitions exist [35, 99],

marginal lands can be broadly categorized as “lands that are not suitable for food-based

agriculture and have limited economic potential for fulfilling other ecosystem services”

[120]. The unsuitability can be attributed to poor physical and chemical soil properties

and/or susceptibility to erosion [121]. Several land use categories could be considered as

70

marginal. Furthermore, there is a lack of consensus regarding choice of suitable energy

crops to grow on urban marginal lands. All energy crops require favorable climatic

conditions and a specific amount of water, nutrients and suitable growing conditions [122].

Some feedstocks, however, are more resilient than others and can grow successfully on

marginal lands and under changing climate scenarios [122].

Past studies have claimed that using marginal lands to produce bioenergy on a global scale

is unfeasible due to lack of economic incentives and threats to biodiversity and

conservation areas [123, 124]. However, marginal production levels become more viable

when considered on the local and regional level as certain regions have high land or

biomass availability, existing infrastructure, and suitable population densities [86].

Regional scale study has the advantage of assessing city by city evaluation of land access,

imperviousness, and soil quality to precisely identify areas suitable for specific crop type.

Every geographic and climatic region is different, and therefore spatial analyses are critical

and chosen technique to assess land use suitability for sustainable bioenergy production

and infrastructure systems development.

This study expands the scope of geospatial model developed by Saha and Eckelman. 2015

[35]; to assess large scale urban region surrounding Boston for marginal land availability

for energy crop production. The model description and outcome are also described in

Chapter 3 of this dissertation. Even though Boston’s identified marginal land is

approximately one quarter of its total [35], due to high land prices and competition with

other current or potential economic activities, access and logistical factors, it is unlikely

71

that all of these land parcels can suitably be used for cultivation. But crops can be grown

outside the city and still be used for inner city bioenergy production as the transportation

requirement and costs are low [88]. Here, total theoretical marginal land availability and

bioenergy potential for 102 cities and towns in Eastern Massachusetts were assessed.

4.2 Methods

4.2.1 Marginal Land Assessments

As mentioned, the urban marginal land assessment model developed in Chapter 3 was

applied in this study for a larger geography that includes 102 cities and towns located in

Eastern Massachusetts. Together, these 102 cities fall within the planning jurisdiction of

Metropolitan Area Planning Council (MAPC), a state agency (Fig. 4-1). Because of its

strategic location, population density and economic importance, the MAPC region is

considered one of the major energy demand centers of the Eastern United States [88]. The

total land area of MAPC region is 330 km2 and includes more than 3 million inhabitants

[89].

Fig. 4-1: Study area (MAPC region)

72

Here, the general geospatial modeling framework developed by Saha and Eckelman [35]

was used for MAPC marginal area assessment. Based on the literature, marginal land was

characterized as specific land use types with marginal soil quality and soil slope. The most

recent land cover data obtained from USGS national land cover database were used as an

input in the screening model [125]. Here a national scale data layer was used in order to

make it easy to perform a similar analysis for a larger geography. The USGS land cover

layer is a nationwide, seamless digital dataset covering wide range of land cover and land

use types, created using semi-automated methods, and based on 0.5 m resolution digital

orthoimagery. The specified land use classification is attributed in the data layer by two

fields: land use description (LU05_DESC) and land use code (LUCODE) (Table A1).

Land cover types defined as ‘residential and commercial underutilized areas, abandoned

agricultural lands, landfills, and junkyards available within city areas are considered as

marginal land.

The input land cover layer was subjected to screening by series of spatially referenced

layers, starting with minimum area (9.3 m2) screen followed by the impervious layer, water

and conservation land, and soil quality respectively. The remaining screening data layers

were obtained from the MassGIS website [126]. The aggregated impervious surface layer

consists of roads, railways, driveways, parking lots, and other areas where soils are

inaccessible [90]. For biogeophysical analysis, land parcels with soil slope <15% to avoid

erosion and soil quality type C or D based on the USDA NRCS soil classification were

selected [35]. The resultant marginal land area information was later used for biomass and

73

bioenergy yield calculation. The spatial projections and coordinate systems used are

described in Table 4-1.

Table 4-1. Description of parameters used to estimate marginal land area

Spatial parameters Scale Projection

Output layer 1: 1,00,000 Massachusetts State Plane Projection System,

ft

Land cover 1 meter Albers conical equal area

Impervious layer 1 meter

Water and

protected areas

1: 1,00,000 North American Datum (NAD) 1983, State

plane Massachusetts, ft

Soil types 1: 1,00,000

World Geodetic System (WGS) 1984

Soil slopes 1: 1,00,000

Extruded 2D

buildings

1: 1,00,000

Fig. 4-2 summarized the step-by-step spatial overlay analyses in panel format. These

include the input land cover layer (Fig. 4-2a) and overlaid screening layers (Fig 4-2b, 4-

2c, 4-2d). The land cover layer consists of multiple land use types represented by different

colors. In greater Boston, land use mostly consists of high intensity residential and

commercial areas (red), low intensity residential and commercial areas (yellow) and forest

(green), respectively (Fig. 4-2a). The detailed geoprocessing steps used and underlying

assumptions are also described in the Methods section of Chapter 3.

74

a) Land cover

b) Impervious layer

c) Water and protected areas

d) Unsuitable parcels from soil

quality and slope analysis

Fig. 4-2: Urban marginal land estimation

4.2.2 Biomass and bioenergy yield

In this study, the theoretical biomass yield was estimated for marginal land across the entire

MAPC region. Three major bioenergy crops miscanthus, poplar and willow were

considered as biomass feedstocks as identified by Massachusetts Department of Energy

Resources [35, 107]. Next, average biomass yield and energy density factors were used to

estimate the total primary bioenergy yield available in an upper-bound, 100% marginal

75

land utilization scenario. The average yield values for each crop type are listed in Table 4-

2, based on the present climatic condition for the region. Total biomass yield Mbio (t yr-1)

was estimated by multiplying each individual marginal land parcel area Ai (ha) with

associated yield data Yi (t ha-1 yr-1) (eq.i) and summing. Bioenergy yield Ebio was calculated

as the product of total using biomass yield (t yr-1) and heat content of energy crops (eq.ii).

𝑀𝑏𝑖𝑜 = ∑ 𝐴𝑖 ∙ 𝑌𝑖𝑖 (i)

𝐸𝑏𝑖𝑜 = 𝑀𝑏𝑖𝑜 ∙ 𝐻𝐿𝑉/𝐻𝑉 (ii)

Table 4-2. Energy-crop yield and heat content information [40]

Energy crops Crop yield (dry t ha-1) Heating value (dry wt, GJ t-1)

Lower (LHV) Higher (HHV)

Miscanthus 7.6 18.1 19.6

Poplar 15.8 17.7 19.7

Willow 11.2 18.4 19.7

4.3 Results and Discussion

4.3.1 MAPC Marginal land

The marginal land assessed are presented in Fig. 4-3. As previously described in Section

2, suitable area estimation for energy crop harvesting was conducted in several steps using

GIS based screening model, bio-geophysical model and shadow analysis. Approximately,

71,200 ha (712 km2) of potential marginal land areas were identified representing 20% of

the total land area of the entire MAPC region (Fig. 4-3). Of this land area, 75% consists of

under-utilized residential and commercial areas that include vacant lots and yards. The

remaining 25% are abandoned cropland, landfill areas, and junkyards that could be

categorized as marginal and repurposed for agricultural purposes. Among 102 cities and

76

towns assessed, Boston, Marshfield, Franklin, and Concord were found to be the four most

suitable cities or towns in terms of total marginal land area availability (Fig. 4-4a, 4-4b, 4-

4c, and 4-4d). Together, these four cities account for 12% of total marginal land assessed

for MAPC region (Table 4-5). For Boston, the available area estimated was 2,910 ha, or

which is 23% of total land area. This value is within one percentage point of the result

estimated by Saha and Eckelman for Boston [35]. This shows successful validation of

spatial model construction in Chapter 3 for a larger geography and using larger scale

slightly different starting input data layer. The city of Reading was found most suitable on

the percentage land intensity basis, with 37% of total land area followed by Needham

(33%), Newton, (32%) and Danvers (31%), respectively (Table A2). Because of close

proximity to Boston, these municipalities could in theory supply bioenergy for the city

center without requiring long hauling distances. Although it contains the largest number of

marginal parcels, Boston was ranked 25th on a percentage basis.

77

Fig. 4-3: Available marginal land in MAPC cities

Boston Marshfield

a) b)

Boston

Marshfield

Franklin

Concord

78

Fig. 4-4: Available marginal land in four municipalities

Table 4-3. MAPC marginal land

Rank City or town Marginal land

area (ha)

Total land

area

(ha)

Marginal %

of total land

area

MAPC region 71,150 362, 419 20%

1 Boston 2,910 12,500 23%

2 Marshfield 2,110 7,446 28%

3 Franklin 1,850 6,988 27%

4 Concord 1,568 6,684 24%

4.3.2 Biomass and Bioenergy Yield

Marginal lands available in the MAPC region could be used to produce nearly 1.1 million

tons of high yield short rotation poplar biomass in a regular growing season (Table 4-4).

Potential biomass yields for willow and miscanthus were found to be 0.8 million tons and

0.5 million tons, respectively. Short rotation poplar was found to be the most promising

energy crop based on heating value as well. Poplar biomass could potentially yield up to

22 PJ yr-1 of primary energy (using high heating values) in a scenario of 100% marginal

land utilization. This is equivalent to ~15% of Massachusetts primary energy demand for

Concord Franklin

c) d)

79

2012 [97]. Boston’s marginal land could alone produce up to 46,000 tons of poplar, 32,600

tons of miscanthus and 22,200 tons of willow, respectively. This may be attributed to

Boston being the largest city with higher amount of land parcels with marginal soil type

(C/D). The results are also closely in line with previously reported values in Saha and

Eckelman [35].

Table 4-4. Biomass and bioenergy yield

Geography Energy crops Biomass

yield (t yr-1)

Primary

Bioenergy yield

(TJ yr-1, HHV)

MAPC Miscanthus 541,000 10,600

Poplar 1,120,000 22,200

Willow 797,000 15,700

Boston Miscanthus 22,100 434

Poplar 46,000 906

Willow 32,600 641

Marshfield Miscanthus 16,000 314

Poplar 33,300 657

Willow 23,600 465

Franklin Miscanthus 14,100 276

Poplar 29,200 576

Willow 20,700 407

Concord Miscanthus 11,900 234

Poplar 24,800 488

Willow 17,600 345

This result shows not insignificant potential for regional bioenergy resources to supply

energy for the highly urbanized and energy-intensive Boston metropolitan area that

includes densely populated and industrialized cities like Cambridge and Somerville.

Introduction of local bioenergy resources and adaptation of existing energy infrastructure

to accommodate biomass could be beneficial in providing local renewable energy (mostly

likely for heating) for one of the nation’s largest educational, health care, and high-tech

clusters.

80

4.4 Conclusion

The geospatial model constructed for assessing urban marginal land in Chapter 3 was

successfully used to validate the method and corroborate results for a different larger scale

geography, using larger-scale data layers. The following outcomes have been summarized:

For MAPC region, total area of marginal land was estimated up to 71,200 ha (712 km2)

that can be suitably used for bioenergy production and represents 20% of the total land

area. This area consists of residential and commercial land parcels, vacant lots, abandoned

cropland, landfill, and junkyards. Among 102 cities assessed, Boston, Marshfield,

Franklin, and Concord were the most suitable urban locations in terms of marginal land

area availability. Three species of bioenergy crops were modeled for total biomass and

bioenergy production, namely herbaceous miscanthus, short rotation poplar, and willow.

The maximum potential bioenergy yield is approximately 22 PJ for short rotation poplar if

100% of the marginal lands identified are used for growing the energy crops. However

even partial development of bioenergy crop production on urban marginal land could fulfill

significant amount of the region’s heating demand during colder days.

The present study assesses the physical potential of bioenergy crops production on regional

scale marginal land and validates the previously developed model outcome for a larger

scale densely settled Eastern Massachusetts. In practice, there exist numerous challenges

and trade-offs when pairings lands and energy resources. Existing popular technologies

like solar production and wind farms facilities will likely compete for the same marginal

81

land. Therefore, land optimized technology pairing can create opportunities for increased

energy production, especially in the case of co-located energy crops and wind farms. Also,

not all available lands can ideally be used for bioenergy production. Logistics such as

transportation, transmission and distribution lines, and access to water and other nutrients

also need to be considered. Optimal use of land resources, biomass feedstock, and energy

production technologies will allow marginal lands to play an increasingly vital role in

fulfilling regional energy demands in a sustainable manner.

82

Chapter 5 Geospatial Assessment of Urban Agriculture Potential in Boston

Urban parcels can be leveraged for developing a local urban food system by growing high

yield food plants. Here, a remote sensing and GIS-based modeling framework was created

to assess Boston’s available area for urban farming, including both rooftop and ground

level areas. Geoprocessing and spatial analysis tools were used to process geographic data

layers for zoning, ownership, slope, soil quality, and adequate light availability. Surface

slope (roof pitch) was determined for all buildings in the city through the creation of a

digital surface map from remotely sensed LiDAR data. Potential parcels from ground level

public and private vacant lands and underutilized residential and commercial areas were

mapped using publicly available datasets. Approximately 922 hectares of rooftop and 1,250

hectares of ground level parcels have been identified, representing 7.4% and 10% of total

land area in Boston, respectively. Finally, food yield values for common urban agricultural

crops were used to estimate the city’s food production potential from the identified parcels.

Despite Boston’s density, the mapped areas have potential to produce enough fresh fruits

and vegetables for 30% of Boston’s population. The study outcome was compared with

results from other regions in North America that might be able to fulfill partial food demand

leveraging local resources.

83

5.1 Introduction

5.1.1 Urban agriculture

Food and energy security are pressing concerns for municipalities in the face of growing

global urban populations. Urban inhabitants represent the majority of global food and

energy demand, with nearly 50% of the total global populations live in cities. Large-scale

industrial agriculture, while inarguably efficient, can incur significant environmental costs,

including deforestation, depletion of cropland, soil, and water pollution, and biodiversity

loss [127, 128]. It is estimated that food in the United States travels an average of 1,500

miles from the farm to plates [129], while food grown in urban areas is coincident with

demand centers. Although dense cities cannot be self-sufficient in food production, urban

food farming could increase food security and help to address various urban environmental,

economic, and social challenges.

Urban agriculture is commonly referred to as the practice of growing, processing and

distributing food in urban and peri-urban areas. Becoming at least partially self-sufficient

through urban agriculture is one way to increase the resiliency of food and energy systems

through diversification of supply, and can bring multiple co-benefits including enhanced

food security, enriched landscapes, local economic development, and improved

environmental quality [130]. Urban agriculture can benefit the local environment through

improvements to urban air quality, increasing rates of carbon sequestration, modulation of

urban heat islands, and mitigation of water pollution problems associated with stormwater

runoff [131]. Producing food locally can also avoid environmental impacts associated with

long-distance food distribution and losses [132]. At the same time, potential economic and

84

social benefits can include employment and local economic activity; redevelopment and

productive use of blighted, marginal urban areas that are frequently located in low-income,

underserved communities; and, depending on the area of the country, noise abatement,

food access and nutrition, and community education [99, 131, 133].

An important preliminary step of planning for local food system development is to estimate

the total available area and potential production volumes for urban agriculture. Urban food

farming can be implemented using both ground level and rooftop areas. Especially in

densely built-up areas where vacant parcels are scarce and with a large number of buildings

available, rooftop farming can be an attractive supplement to more conventional urban

farms and community gardens, but is still a niche form urban farming that has yet to gain

popularity on a large scale. However, there exists a lack of consensus in defining the extent

of a local food system. According to one US Department of Agriculture (USDA) definition,

a local urban food system may consist of an area as large as 644 km2 [134]. Many

municipalities consider their administrative boundaries for delineation of their local food

systems. ‘Foodshed’ is another common framework for regional food sourcing and

distribution system. Kloppenburg et al. suggest that “the foodshed can provide a place for

us to ground ourselves in the biological and social realities of living on the land and from

the land in a place that we can call home, a place to which we are or can become native.”

[135]

Compared to efforts in the developing world, cities in North America only recently have

begun to pay attention to the integration of urban agriculture and land use planning [136],

85

with notable progress in New York, San Francisco, Portland, and Vancouver [137]. Boston

is a comparative latecomer in this regard, though up until the 20th century it was one of the

largest agricultural centers in Massachusetts [138]. Following the creation of the City’s

Office of Food Initiatives in 2010 has made significant efforts in focusing attention on land

use planning that includes agricultural uses. In Boston, the Mayor’s Office of Food

Initiatives supports urban agriculture, because it “improves access to fresh, healthy, and

affordable food, with decreased transportation costs and lower carbon emissions.

Furthermore, new farming endeavors can bring communities together, empower small

entrepreneurs, and improve access to fresh food for all Bostonians.” [139] To support

commercial-scale urban agriculture, the City in 2013 passed Zoning Article 89 for urban

agriculture, providing necessary guidelines about urban farming implementation and

municipal support for local food distribution [140]. This was followed by a city-wide

visioning document based on extensive stakeholder engagement, published in 2015 [141].

5.1.2 Spatial analysis

Spatial research methods are central to understanding and evaluating the different

components of local food systems. On the supply side, spatial analysis has primarily

included urban food system mapping and land suitability studies. Notable efforts have been

taken in foodshed analysis in various locations over last several decades [142-144].

Different analytical, numerical, statistical, and artificial intelligence approaches have been

investigated to assess urban agriculture viability and potential extent. Table 5-1

summarizes the study characteristics for different cities in North America. GIS-based

86

analysis using tax assessor or land use layers has been most commonly employed,

occasionally coupled with remote sensing data products for further screening or validation.

87

Table 5-1. Review of studies on urban agriculture potential in North America

Author(s) Location Agriculture

type

Used

remotely

sensed data

% of total

municipal

land area

Comments

Ackerman et al. [145] New York, NY Ground level Yes 2.6% Vacant lots

Ballmer et al. [146] Portland, OR Ground level Yes n/a 47 sites < ¼ acre

242 sites > ¼ acre

Berger [147] New York, NY Rooftop Yes n/a N. Brooklyn Industrial

Business Zone

Colasanti and Hamm [148] Detroit, MI Ground level No 5.5% Vacant lots

Eanes and Ventura [149] Madison, WI Ground level No 2.2% Vacant lots

Horst [150] Seattle, WA Ground level Yes 0.02% Public vacant, unused,

or excess right-of-way

Kaethler [151] Vancouver, BC Both Yes 0.29% Public vacant, unused,

or excess right-of-way

Kremer and DeLiberty

[152] Philadelphia, PA Ground level Yes 7.8%

Residential lots

Grewal and Grewal [153] Cleveland, OH Both No

37%;

6.9%

5.8%

Ground level;

Vacant lots;

Rooftop Industrial

Habermen et al. [154] Montreal, QB Both No n/a Vacant lots, residential

yards, industrial roofs

MacRae et al. [155] Toronto, ON Ground level Yes 1.3% Two areas zoned for

agriculture

McClintock et al. [156] Oakland, CA Ground level Yes 4.3% Public land and private

vacant lots

Ryerson [157] Atlanta, GA Ground level Yes 0.01% Single candidate site

Taylor and Lovell [81] Chicago, IL Ground level Yes 0.04% Existing community

and private gardens

Remote sensing imagery has typically been processed manually rather than through

automated means, either to detect existing urban agriculture activity [81] or to identify

potentially suitable parcels [147, 156]. For example, on ground level suitability, Kremer

and DeLiberty [152] analyzed the city of Philadelphia using a procedure integrating

classification techniques for remotely sensed land cover data with GIS data layers

including physical and administrative information about Philadelphia’s parcels, buildings,

and zoning. The authors used a supervised pixel-based classification method and discussed

88

the importance of comparing it with object-based classification method for the purpose of

detecting urban yards, and including shape, size, and shade for refining the detection. On

rooftop suitability, Berger [147] looked at potential site suitability of rooftop agriculture in

New York City, which presents many challenges to ground level urban agriculture, primar-

ily through the lack of affordable open space for commercial urban farms due to the densely

built landscape. The model utilizes publicly available datasets to identify the buildings with

the greatest potential for rooftop farms, greenhouses, or intensive green roofs (including

structural considerations), combined with aerial imagery for validation and estimation of

usable area. While not focused on food, Kodysh et al. [158] used light detection and

ranging (LiDAR) data and a geographic information system to conduct a rooftop suitability

study for PV solar installation, considering screening parameters such as elevation, slope,

and shadow effect that influence solar intensity on the building roof. Such LiDAR-based

tools developed for the solar energy industry have also been applied to consider suitability

for rooftop agriculture, such as the NYC Solar Map used by Berger [147].

5.1.3 Connections to self-sufficiency and resilience

Resilience is now a central concept in city planning; Boston was one of the first cities in

the US to appoint a Chief Resilience Officer in 2015. Also in 2015, the City released the

first Boston Food System Resilience Study, describing in mostly qualitative terms some of

the potential risks facing the city’s residents in the case of disruptions [159]. The

vulnerability of Boston’s food system to natural disasters is of interest to decision makers,

due in part to recent multiple extreme events such as Hurricane Sandy in 2012, and the

record-breaking snowfall in the early months of 2015.

89

The effects of climate change on agriculture are being assessed in many countries, but the

ability of the existing food systems to recover from natural extremes is not generally

considered in most metropolitan resilience planning [160]. The food distribution network

constitutes a critical ‘lifeline’ system for cities. Food distribution systems that are disrupted

by disasters may not return to normal operations for long periods, which could cause

significant food availability and food access issues. Recent studies have suggested local

and regional food systems as a partial solution to these issues [101, 134, 161], though of

course these may also be disrupted during extreme events. It is estimated that 90% of the

food consumed in Boston is produced outside of the region. While growing, urban

agriculture is only a small portion of local food supply. However, there is significant

interest in Boston to expand local food production and processing in the city and the

metropolitan region. Policies that enable or incentivize urban agriculture may decrease

food shortage risks associated with local natural disasters, while in addition increasing

economic resilience by supporting local food cultivation, processing, and distribution

companies that create local jobs [160].

Following in this paper, we describe an automated procedure for combining GIS and

remote sensing data to determine the suitable areas for urban agriculture at ground level

and rooftops in Boston. Additional attribute-based and geospatial modeling tools were

employed, and to our knowledge, this is the first urban-scale study to conduct both rooftop

and ground level productive area mapping with the detailed exclusion of ownership,

zoning, parcel size, slope, soil quality and shadow analysis for an entire city. This modeling

framework can be extended to other cities with similar data availability and can be used to

90

provide detailed estimates of urban agricultural potential, as well as to identify or prioritize

specific parcels or neighborhoods for more in-depth analysis. In addition, the outcome of

the study looks to provide urban planners and food system stakeholders with additional

information about urban farming scale, while also contributing to the larger research

question of urban food self-sufficiency.

91

5.2 Methods

5.2.1 Study area and datasets

The current study is focused on developing a GIS-based model to conduct site suitability

analysis of urban farming on rooftops and ground level areas in Boston. Both currently

utilized agricultural sites and potentially suitable sites were mapped. The Boston area was

chosen because the city has a diverse range of buildings and land use types that could be

suitable for urban food production and because of supporting policy on urban agriculture

and food security. Currently, there are nearly 200 community gardens in operation for

vegetable and fresh fruit production in Boston. However, there are very few private

operations in the City currently for commercial-scale food production [94]. Specifically,

there are currently six commercial urban farms in Boston operating on 14 plots throughout

the city named: “Allandale Farm, City Growers, Corner Stalk, The Food Project,

Katsiroubas Brothers Fruit and Produce, and ReVision Urban Farms” [94].

This study was divided into three distinct steps (Fig. 5-1). First, a geospatial model was

developed to map the potential flat rooftop and ground level areas that can be used for

urban farming, using a variety of public geospatial data layers and remote sensing data,

described in detail below. Second, the total theoretical fruit and vegetable yield from these

potentially productive lands was assessed. Finally, validation analysis was conducted to

check the model accuracy by comparing with records from the latest Tax Assessor’s

database for the city that includes characteristics of all buildings and parcels. The input

data sources, estimation approaches, and validation techniques are explained in detail

below for each step.

92

Fig. 5-1: Boston rooftops and ground level parcels extraction steps

The spatial extent of study area was easting 656,167 feet and northing 2,460,625 feet (Fig.

5-2). The projected coordinate and projection system of the spatial data frame was NAD

1983 State Plane Massachusetts Mainland FIPS 2001 Feet and Lambert Conformal Conic

respectively. The required datasets for analysis were primarily available from MassGIS,

the state GIS clearinghouse for the Commonwealth of Massachusetts [162].

5.2.2 Mapping flat rooftops

Identification of flat rooftops was conducted using several geoprocessing, spatial analyst

and raster analysis tools in ArcMap 10.3 software package (ESRI Inc., Redlands, CA). The

overall modeling process was then automated using ArcMap Model Builder and python

93

script. The flowchart in Fig. 5-1 summarizes the steps involved in rooftop mapping. Here,

suitable flat rooftop for urban farming is defined as roof areas that are located within

permitted zones for urban agriculture according to a recent zoning ordinance [140], with a

minimum roof surface area (1000 ft2), and a maximum building height (100 ft), and surface

slope (< 5o). Typically, rooftop conditions above 100 ft are assumed as being less

hospitable to plants due to high winds, and there are logistical and safety concerns with

access for people and supplies [147]. The first step was to overlay building footprints and

zoning layers. The most recent Boston building footprint data layer was obtained from the

City of Boston GIS Datahub [163]. Surface area and height constraints were then applied

to delineate the suitable buildings. Selections by attributes and Selection by location, Clip,

and Erase operations in ArcMap were used for delineation functions. It was assumed

throughout that the roof area of buildings is identical to the building footprint. The building

footprint area is automatically calculated as the shape geometry in ArcMap. The model

was allowed to combine adjacent rooftops. In those cases, the continuous roof area is

considered same as their total combined building footprint. This relaxation of the single-

parcel area constraint result allows for interesting opportunities for rooftop agriculture

through multi-building partnerships.

94

Fig. 5-2: a) Delineated buildings, b) The division of the Boston area for LiDAR data

The delineated building layer from the previous step was then subjected to slope analysis

(Fig. 5-2a). For this purpose, a digital surface model (DSM) of Boston buildings was

created using remotely sensed LiDAR point cloud dataset and the identified building

layers. The LiDAR point cloud database was obtained from MassGIS in LAZ (compressed

LAS) file format. The extracted LAS files are originated from airborne LiDAR cloud data

sources that includes elevation and intensity of the topographical features taken from the

first and last pulse returns from a LiDAR instrument flown on a rotary wing platform [164].

These LAS files can be displayed and analyzed in ArcMap as point clouds. For this study,

373 separate LiDAR LAS files are extracted for Boston which consist of total 698,553,071

measured known points (Fig. 5-2b). The resolution of each LAS file is 10 ft × 10 ft. These

dataset were later combined into a single LiDAR database (.lasd) file in ArcMap using

Create LAS dataset tool. The resultant layer was than post-processed using LAS

Reclassification tools to identify the building features for slope analysis. Table 5-2

summarizes the spatial extent of LiDAR cloud dataset.

a) b)

95

Table 5-2. Description of parameters used to estimate rooftop area

Spatial

parameters

Scale Projection Description

Output layer 1: 1,00,000 World Equal Area

projection

Feature layer

Land parcels 1: 1,00,000

North American Datum

1983, State plane

Massachusetts, ft

Feature layer

Buildings 1: 1,00,000 Feature layer

Extruded 2D

buildings

1: 1,00,000 Feature layer

LiDAR Point

Clouds

1: 1,00,000 LAZ layer

The corrected LAS data-file was then converted to raster database using LAS to raster tool.

Elevation was used as a conversion value field and Nearest neighbor raster interpolation

method. Previous research has found the Elevation parameter to be useful in extracting the

height variation within a building with a flat roof and Nearest neighbor interpolation

method suitable over inverse distance weighted methods [158]. The resultant DSM layer

of Boston was then further processed for slope and aspect analysis (Fig. 5-3).

Slope analysis of was performed using the Surface slope tool of ArcMap spatial analyst.

The DSM layer was used as an input and degree (pitch) was chosen as output slope layer

unit. Previous work has found that, for a typical urban region, DSM slope ranges from 0-

25°, 25-50° and 50-90° correlated well with actual rooftop types [164]. Slopes of less than

5° usually indicated a flat roof, 25-50°, a pitched roof, and slopes that were greater than

50° were usually associated with a breakline. Therefore, we reclassified all the building

polygons with average slopes of less than 5o as flat roofs using the Reclassify tool. The

reclassified slope layer was masked using the Extract by mask tool where the slope layer

96

was used as an input and building footprint layer as masked layer to extract the roofs with

slope ranges of 0-5°. Wherever there was a sudden change in elevation―between the

ground and the roof of a building or between two levels of the same building―the raster

pixel has a slope greater than 50o.

Fig. 5-3: a) Boston downtown geographic area of interest; b) High resolution LiDAR

data; c) Extracted building footprint layer

5.2.3 Mapping ground level parcels

Mapping of suitable ground level areas for urban farming was also performed in several

steps (Fig. 5-1). Another GIS-based site suitability model was developed using the Model

Builder automation tool and adapted from the spatial analysis framework constructed in

our previous study on bioenergy crop potential in the city [35]. Boston’s land parcel layer

was used as an input into GIS model and a series of linear combinations of spatially

referenced layers were used as screening layers, with some boundary conditions to identify

land parcels that can be suitably used. Here, ground level urban farmland was defined as

parcels that are in areas zoned as suitable for primary agriculture, with a minimum parcel

size of 100 ft2, a soil slope <15%, and adequate soil quality and sunlight hours (6 hours in

growing season). Land use types that fit these criteria can be diverse, especially within built

up areas. For Boston, these included public and private vacant lands, as well as private

residential and commercial underutilized areas.

a) b) c)

97

First, an intermediate layer for suitable parcels were identified using spatial screening of

unsuitable areas from the input parcel layer (Table 3), which features of all Boston parcels

estimated for 2014. Screening layers were spatially joined using the Join tool and overlaid

with the input layer using the Erase tool to exclude areas unsuitable for farming because

of zoning, ownership and accessibility restrictions. These screening layers included area

occupied by buildings, driveways, water features, and conservation lands. These layers

were obtained from the City of Boston Department of Innovation and Technology [163],

MassGIS [162], and the Tufts University Boston GIS Database [165]. World equal area

projection system was chosen as the default projected coordinate system for ArcMap data

frame properties to conduct the necessary geoprocessing calculations (Table 5-3). The

scale of all the layers was 1:1,000,000. Streets and highways were pre-excluded from the

input parcel layer and so did not require subsequent exclusion.

Next, the intermediate parcel layer was subjected to bio-geophysical analysis. The bio-

geophysical model performed minimum area screening, marginal slope, soil quality, and

shadow analysis. Minimum parcel size was chosen accordingly to comply with city zoning

restriction and a marginal slope to minimize runoff and erosion and for ease of cultivation

and harvesting, following previous work [152].

98

Table 5-3. Description of parameters used to estimate ground level area

Spatial parameters Scale Projection

Output layer 1: 1,000,000 World equal area projection system

Land parcels 1: 1,000,000

North American Datum (NAD) 1983,

State plane Massachusetts, ft

Buildings 1: 1,000,000

Drive ways 1: 1,000,000

Water and protected

areas

1: 1,000,000

Soil types 1: 1,000,000

World Geodetic System (WGS) 1984

Soil slopes 1: 1,000,000

Extruded 2D buildings 1: 1,000,000

Soil quality information for Boston was obtained from the United States Department of

Agricultural Natural Resources Conversation Services (USDA-NRCS) soil survey

geographic database (SSURGO) [166]. The SSURGO database contains information about

soil types based on a soil classification for the entire United States. There are four distinct

classes of soil types (A, B, C and D) and three major farmland classifications available that

include prime farmland, farmland of statewide importance and not prime farmland. Areas

with soils listed as ‘prime farmland’ and ‘farmland of statewide importance’ were

classified as suitable for urban agriculture (Table 5-4). The geoprocessing steps included

selection and simultaneous Erase of unsuitable parcels that are classified as ‘not prime

farmland’, soil type C and D and slope >15%.

Table 5-4. USDA-NRCS soils classification

USDA NRCS classification Soil type Slope (%) Soil type

Prime farmland Silt loam/Shaly silt/loam/clay

loam

0-8% A

Farmland of statewide

importance

Silt loam/Shaly silt/loam/clay

loam

2-15% B

Not prime farmland Urban/industrial dump/gravel

pits

0-35% C/D

99

Finally, sun-shadow analysis was conducted to estimate the duration and intensity of

sunlight over the previously identified suitable parcels, based on heights and footprints of

existing buildings. Trees were excluded from the sun-shadow analysis as these could be

removed to prioritize urban agriculture. 3D sun-shadow tool was used to identify the

suitable parcels that receive a minimum of six hours of sunlight during the middle of the

growing season (Fig. 5-4a). Generally, land parcels with full sun exposure on the south,

southeast and southwest with no buildings or other obstruction on those sides were

considered optimal, while parcels with shade during any part of the day were considered a

shadow area and removed (Fig. 5-4b). 3D sun-shadow analysis required extruded 2D

buildings as an input features, created from attribute data from the Boston building

footprint data layer. The analysis was conducted on extruded buildings for July 21st

between 10AM – 4PM. July 21st was considered as the representative day for analysis that

lies at the midpoint of the seven-month growing season (April – October) in Boston. The

analysis yielded composite 2D projections created by shadow volumes on each parcel for

the specified time duration. The projection multipatch features were converted to polygon

features using Multipatch to polygon tool to create the shadow vector map layer, which

was used as an Erase layer in the bio-geophysical model along with unsuitable soil quality

and slope layers.

100

Fig. 5-4: Shadow distribution on Boston Downtown at a) 10 AM and b) 4 PM on July

21st

5.2.4 Estimating food yields

As an urban coastal area, Boston has a subtropical climate and can support production of a

multitude of fresh fruits and vegetables, especially between April – October [160]. In this

study, Boston’s total urban food potential was estimated using spatially averaged crop yield

results. Suitable plants species were identified based on agricultural zone information

derived from United States Department of Agriculture (USDA) plant hardiness zone map

[167]. The plant hardiness zone map is the standard by which gardeners and growers can

determine which plants are most likely to thrive at a particular location. The map is based

on the average annual minimum winter temperature, divided into 10-degree Fahrenheit

zones. According to that map, Boston falls into zone 5b, with average minimum

temperatures in the -15 to -10 °F range. Suitable food-plant yield values for Boston were

obtained for both conventional urban gardening and hydroponic rooftop gardening from

literature. These average constant yield values for current climate conditions are listed in

Table 5-5. Total food yield Mf (t yr-1) was estimated by multiplying each individual rooftop

a) b)

101

Ar (ha) and land parcel area Ag (ha) with associated yield data Yi (kg m-2 yr-1) (eqs. i and ii)

and summing.

Table 5-5. Plant yield information

Production practice Food types Yield (kg m-2 yr-1) Source

Conventional urban

gardening, Yg

Dark green

vegetables; 1.35 Duchemin et al.

[168] Tree fruits

Hydroponic rooftop

gardening, Yr

Dark green

vegetables 19.5

Grewal and

Grewal [153]

𝑀𝑟 = ∑ 𝐴𝑟 ∙ 𝑌𝑟𝑖 (i)

𝑀𝑔 = ∑ 𝐴𝑔 ∙ 𝑌𝑔𝑖 (ii)

5.2.5 Validation

Validation of results was conducted to ensure accuracy of the spatial models and ground-

truthing of identified rooftop and ground level parcels. To accomplish this, four suitable

rooftops and four ground level parcels were randomly selected from the identified rooftop

and land parcel layers using the Create random point tools of ArcMap. The python-based

geo-processing tool generated 20 random points from the original input parcel layer; 8 out

of 20 points coincide with centroids of the parcels identified. The resultant parcels were

further layered with high resolution (15 cm) orthoimagery for visual comparison and

validation and compared against existing Tax Assessor’s data obtained from the Boston

Redevelopment Authority. The Find similar tool was also used to determine the extent of

similarities and false negative parcels not identified in spatial analysis. Validation could

be further bolstered through on-site inspection, but this was not done here so as to

emphasize automated methods.

102

5.3 Results

5.3.1 Rooftop area mapping

The roof area estimation steps are displayed in panel format in Fig. 5-5. This shows the a)

input buildings, b) reclassified LiDAR points with ‘buildings’ object classified as green

colored feature c) digital surface model layer with delineated buildings, d) slope analysis

with color coding represent the slope range (green 0-5°; yellow 5-50°; blue 50-90°), e)

building with slope <5o in green, and f) final flat roof layers in red.

Fig. 5-5: Flat rooftop mapping steps

Fig. 5-6 visualizes the distribution of suitable flat roofs for urban agriculture across the

city. Approximately 922 ha (9.22 km2) of potential roof area was identified, representing

42% of total the roof area and 7.4% of the total land area in Boston, respectively.

Examining rooftop suitability by neighborhood, the communities of Dorchester, Roxbury,

a) Building polygons b) LiDAR point

cloud

f) Extraction of flat rooftops

e) Buildings with slope

<5o

d) Slope analysis of DSM

c) Raster digital surface

model

103

South Boston, Jamaica Plain, and Brighton were found most suitable locations for rooftop

urban farming, where approximately 60% of identified rooftops are located (Fig. 5-7). The

North End, Financial District, Charlestown, and East Boston communities have a lower

proportion of suitable roofs due to lack of sunlight availability and height and zoning

restrictions.

Fig. 5-6: Flat Roofs in Boston

Fig. 5-7: a) Flat roof distribution in Boston’s neighborhoods, by number; b) overlaid

aerial image of flat roofs in the Dorchester neighborhood (suitable roofs filled in grey

with red outline)

a) b)

104

5.3.2 Ground level farmland mapping

The ground level area estimation steps are displayed in panel format in Fig. 5-8. This shows

the a) input layer of Boston parcels, all the exclusion layers including b) buildings, c) water

and protected areas, d) driveways, and e) parcels that are not suitable due to inferior soil

quality, slope and inadequate sunlight, resulting in f) final available ground level areas.

Fig. 5-9 visualizes the total ground level land availability across the city. Approximately,

1,250 ha (12.5 km2) of potential urban farmland was mapped, representing 10% of

Boston’s total land area. Of this, 22% consists of vacant lands and the remainder is under-

utilized residential and commercial areas. There are approximately 184 ha (1.84 km2) of

public and 100 ha (1 km2) of private vacant lands available for immediate ground level

food production. Fully 78% of the potential land areas identified are under-utilized parcels,

nearly evenly split between residential areas (458 ha) such as yards (excluding paved

driveways), and government and commercial areas (508 ha) that could potentially be

categorized as urban farmland and used for agricultural purposes.

Suitable ground level farmland resources are concentrated in eastern and southern part of

the city. Due to sufficient sunshine and low building and population density relative to the

rest of the city, the southern neighborhoods of Dorchester, Roxbury, Mattapan, and West

Roxbury are promising areas for growing fruits and vegetables. These areas have also been

identified by the City of Boston as potential location for developing urban agriculture [94].

On the other hand, the highly urbanized and densely populated core of Boston was not

found suitable due to the high proportion of impervious areas, limited parcel sizes, and

large shadow areas associated with buildings; however, this part of the city has some

105

potential for rooftop urban agriculture. The results of this study suggest that Boston has

significant amount of productive farmland and roof area available for neighborhood and

commercial scale food production.

a) Land parcels (input layer)

b) Buildings

c) Driveways and parking

lots

d) Water and protected

areas

e) Unsuitable parcels from

slope and shadow analysis

f) Ground level areas

(output layer)

Fig. 5-8: Ground level farmland estimation in Boston

106

Fig. 5-9: Available ground level areas in Boston

5.3.4 Food yield potential

Evaluation of maximum food yield potential in Boston are shown in Table 5-6. It was

estimated that Boston’s ground level parcel and rooftop area have a capacity to produce

annually nearly 17,000 metric tons and 180,000 tons of high-yield fresh fruits and

vegetables respectively, if 100% of available areas is used in a regular growing season.

Potential food yields for hydroponic rooftop-based gardening are higher compared to

conventional ground-based urban gardening (Table 5-6). Actual yields may vary based on

soil quality as well as future climate conditions. For context, this amount of food can

potentially provide enough vegetables and fruits for approximately 280,000 people (30%

of Boston’s population) with an annual average intake of 698 kg fresh fruit and vegetables

[106]. The study shows promising results for increasing local production in Boston’s food

system.

107

Table 5-6: Potential food production in Boston

Production practice Yield

(t ha-1yr-1)

Area

(ha)

Food production

(t yr-1)

Conventional urban gardening, Yg 13.5 1250 17,000

Hydroponic rooftop gardening, Yr 195 922 180,000

5.3.5 Validation

Each of the 8 randomly-selected potential ground level and rooftop parcels passed both

visual inspection and validation against data in Boston’s Tax Assessor’s database. Aerial

images, with building id, parcel id and addresses of selected buildings and parcels are

presented in panel format in Fig. 5-10. One of the selected flat-roofed buildings (45 Nevins

St) was a parking garage with rooftop spaces that could be used for greenhouse and/or

hydroponic agriculture but may also be appropriate to screen out. The use of even higher

resolution LiDAR data with precise classification could further improve the assessment

and validation, since features such as rooftop access could be characterized. One of the

selected ground level parcels (ID: 1401121000) was adjacent to several other small parcels,

which could be farmed independently or through community agriculture on shared private

lands.

108

Fig. 5-10: Randomly selected rooftop (top row) and ground level (bottom row)

parcels

5.4 Discussion and Implications

The study shows a step-by-step geospatial modeling approach of assessing local food

potential in Boston and how it can supplement the City’s existing food system. In

estimating total potential yield, it was assumed that 100% of available area will be used

towards developing urban agriculture; so these results should be seen as an upper bound,

as not all farm land available in the city may be practically be used. Future changes in land

use patterns and local climate conditions may affect the suitability of the different food

crops and their associated yields. Community acceptance, soil contamination, safety, and

traffic considerations could reduce the amount of available land that is practically usable.

Several previous analyses have incorporated estimates of the proportion of potentially

productive urban land that could practically be utilized, either through application of a city-

875 Morton St, Boston 115 Norfolk St, Boston 761 Gallivan Bl, Boston 45 Nevins St, Boston

Parcel ID: 1601229000 Parcel ID: 1401121000 Parcel ID: 1700979000 Parcel ID: 800943000

109

wide average (e.g., 75% for parcels in Oakland by McClintock et al. [156]) or through

creation of different production scenarios (e.g., Grewal and Grewal for Cleveland [153]).

Previous urban agricultural land assessments have taken a diversity of approaches (Table

5-1), with different land types included and different screening criteria. Some have

included only public lands, investigated existing sites, or screened with the intention of

identifying a small number of promising sites. Of those that conducted a city-wide

inventories of potential agricultural land, namely Cleveland [153], Oakland [156], Detroit

[148], and Philadelphia [152], researchers found a range of 2.6-7.8% of total municipal

land areas, with a mix of public and private vacant lots depending on the study. For

Cleveland, the inclusion of occupied residential parcels (i.e., yards) brought the ground

level total to 37% of the city’s total land area, while the flat industrial rooftops were

assessed at 5.8%. The results presented here for Boston for vacant lots correspond to 2.3%,

below the range reported in previous work. This could be due in part to the relative

economic health of Boston compared to the other cities, leading to active development or

conversion of vacant parcels. Boston’s result for total ground level availability of 10% is

well below that for Cleveland, likely due to the multiple screening criteria of slope, soil

quality, impervious surfaces, and shadow included here that were not considered for

Cleveland. The result for rooftops of 7.4% moderately exceeds that for Cleveland but

includes residential and commercial properties, whereas the latter only considers industrial

roofs [153].

110

This recent research indicates that combining high resolution remotely sensed data and GIS

methods in assessment of precision agriculture and urban food potential mapping can

support a wide range of policy and planning efforts. In discussing early efforts in Portland

and Vancouver, Mendes et al. suggest that the very process of carrying out an inventory

can promote stakeholder engagement and assist in consensus community planning [137],

particularly when it is linked to official goals for agricultural land use. Implementation of

this local scale production requires detail food system modeling. The total number of food

based community gardens are growing but represents a very small share of the city’s food

supply, the great majority of which is produced outside of the region [160]. While only a

small portion of the city’s food is currently sourced locally, a network of over 200

community gardens and several dozen local food vendors grow and distribute local fresh

fruits and vegetables in the city. Research on the urban gardening movement in the city

suggests that personal food production in community gardens is helping to fill the gap

particularly in low-income, inner city neighborhoods [94]. Integrating on-the-ground

surveys and interviews with spatial analysis may also help refine modeling efforts and lead

to targeted recommendations for the development of local food systems.

However, several potential challenges need to be carefully considered in expanding the

current scale of city’s local food production activity. Surface runoff and water quality

degradation due to possible additional use of pesticides and fertilizers in the city may

impact local water bodies, especially the Charles River watershed which is already

subjected to cyanobacteria blooms and has a total maximum daily load (TMDL) framework

in place [169]. Neighborhood noise pollution of farming is another key issue that many

111

communities have faced in many other urban regions [133]. Access to adequate fresh water

for some urban area is an important factor that may reduce the rate of community garden

expansion. Also, rooftop farming imposes additional live loads on roofs that may increase

the rate of structural deterioration and maintenance cost.

While Boston was used as a test case in this study, the methods were designed to be applied

to any city or metropolitan area. Urban agriculture inventories may be of particular interest

for cities that are expecting climatic shifts towards more temperate conditions, and/or are

undergoing depopulation and so have a high proportion of underutilized lands and lower

economic rents than Boston.

5.5 Conclusions

Urban agriculture has numerous and varied potential benefits, including community and

economic development, improvements in environmental quality, and partial fulfilment of

food demand, potentially enhancing food security. Here, an automated modeling

framework utilizing GIS tools and remote sensing data was described and used to estimate

potential agricultural land in within Boston, encompassing both ground level parcels and

flat rooftops.

The total available rooftop area was estimated at 922 ha (9.2 km2) that can be suitably used

for food production, representing 42% of city’s total roof area. At ground level, the

estimated area is 1,250 ha (12.5 km2) representing 10% of city’s total available land area,

112

of which 78% is private residential and commercial underutilized areas and 22% is public

and private vacant lands.

If all of the available ground level parcels were converted to agriculture and intensive

hydroponic systems were installed on all suitable roofs, Boston has the capacity to produce

nearly 17,000 metric tons/yr and 180,000 tons/yr of high yield food plants, respectively, in

a regular growing season. This amount is sufficient to supply roughly 30% of Boston’s

fruit and vegetable demand annually. Even partial development of urban agriculture on

these lands could fulfill a significant proportion of the city’s food demand, corroborating

similar results for other cities in North America.

113

Chapter 6 Conclusion and Future Works

The dissertation focused on conducting a series of case studies related to the application of

geographic information system in quantifying the extent of urban system’s response to

climate change induced the atmospheric effect on infrastructure and resource self-

sufficiency. Following the research objectives mentioned in the introduction section, four

specific studies were carried out on different aspects of urban stock assessment and

geospatial modeling.

In chapter 2, describes a study on 4D-GIS based assessment of mapping vulnerable

building stocks in Boston under different projected climate change scenarios. Novel

corrosion model was constructed to assess building and block-level vulnerability of urban

concrete buildings, related maintenance needs, and to project cover thickness degradation

for the existing building stock. This is the first ever urban scale study that looked at both

spatial and temporal aspects of atmospheric CO2 and Chloride induced corrosion

phenomena. The results suggest that climate change may reduce the average design service

life of concrete buildings, potentially requiring extensive repairs well within the average

design service life. Chapter 3 describes another geospatial assessment for quantifying city’s

extent of energy self-reliance from leveraging local resources. Here, a comprehensive,

novel and scalable GIS-based modeling framework was created to assess potential urban

marginal lands that considered ownership, zoning, under-utilized public and private areas

with adequate soil quality and sunlight. This is the most sophisticated screening tool to date

to examine parcel suitability for bioenergy applications, specifically through the

implementation of 3D sun-shade analysis. The detailed analysis found nearly a quarter of

114

city’s land parcel suitable for energy crop agriculture. Chapter 4 describes an application

and validation of GIS framework developed in chapter 3 on a regional scale. Here a site-

specific marginal land assessment for bioenergy production was conducted for the MAPC

cities. The outcomes of this study confirms the model accuracy and provides

comprehensive information about marginal land availability for using in the demand

center. Finally, Chapter 5 describes an automated geospatial analysis of urban land

availability for food-based agriculture purpose. The study considers both ground level and

rooftop areas and estimate the total potential for local food production in the city. High-

resolution LiDAR imagery was used for accurate estimation of suitable rooftops. Overall,

the findings could potentially aid in decision making and prioritize policies regarding urban

system engineering.

The works depicted in this dissertation have made novel contributions to the urban

metabolism and GIS body of knowledge. There is also a novelty in creating spatial

framework and models for assessing urban scale vulnerable infrastructure and sustainable

resource that may be beneficial for improving urban ecosystem health. Especially, a

building-by-building corrosion model was developed for an entire city and the most

sophisticated sub-parcel level screening tool to date to examine parcel suitability for

bioenergy and urban food applications, specifically through the implementation of sun-

shade analysis in custom model builder routine. The models are scalable and robust enough

to explore other urban regions with similar scenarios. With the increase in disaster

dynamics, it’s important to have enough tools to sustain as a resourceful and low impact

city. Therefore, the question of self-sufficiency and resiliency are important. This work

115

fills an important gap of how to leverage existing knowledge of GIS and urban stock

assessment in answering these questions.

There is a wide range future studies that could grow out of this dissertation work. To

realistically include the concept of future climate change impacts and self-sufficiency into

urban engineering, more and more city-specific spatial and temporal study of infrastructure

vulnerability and resource productivity assessment are required. Comparative studies that

apply a unified methodology may help to identify adaptation and other planning measures

(such as changes in building codes or land use zoning) that are appropriate in different

cities.

Several modeling extensions could be made. These studies used a deterministic modeling

framework with average values or fixed ranges. A stochastic analysis using assigned or

fitted parameter value probability distributions could assess uncertainty in the results using

Monte Carlo simulation, while model parametrization and sensitivity analysis could be

performed to determine the most sensitive parameters and test model robustness. For the

concrete buildings work, advanced degradation models are being developed with more

detailed chemistry, specifically that couple the carbonation and chlorination processes,

which could be used in the framework presented in this dissertation. In addition, model-

predicted carbonation and chlorination depths can and should be validated with

experimental site data. For the geospatial modeling of historical concrete building stocks,

instead of constant concrete properties assumed in this work, the temporal variation of code

recommended water-cement ratio, concrete strength, and corrosion protected cover

116

thickness could be considered, or, better, yet, historical construction and material testing

documents could be assembled to determine these modeling parameters on a building-by-

building basis. In addition to revising code modification, as suggested in Chapter 2, the

projected results for corrosion initiation in existing buildings outcome can be used for

devising appropriate building inspection programs.

Possible extensions of urban resource models may include scenario-based land use

combinations for assessing biomass and bioenergy yield. As not all of the identified areas

can be practically used, for reasons described in Chapters 3-5, it could be fruitful to engage

local stakeholders to create probable usage scenarios for the different percentages of

available residential and commercial lands use types, and then estimate potential bioenergy

or food production potential. For example, for total available marginal land assessment for

bioenergy crops, possible scenarios might include only vacant lands, or vacant lands and

50% of commercial under-utilized areas, depending on municipal development and zoning

priorities. For estimating bioenergy and/or food yield, average values reflecting current

climatic conditions were used here, but in a future climate, different crop types and

productivities may be appropriate to use should these studies be extended into the future.

Engineering analysis is just the first step; community scale surveys also need to be

conducted before large-scale urban bioenergy or food agriculture could practically develop.

Successful implementation also requires site validation and planning through soil testing,

logistics planning for harvesting and processing, and financial modeling. A final extension

in scope would be to create a comprehensive model capable of assessing national scale

urban site suitability for urban agriculture implementation. However, this type of modeling

117

will require large scale data acquisition, manipulation, and modeling, using national input

data layers if possible. As more and higher-resolution geospatial data become available, it

will become easier for cities, regions, and entire nations to conduct bottom-up analyses of

buildings, land, and other resources in order to plan for climate change impacts and assess

potential resource productivity.

118

REFERENCES

[1] Kammen DM, Sunter DA. City-integrated renewable energy for urban sustainability.

Science. 2016;352:922-8.

[2] Roggema R. Towards Enhanced Resilience in City Design: A Proposition. Land.

2014;3:460-81.

[3] Wolman A. The metabolism of cities. Scientific American. 1965;213:179-90.

[4] Bristow DN, Kennedy CA. Urban Metabolism and the Energy Stored in Cities.

Journal of Industrial Ecology. 2013;17:656-67.

[5] ICIC. Resilient food systems, resilient cities: Recommendations for the City of

Boston. 2015.

[6] Sharifi A, Yamagata Y. Principles and criteria for assessing urban energy resilience:

A literature review. Renewable and Sustainable Energy Reviews. 2016;60:1654-77.

[7] Tanikawa H, Hashimoto S. Urban stock over time: spatial material stock analysis

using 4d-GIS. Building Research & Information. 2009;37:483-502.

[8] Godschalk DR. Urban hazard mitigation: creating resilient cities. Natural hazards

review. 2003;4:136-43.

[9] Krumdieck S. Transition Engineering of urban transportation for resilience to Peak.

Oil risks. ASME 2011 International Mechanical Engineering Congress and Exposition:

American Society of Mechanical Engineers; 2011. p. 387-400.

119

[10] Fleischhauer M. The role of spatial planning in strengthening urban resilience.

Resilience of Cities to Terrorist and other Threats Springer Science+ Business Media BV.

2008:273-98.

[11] Milman A, Short A. Incorporating resilience into sustainability indicators: An

example for the urban water sector. Global Environmental Change. 2008;18:758-67.

[12] Frazier TG, Thompson CM, Dezzani RJ, Butsick D. Spatial and temporal

quantification of resilience at the community scale. Applied Geography. 2013;42:95-107.

[13] Ouyang M, Dueñas-Osorio L, Min X. A three-stage resilience analysis framework

for urban infrastructure systems. Structural Safety. 2012;36:23-31.

[14] Vale LJ, Campanella TJ. The resilient city: How modern cities recover from

disaster: Oxford University Press; 2005.

[15] Cutter SL, Ash KD, Emrich CT. The geographies of community disaster resilience.

Global Environmental Change. 2014;29:65-77.

[16] Kennedy C, Pincetl S, Bunje P. The study of urban metabolism and its applications

to urban planning and design. Environmental Pollution. 2011;159:1965-73.

[17] Wallsten B, Carlsson A, Frändegård P, Krook J, Svanström S. To prospect an urban

mine – assessing the metal recovery potential of infrastructure “cold spots” in

Norrköping, Sweden. Journal of Cleaner Production. 2013;55:103-11.

[18] Zhu X. GIS and urban mining. Resources. 2014;3:235-47.

[19] Van Beers D, Graedel T. Spatial characterisation of multi-level in-use copper and

zinc stocks in Australia. Journal of Cleaner Production. 2007;15:849-61.

120

[20] Krook J, Carlsson A, Eklund M, Frändegård P, Svensson N. Urban mining:

hibernating copper stocks in local power grids. Journal of Cleaner Production.

2011;19:1052-6.

[21] Gordon RB, Bertram M, Graedel T. Metal stocks and sustainability. Proceedings of

the National Academy of Sciences. 2006;103:1209-14.

[22] Hashimoto S, Tanikawa H, Moriguchi Y. Where will large amounts of materials

accumulated within the economy go?–A material flow analysis of construction minerals

for Japan. Waste Management. 2007;27:1725-38.

[23] Schandl H, West J. Material flows and material productivity in China, Australia, and

Japan. Journal of Industrial Ecology. 2012;16:352-64.

[24] Bonham-Carter GF. Geographic information systems for geoscientists: modelling

with GIS: Elsevier; 2014.

[25] Chang K-t. Introduction to geographic information systems: McGraw-Hill Higher

Education Boston; 2006.

[26] Dueker KJ. Land resource information systems: a review of fifteen years experience.

Geo-Processing (Netherlands). 1979.

[27] Geographic information systems and science. International Journal of Digital Earth.

2011;4:360-1.

[28] Panigrahi N. Computing in Geographic Information Systems: CRC Press; 2014.

[29] Jat MK, Garg PK, Khare D. Monitoring and modelling of urban sprawl using remote

sensing and GIS techniques. International journal of Applied Earth Observation and

Geoinformation. 2008;10:26-43.

121

[30] Weng Q. A remote sensing? GIS evaluation of urban expansion and its impact on

surface temperature in the Zhujiang Delta, China. International Journal of Remote

Sensing. 2001;22:1999-2014.

[31] Sugimoto K, Morita H, Tanikawa H. Time-series Analysis and Prediction of

Building Material Stock and Flow Using 4d-GIS CUPUM. 2015.

[32] Stewart MG, Wang X, Nguyen MN. Climate change adaptation for corrosion control

of concrete infrastructure. Structural Safety. 2012;35:29-39.

[33] Niblick B, Monnell JD, Zhao X, Landis AE. Using geographic information systems

to assess potential biofuel crop production on urban marginal lands. Applied Energy.

2013;103:234-42.

[34] Saha M, Eckelman MJ. Urban scale mapping of concrete degradation from projected

climate change. Urban Climate. 2014;9:101-14.

[35] Saha M, Eckelman MJ. Geospatial assessment of potential bioenergy crop

production on urban marginal land. Applied Energy. 2015;159:540-7.

[36] McMichael AJ, Woodruff RE, Hales S. Climate change and human health: present

and future risks. The Lancet. 2006;367:859-69.

[37] Rosenzweig C, Solecki WD, Hammer SA, Mehrotra S. Climate change and cities:

First assessment report of the urban climate change research network: Cambridge

University Press; 2011.

[38] Cole IS, Paterson DA. Possible effects of climate change on atmospheric corrosion

in Australia. Corrosion Engineering, Science and Technology. 2010;45:19-26.

122

[39] Stocker T, Qin D, Plattner G-K. Climate change 2013 - The physical science basis:

working group I contribution to the fifth assessment report of the IPCC. Cambridge

University Press. 2013.

[40] The Next Four Decades, the older population in the United States: 2010 to 2050,

population estimates and projections, US Census Bureau. Available from:

<https://www.census.gov/prod/2010pubs/p25-1138.pdf>, Accessed on June 2014 2010.

[41] Solecki W, Marcotullio PJ. Climate change and urban biodiversity vulnerability. In:

Elmqvist T, Fragkias M, Goodness J, Güneralp B, Marcotullio PJ, McDonald RI, et al.,

editors. Urbanization, Biodiversity and Ecosystem Services: Challenges and

Opportunities: Springer Netherlands; 2013. p. 485-504.

[42] Wilbanks T. Climate change and infrastructure, urban systems, and vulnerabilities.

Technical Report to the US Department of Energy in Support of the National Climate

Assessment 2012.

[43] McCrea R, Stimson R, Marans RW. The evolution of integrative approaches to the

analysis of quality of urban life. In: Marans RW, Stimson RJ, editors. Investigating

Quality of Urban Life: Springer Netherlands; 2011. p. 77-104.

[44] Nijland TG, Adan OCG, van Hees RPJ, van Etten BD. Evaluation of the effects of

expected climate change on the durability of building materials with suggestions for

adaptation. Heron. 2009;54:37.

[45] Anderson DMaB, D. Global climate change impacts in the United States. 2009.

[46] Zivica Vr, Bajza A. Acidic attack of cement based materials - a review.: Part 1.

Principle of acidic attack. Construction and Building Materials. 2001;15:331-40.

123

[47] Greaver TL, Sullivan TJ, Herrick JD, Barber MC, Baron JS, Cosby BJ, et al.

Ecological effects of nitrogen and sulfur air pollution in the US: what do we know?

Frontiers in Ecology and the Environment. 2012;10:365-72.

[48] Talukdar S, Banthia N, Grace JR. Carbonation in concrete infrastructure in the

context of global climate change – Part 1: Experimental results and model development.

Cement and Concrete Composites. 2012;34:924-30.

[49] Mehta P, Monteriro P. Concrete, microstructure, properties, and materials. Third ed.

New York: McGraw-Hill Companies, Inc.; 2006.

[50] Yoon I-S, Çopuroğlu O, Park K-B. Effect of global climatic change on carbonation

progress of concrete. Atmospheric Environment. 2007;41:7274-85.

[51] Stewart MG, Peng J. Life-cycle cost assessment of climate change adaptation

measures to minimise carbonation-induced corrosion risks | NOVA. The University of

Newcastle's Digital Repository, Life-cycle cost assessment of climate change adaptation

measures to minimise carbonation-induced corrosion risks. 2010.

[52] Stewart MG, Wang X, Nguyen MN. Climate change impact and risks of concrete

infrastructure deterioration. Engineering Structures. 2011;33:1326-37.

[53] Bastidas-Arteaga E, Chateauneuf A, Sánchez-Silva M, Bressolette P, Schoefs F.

Influence of weather and global warming in chloride ingress into concrete: A stochastic

approach. Structural Safety. 2010;32:238-49.

[54] Talukdar S, Banthia N, Grace JR, Cohen S. Carbonation in concrete infrastructure in

the context of global climate change: Part 2 – Canadian urban simulations. Cement and

Concrete Composites. 2012;34:931-5.

124

[55] Ruth M. Modelling Infrastructure Vulnerabilities and Adaptation to Climate Change

in Urban Systems: Methodology and Application to Metropolitan Boston. European

Regional Science Association Meetings, Jyväskylä, Finland2003.

[56] MEMA. Commonwealth of Massachusetts state hazard mitigation plan. 2010.

[57] Allen MR, Stott PA, Mitchell JFB, Schnur R, Delworth TL. Quantifying the

uncertainty in forecasts of anthropogenic climate change. Nature. 2000;407:617-20.

[58] Nakicenovic N, Alcamo J, Davis G, de Vries B, Fenhann J, Gaffin S, et al. Special

report on emissions scenarios: A special report of working group Iii of the

intergovernmental panel on climate change. Pacific Northwest National Laboratory,

Richland, WA (US), Environmental Molecular Sciences Laboratory (US); 2000.

[59] Wigley TML, Richels R, Edmonds JA. Economic and environmental choices in the

stabilization of atmospheric CO2 concentrations. Nature. 1996;379:240-3.

[60] George K, Ziska LH, Bunce JA, Quebedeaux B. Elevated atmospheric CO 2

concentration and temperature across an urban–rural transect. Atmospheric Environment.

2007;41:7654-65.

[61] Stoner AMK, Hayhoe K, Yang X, Wuebbles DJ. An asynchronous regional

regression model for statistical downscaling of daily climate variables. International

Journal of Climatology. 2013;33:2473-94.

[62] Hayhoe K, Wake C, Anderson B, Liang X-Z, Maurer E, Zhu J, et al. Regional

climate change projections for the Northeast USA. Mitigation and Adaptation Strategies

for Global Change. 2008;13:425-36.

125

[63] DuraCrete. Modelling of degradation. DuraCrete - probabilistic performance based

durability design of concrete structures. EU-Brite EURam III Contract BRPR-CT95-0132

Proj. BE95-1327R4-5 174. 1998.

[64] Kada-Benameur H, Wirquin E, Duthoit B. Determination of apparent activation

energy of concrete by isothermal calorimetry. Cement and Concrete Research.

2000;30:301-5.

[65] Code A. Building code requirements for structural concrete and commentary, in:

ACI 318-11. p. 503. 2011.

[66] Val DV, Stewart MG. Life-cycle cost analysis of reinforced concrete structures in

marine environments. Structural Safety. 2003;25:343-62.

[67] Stanish KD, Hooton RD, Thomas MDA. Testing the chloride penetration resistance

of concrete: A literature Review, FHWA Congract DTFH61-97-R-00022 "Prediction of

chloride penetration in concrete". Toronto, Ontario, Canada: University of Toronto; 2006.

[68] Luguang S, Wei S, Gao J. Time dependent chloride diffusion coefficient in concrete.

Journal of Wuhan University of Technology-Mater Sci Ed. 2013;28:314-9.

[69] DuraCrete. Statistical quantification of the variables in the limit state functions.

DuraCrete-probabilistic performance based durability design of concrete structures. EU-

Brite EURam III Contract BRPR-CT95-0132 Proj. BE95-1347R9 130. 2000.

[70] The Emission and Generation Resource Integrated Database (eGRID), US EPA,

Available from: <http://www.epa.gov/cleanenergy/energy-resources/egrid>, Accessed on

06/03/2015.

126

[71] Muth Jr DJ, Bryden KM, Nelson RG. Sustainable agricultural residue removal for

bioenergy: A spatially comprehensive US national assessment. Applied Energy.

2013;102:403-17.

[72] MassGIS Data Layer. Available from: <http://www.mass.gov/anf/research-and-

tech/it-serv-and-support/application-serv/office-of-geographic-information-

massgis/datalayers/layerlist.html>, Accessed on December 11, 2013. 2013.

[73] Grewal PS, Grewal PS. Can cities become self-reliant in energy? A technological

scenario analysis for Cleveland, Ohio. Cities. 2013;31:404-11.

[74] Talukdar S, Banthia N. Carbonation in concrete infrastructure in the context of

global climate change: Development of a service lifespan model. Construction and

Building Materials. 2013;40:775-82.

[75] Banthia N. American concrete institute committee 130 technical agenda meeting,

Reno, Nevada, Available from:

<http://www.concrete.org/Committees/GetAgendaMinutes.aspx?DocID=78783>,

accessed on July 2, 2014.

[76] Zafeiropoulou T. Carbonation resistance and anticorrosive properties of organic

coatings for concrete structures. Journal of Surface Engineered Materials and Advanced

Technology. 2013;03:67-74.

[77] Andersson R, Fridh K, Stripple H, Häglund M. Calculating CO2 uptake for existing

concrete structures during and after service Life. Environmental Science & Technology.

2013;47:11625-33.

[78] Neville A. Properties of concrete. 5th ed: Pearson Education Limited; 2011.

127

[79] Pantaleo AM, Giarola S, Bauen A, Shah N. Integration of biomass into urban energy

systems for heat and power. Part I: An MILP based spatial optimization methodology.

Energy Convers and Manage. 2014;83:347-61.

[80] Manfren M, Caputo P, Costa G. Paradigm shift in urban energy systems through

distributed generation: Methods and models. Applied Energy. 2011;88:1032-48.

[81] Taylor JR, Lovell ST. Mapping public and private spaces of urban agriculture in

Chicago through the analysis of high-resolution aerial images in Google Earth.

Landscape and Urban Planning. 2012;108:57-70.

[82] Zhuang D, Jiang D, Liu L, Huang Y. Assessment of bioenergy potential on marginal

land in China. Renewable and Sustainable Energy Review. 2011;15:1050-6.

[83] Milbrandt A. A Geographic perspective on the current biomass resource availability

in the United States. Available from: <http://www.nrel.gov/docs/fy06osti/39181.pdf>.

National Renewable Energy Laboratory; 2005.

[84] Schmidt J, Haifly A. Delivering on renewable energy around the world; Natural

resources defense council. Available from: <http://www.nrdc.org/energy/files/delivering-

renewable-energy.pdf>, Washington, DC. 2012.

[85] Shortall OK. “Marginal land” for energy crops: Exploring definitions and embedded

assumptions. Energy Policy. 2013;62:19-27.

[86] Lewis SM, Kelly M. Mapping the potential for biofuel production on marginal

lands: Differences in definitions, data and models across scales. isprs International

Journal of Geo-Information. 2014;3:430-59.

128

[87] Madlener R, Bachhiesl M. Socio-economic drivers of large urban biomass

cogeneration: Sustainable energy supply for Austria's capital Vienna. Energy Policy.

2007;35:1075-87.

[88] Bowman AOM, Pagano MA. Vacant land in cities: An urban resource. Available

from: <http://content.knowledgeplex.org/kp2/img/cache/kp/1114.pdf>. The Brookings

Institution. 2001.

[89] Wilson TO, McNeal FM, Spatari S, G. Abler D, Adler PR. Densified biomass can

cost-effectively mitigate greenhouse gas emissions and address energy security in thermal

applications. Environmental Science & Technology. 2012;46:1270-7.

[90] Angelis-Dimakis A, Biberacher M, Dominguez J, Fiorese G, Gadocha S,

Gnansounou E, et al. Methods and tools to evaluate the availability of renewable energy

sources. Renewable & Sustainable Energy Review. 2011;15:1182-200.

[91] Gopalakrishnan G, Cristina Negri M, Snyder SW. A novel framework to classify

marginal land for sustainable biomass feedstock production. Journal of Environmental

Quality. 2011;40:1593-600.

[92] Gelfand I, Sahajpal R, Zhang X, Izaurralde RC, Gross KL, Robertson GP.

Sustainable bioenergy production from marginal lands in the US Midwest. Nature.

2013;493:514-7.

[93] Milbrandt AR, Heimiller DM, Perry AD, Field CB. Renewable energy potential on

marginal lands in the United States. Renewable & Sustainable Energy Review.

2014;29:473-81.

[94] Chin D, Infahsaeng T, Jakus I, Oorthuys V. Urban farming in Boston: A survey of

opportunities. Available from:

129

<http://ase.tufts.edu/uep/degrees/field_project_reports/2013/Team_8_Final_Report_2013

.pdf>. Tufts University; 2013.

[95] Lovett AA, Sünnenberg GM, Richter GM, Dailey AG, Riche AB, Karp A. Land use

implications of increased biomass production identified by GIS-based suitability and

yield mapping for miscanthus in England. Bioenergy Research. 2009;2:17-28.

[96] Van Meerbeek K, Ottoy S, De Meyer A, Van Schaeybroeck T, Van Orshoven J,

Muys B, et al. The bioenergy potential of conservation areas and roadsides for biogas in

an urbanized region. Applied Energy. 2015;154:742-51.

[97] Arodudu O, Ibrahim E, Voinov A, van Duren I. Exploring bioenergy potentials of

built-up areas based on NEG-EROEI indicators. Ecological Indicators. 2014;47:67-79.

[98] Comber A, Dickie J, Jarvis C, Phillips M, Tansey K. Locating bioenergy facilities

using a modified GIS-based location–allocation-algorithm: Considering the spatial

distribution of resource supply. Applied Energy. 2015;154:309-16.

[99] Zhao X, Monnell JD, Niblick B, Rovensky CD, Landis AE. The viability of biofuel

production on urban marginal land: An analysis of metal contaminants and energy

balance for Pittsburgh's Sunflower Gardens. Landscape Urban Plan. 2014;124:22-33.

[100] Sharifi A, Murayama A. Neighborhood sustainability assessment in action: Cross-

evaluation of three assessment systems and their cases from the US, the UK, and Japan.

Building and Environment. 2014;72:243-58.

[101] Colding J, Barthel S. The potential of ‘Urban Green Commons’ in the resilience

building of cities. Ecological Economics. 2013;86:156-66.

[102] Tang Y, Xie J-S, Geng S. Marginal land-based biomass energy production in

China. International Journal of Plant Biology. 2010;52:112-21.

130

[103] Wild LG, Flisher AJ, Robertson BA. Risk and resilience in orphaned adolescents

living in a community affected by AIDS. Youth & Society. 2011:0044118X11409256.

[104] Pantaleo AM, Giarola S, Bauen A, Shah N. Integration of biomass into urban

energy systems for heat and power. Part II: Sensitivity assessment of main techno-

economic factors. Energy Conversion and Management. 2014;83:362-76.

[105] Lee J, Zlatanova S. Solar radiation over the urban texture: LIDAR data and image

processing techniques for environmental analysis at city scale. In: Lee J, Zlatanova S,

editors. 3D Geo-Information Sciences: Springer Berlin Heidelberg; 2009. p. 319-40.

[106] Ackerman K. The potential for urban agriculture in New York city, growing

capacity, food security, & green Infrastructure. Available from:

<http://urbandesignlab.columbia.edu/>. Columbia University; 2012.

[107] Timmons D, Geoff A, Damery D. Biomass energy crops: Massachusetts potential,

Massachusetts Division of Energy Resources. Available from:

<http://www.mass.gov/eea/docs/doer/renewables/biomass/bio-ma-potential-crop.pdf>.

2008.

[108] Springer TL. Biomass yield from an urban landscape. Biomass and Bioenergy.

2012;37:82-7.

[109] Miguez FE, Maughan M, Bollero GA, Long SP. Modeling spatial and dynamic

variation in growth, yield, and yield stability of the bioenergy crops

Miscanthus × giganteus and Panicum virgatum across the conterminous United States.

GCB Bioenergy. 2012;4:509-20.

131

[110] Wang D, LeBauer D, Dietze M. Predicting yields of short-rotation hybrid poplar

(Populus spp.) for the United States through model-data synthesis. Applied Ecology.

2013;23:944-58.

[111] Jenkins BM, Baxter LL, Miles Jr TR, Miles TR. Combustion properties of biomass.

Fuel Process Technology. 1998;54:17-46.

[112] Biomass based combined heat and power generation, Available from:

<http://www.videncenter.dk/exportcat/combined_heat_and_power.pdf>, Accessed on

08/07/2015.

[113] Lund H, Möller B, Mathiesen BV, Dyrelund A. The role of district heating in

future renewable energy systems. Energy. 2010;35:1381-90.

[114] MacFarlane DW. Potential availability of urban wood biomass in Michigan:

Implications for energy production, carbon sequestration and sustainable forest

management in the U.S.A. Biomass and Bioenergy. 2009;33:628-34.

[115] McKendry P. Energy production from biomass (part 2): conversion technologies.

Bioresource Technology. 2002;83:47-54.

[116] Lu X, Withers MR, Seifkar N, Field RP, Barrett SR, Herzog HJ. Biomass logistics

analysis for large scale biofuel production: case study of loblolly pine and switchgrass.

Bioresource Technology. 2015;183:1-9.

[117] Mathiesen BV, Lund H, Connolly D. Limiting biomass consumption for heating in

100% renewable energy systems. Energy. 2012;48:160-8.

[118] Niblick B, Landis AE. Assessing renewable energy potential on United States

marginal and contaminated sites. Renewable and Sustainable Energy Reviews.

2016;60:489-97.

132

[119] Springer TL. Biomass yield from an urban landscape. Biomass and Bioenergy.

2012;37:82-7.

[120] Kraxner F, Aoki K, Kindermann G, Leduc S, Albrecht F, Liu J, et al. Bioenergy

and the city–What can urban forests contribute? Applied Energy. 2016;165:990-1003.

[121] Kang S, Post W, Wang D, Nichols J, Bandaru V, West T. Hierarchical marginal

land assessment for land use planning. Land Use Policy. 2013;30:106-13.

[122] Xue S, Lewandowski I, Wang XY, Yi ZL. Assessment of the production potentials

of Miscanthus on marginal land in China. Renewable & Sustainable Energy Reviews.

2016;54:932-43.

[123] Field CB, Campbell JE, Lobell DB. Biomass energy: the scale of the potential

resource. Trends in Ecology & Evolution.23:65-72.

[124] Bryngelsson DK, Lindgren K. Why large-scale bioenergy production on marginal

land is unfeasible: A conceptual partial equilibrium analysis. Energy Policy.

2013;55:454-66.

[125] Gibbs HK, Salmon JM. Mapping the world's degraded lands. Applied Geography.

2015;57:12-21.

[126] MassGIS. Office of Geographic Information (MassGIS), Available from:

<http://www.mass.gov/anf/research-and-tech/it-serv-and-support/application-serv/office-

of-geographic-information-massgis/> Accessed on 06/01/2015. 2015.

[127] Knudsen MT, Halberg N, Olesen JE, Byrne J, Iyer V, Toly N. Global trends in

agriculture and food systems. Global development of organic agriculture: challenges and

promises Cabi Publishing, Wallingford. 2006:1-48.

133

[128] Nellemann C. The environmental food crisis: the environment's role in averting

future food crises: a UNEP rapid response assessment: UNEP/Earthprint; 2009.

[129] Halweil B. Home grown: The case for local food in a global market: Worldwatch

Institute; 2002.

[130] Hodgson K, Campbell MC, Bailkey M. Urban agriculture: growing healthy,

sustainable places: American Planning Association; 2011.

[131] Lovell ST. Multifunctional urban agriculture for sustainable land use planning in

the United States. Sustainability. 2010;2:2499-522.

[132] Kulak M, Graves A, Chatterton J. Reducing greenhouse gas emissions with urban

agriculture: a life cycle assessment perspective. Landscape and Urban Planning.

2013;111:68-78.

[133] Hendrickson MK, Porth M. Urban Agriculture—Best Practices and Possibilities.

University of Missouri. 2012:1-52.

[134] Clancy K, Ruhf K. Is local enough? Some arguments for regional food systems.

Choices. 2010;25:123-35.

[135] Kloppenburg Jr J, Hendrickson J, Stevenson GW. Coming in to the foodshed.

Agriculture and Human Values. 1996;13:33-42.

[136] World Bank. Urban agriculture: Findings from four city case studies. Urban

development series. Washington, DC2013.

[137] Mendes W, Balmer K, Kaethler T, Rhoads A. Using land inventories to plan for

urban agriculture: experiences from Portland and Vancouver. Journal of the American

Planning Association. 2008;74:435-49.

134

[138] Neyfakh L. If urban farming took off, what would Boston look like? The Boston

Globe. Boston, MA2014.

[139] City of Boston Office of Food Initiatives.

[140] Boston Redevelopment Authority. Article 89. 2013. Available from:

<http://www.bostonredevelopmentauthority.org/getattachment/8405c72c-7520-43ad-

a969-0e27dddae7a2>, Accessed on 06/03/2015.

[141] Northbound Ventures L. Urban Ag Visioning: A City’s Vision for Growing Food,

Community and the Economy. Boston, MA: City of Boston; 2015.

[142] Peters CJ, Bills NL, Wilkins JL, Fick GW. Foodshed analysis and its relevance to

sustainability. Renewable Agriculture and Food Systems. 2009;24:1-7.

[143] Horst M, Gaolach B. The potential of local food systems in North America: A

review of foodshed analyses. Renewable Agriculture and Food Systems. 2015;30:399-

407.

[144] Thompson Jr E, Harper AM, Kraus S. Think globally—Eat locally: San Francisco

foodshed assessment. Sacramento, CA: American Farmland Trust; 2008.

[145] Ackerman K, Conard M, Culligan P, Plunz R, Sutto M-P, Whittinghill L.

Sustainable food systems for future cities: The potential of urban agriculture. The

Economic and Social Review. 2014;45:189-206.

[146] Balmer K, Gill J, Kaplinger H, Miller J, Peterson M, Rhoads A, et al. The Diggable

City: Making urban agriculture a planning priority. Portland, OR: Department of Urban

and Regional Planning, Portland State University; 2005.

[147] Berger D. A GIS suitability analysis of the potential for rooftop agriculture in New

York City. Master's Thesis: Columbia University; 2013.

135

[148] Colasanti K, Hamm M. Assessing the local food supply capacity of Detroit,

Michigan. Journal of Agriculture, Food Systems, and Community Development. 2010;1.

[149] Eanes F, Ventura SJ. Inventorying Land Availability and Suitability for

Community Gardens in Madison, Wisconsin. Cities and the Environment. 2015;8:2.

[150] Horst M. Growing Green: An Inventory of Public Lands Suitable for Community

Gardening in Seattle, Washington. University of Washington, College of Architecture

and Urban Planning; 2008.

[151] Kaethler TM. Growing space: The potential for urban agriculture in the city of

Vancouver: School of Community and Regional Planning University of British

Columbia; 2006.

[152] Kremer P, DeLiberty TL. Local food practices and growing potential: Mapping the

case of Philadelphia. Applied Geography. 2011;31:1252-61.

[153] Grewal SS, Grewal PS. Can cities become self-reliant in food? Cities. 2012;29:1-

11.

[154] Haberman D, Gillies L, Canter A, Rinner V, Pancrazi L, Martellozzo F. The

Potential of Urban Agriculture in Montréal: A Quantitative Assessment. ISPRS

International Journal of Geo-Information. 2014;3:1101-17.

[155] MacRae R, Gallant E, Patel S, Michalak M, Bunch M, Schaffner S. Could Toronto

provide 10% of its fresh vegetable requirements from within its own boundaries?

Matching consumption requirements with growing spaces. Journal of Agriculture, Food

Systems, and Community Development. 2010;1.

136

[156] McClintock N, Cooper J, Khandeshi S. Assessing the potential contribution of

vacant land to urban vegetable production and consumption in Oakland, California.

Landscape and Urban Planning. 2013;111:46-58.

[157] Ryerson NB. Possibilities For the Urban Grower: Finding Sites in the City of

Atlanta using Geographic Information Systems. Thesis, Georgia State University; 2015.

[158] Kodysh JB, Omitaomu OA, Bhaduri BL, Neish BS. Methodology for estimating

solar potential on multiple building rooftops for photovoltaic systems. Sustainable Cities

and Society. 2013;8:31-41.

[159] ICIC. Boston Food System Resilience Study. Boston, MA: Initiative for a

Competitive Inner City; 2015. p. 36.

[160] ICIC. Resilient Food Systems, Resilient Cities: Recommendations for the City of

Boston. Boston, MA: Initiative for a Competitive Inner City; 2015.

[161] Allen P, FitzSimmons M, Goodman M, Warner K. Shifting plates in the agrifood

landscape: the tectonics of alternative agrifood initiatives in California. Journal of rural

studies. 2003;19:61-75.

[162] Massachusetts Office of Geographic Information (MassGIS). MassGIS Data.

Available from: <http://www.mass.gov/anf/research-and-tech/it-serv-and-

support/application-serv/office-of-geographic-information-massgis/> Accessed on

06/01/2015.

[163] City of Boston Department of Innovation and Technology. City of Boston Data

Portal. Available from: <https://data.cityofboston.gov/>, Accessed on 06/01/2015.

137

[164] Alexander C, Smith-Voysey S, Jarvis C, Tansey K. Integrating building footprints

and LiDAR elevation data to classify roof structures and visualise buildings. Computers,

Environment and Urban Systems. 2009;33:285-92.

[165] Tufts University. Boston GIS Database. Available from:

<http://sites.tufts.edu/gis/>, Accessed on 06/01/2015.

[166] USDA Natural Resources Conservation Service. Soil Survey Geographic Database

- Suffolk County. 2014. Available from:

<http://websoilsurvey.sc.egov.usda.gov/App/WebSoilSurvey.aspx>, Accessed on

06/01/2015.

[167] USDA Agricultural Research Service. Plant hardiness interactive map. Available

from: <http://planthardiness.ars.usda.gov/phzmweb/interactivemap.aspx> Accessed on

05/15/2016.

[168] Duchemin E, Wegmuller F, Legault A-M. Urban agriculture: multi-dimensional

tools for social development in poor neighbourhoods. Field Actions Science Reports: The

Journal of Field Actions. 2008;1.

[169] Howarth R, Anderson D, Cloern J, Elfring C, Hopkinson C, Lapointe B, et al.

Nutrient pollution of coastal rivers, bays, and seas. Issues in Ecology. 2000;7:1-14.

138

APPENDIX

Table A1: USGS land cover classification [126]

Land use

Code

Land use

description

Detailed definition

1 Cropland Generally tilled land used to grow row crops.

Boundaries follow the shape of the fields and

include associated buildings (e.g., barns). This

category also includes turf farms that grow sod.

2 Pasture Fields and associated facilities (barns and other

outbuildings) used for animal grazing and for the

growing of grasses for hay.

3 Forest Areas where tree canopy covers at least 50% of the

land. Both coniferous and deciduous forests belong

to this class.

4 Non-Forested

Wetland

DEP Wetlands (1:12,000) WETCODEs 4, 7, 8, 12,

23, 18, 20, and 21.

5 Mining Includes sand and gravel pits, mines and quarries.

The boundaries extend to the edges of the site’s

activities, including on-site machinery, parking lots,

roads and buildings.

6 Open Land Vacant land, idle agriculture, rock outcrops, and

barren areas. Vacant land is not maintained for any

evident purpose and it does not support large plant

growth.

7 Participation

Recreation

Facilities used by the public for active recreation.

Includes ball fields, tennis courts, basketball courts,

athletic tracks, ski areas, playgrounds, and bike

paths plus associated parking lots. Primary and

secondary school recreational facilities are in this

category, but university stadiums and arenas are

considered Spectator Recreation. Recreation

facilities not open to the public such as those

belonging to private residences are mostly labeled

with the associated residential land use class not

participation recreation. However, some private

facilities may also be mapped.

139

8 Spectator

Recreation

University and professional stadiums designed for

spectators as well as zoos, amusement parks, drive-

in theaters, fairgrounds, race tracks and associated

facilities and parking lots.

9 Water-Based

Recreation

Swimming pools, water parks, developed freshwater

and saltwater sandy beach areas and associated

parking lots. Also included are scenic areas

overlooking lakes or other water bodies, which may

or may not include access to the water (such as a

boat launch). Water-based recreation facilities

related to universities are in this class. Private pools

owned by individual residences are usually included

in the Residential category. Marinas are separated

into code 29.

10 Multi-Family

Residential

Duplexes (usually with two front doors, two

entrance pathways, and sometimes two driveways),

apartment buildings, condominium complexes,

including buildings and maintained lawns.

Note: This category was difficult to assess via photo

interpretation, particularly in highly urban areas.

11 High Density

Residential

Housing on smaller than 1/4 acre lots. See notes

below for details on Residential interpretation.

12 Medium Density

Residential

Housing on 1/4 - 1/2 acre lots. See notes below for

details on Residential interpretation.

13 Low Density

Residential

Housing on 1/2 - 1 acre lots. See notes below for

details on Residential interpretation.

14 Saltwater Wetland DEP Wetlands (1:12,000) WETCODEs 11 and 27.

15 Commercial Malls, shopping centers and larger strip commercial

areas, plus neighborhood stores and medical offices

(not hospitals). Lawn and garden centers that do not

produce or grow the product are also considered

commercial.

16 Industrial Light and heavy industry, including buildings,

equipment and parking areas.

17 Transitional Open areas in the process of being developed from

one land use to another (if the future land use is at

all uncertain). Formerly identified as "Urban Open".

18 Transportation Airports (including landing strips, hangars, parking

areas and related facilities), railroads and rail

stations, and divided highways (related facilities

would include rest areas, highway maintenance

areas, storage areas, and on/off ramps). Also

140

includes docks, warehouses, and related land-based

storage facilities, and terminal freight and storage

facilities. Roads and bridges less than 200 feet in

width that are the center of two differing land use

classes will have the land use classes meet at the

center line of the road (i.e., these roads/bridges

themselves will not be separated into this class).

19 Waste Disposal Landfills, dumps, and water and sewage treatment

facilities such as pump houses, and associated

parking lots. Capped landfills that have been

converted to other uses are coded with their present

land use.

20 Water DEP Wetlands (1:12,000) WETCODEs 9 and 22.

23 Cranberry bog Both active and recently inactive cranberry bogs and

the sandy areas adjacent to the bogs that are used in

the growing process. Impervious features associated

with cranberry bogs such as parking lots and

machinery are included. Modified from DEP

Wetlands (1:12,000) WETCODE 5.

24 Powerline/Utility Powerline and other maintained public utility

corridors and associated facilities, including power

plants and their parking areas.

25 Saltwater Sandy

Beach

DEP Wetlands (1:12,000) WETCODEs 1, 2, 3, 6,

10, 13, 17 and 19

26 Golf Course Includes the greenways, sand traps, water bodies

within the course, associated buildings and parking

lots. Large forest patches within the course greater

than 1 acre are classified as Forest (class 3). Does

not include driving ranges or miniature golf courses.

29 Marina Include parking lots and associated facilities but not

docks (in class 18)

31 Urban

Public/Institutional

Lands comprising schools, churches, colleges,

hospitals, museums, prisons, town halls or court

houses, police and fire stations, including parking

lots, dormitories, and university housing. Also may

include public open green spaces like town

commons.

34 Cemetery Includes the gravestones, monuments, parking lots,

road networks and associated buildings.

35 Orchard Fruit farms and associated facilities.

141

36 Nursery Greenhouses and associated buildings as well as any

surrounding maintained lawn. Christmas tree (small

conifer) farms are also classified as Nurseries.

37 Forested Wetland DEP Wetlands (1:12,000) WETCODEs 14, 15, 16,

24, 25 and 26.

38 Very Low Density

Residential

Housing on > 1 acre lots and very remote, rural

housing. See notes below for details on Residential

interpretation.

39 Junkyard Includes the storage of car, metal, machinery and

other debris as well as associated buildings as a

business.

40 Brushland/Success

ional

Predominantly (> 25%) shrub cover, and some

immature trees not large or dense enough to be

classified as forest. It also includes areas that are

more permanently shrubby, such as heath areas,

wild blueberries or mountain laurel.

142

Table A2: Estimated marginal land for MAPC cities

Town

Marginal land

(ha)

Total land area

(ha)

Land

Percentage

BOSTON 2914 12,825 23%

MARSHFIELD 2112 7446 28%

FRANKLIN 1850 6988 26%

CONCORD 1569 6684 23%

MARLBOROUGH 1548 5706 27%

FRAMINGHAM 1546 6869 23%

NEWTON 1527 4711 32%

ACTON 1375 5247 26%

HOPKINTON 1350 7219 19%

SUDBURY 1346 6411 21%

WALPOLE 1312 5467 24%

IPSWICH 1265 8680 15%

NATICK 1236 4138 30%

PEMBROKE 1229 6100 20%

WEYMOUTH 1219 4581 27%

DUXBURY 1217 6273 19%

HANOVER 1190 4060 29%

LEXINGTON 1168 4309 27%

NORWELL 1158 5490 21%

SCITUATE 1122 4475 25%

HOLLISTON 1108 4933 22%

NEEDHAM 1102 3303 33%

DANVERS 1099 3578 31%

BRAINTREE 1075 3720 29%

BELLINGHAM 1039 4898 21%

SHARON 1012 6324 16%

CANTON 996 5053 20%

READING 953 2583 37%

WILMINGTON 947 4442 21%

WESTON 903 4483 20%

NORTH READING 901 3496 26%

MEDWAY 892 3021 30%

HINGHAM 881 5892 15%

FOXBOROUGH 881 5399 16%

LITTLETON 859 4541 19%

PEABODY 850 4350 20%

BEDFORD 847 3589 24%

WAYLAND 838 4104 20%

STOUGHTON 809 4262 19%

143

ROCKLAND 803 2624 31%

BEVERLY 799 4006 20%

QUINCY 771 4335 18%

WRENTHAM 759 5859 13%

MEDFIELD 727 3793 19%

WALTHAM 723 3565 20%

GLOUCESTER 712 6939 10%

MILTON 709 3418 21%

NORFOLK 684 3987 17%

ASHLAND 670 3334 20%

HAMILTON 647 3873 17%

WELLESLEY 630 2720 23%

STOW 625 4672 13%

SHERBORN 622 4180 15%

TOPSFIELD 616 3314 19%

WESTWOOD 614 2889 21%

MILLIS 599 3180 19%

DOVER 594 3998 15%

CARLISLE 576 4012 14%

ESSEX 575 3769 15%

WOBURN 573 3357 17%

MIDDLETON 573 3751 15%

DEDHAM 572 2765 21%

HUDSON 562 3074 18%

LINCOLN 549 3881 14%

BROOKLINE 520 1767 29%

WAKEFIELD 518 2066 25%

AYER 484 2461 20%

COHASSET 435 2604 17%

LYNNFIELD 422 2711 16%

BOXBOROUGH 392 2691 15%

SAUGUS 368 2957 12%

ROCKPORT 363 1849 20%

HOLBROOK 345 1918 18%

WENHAM 344 2102 16%

WINCHESTER 336 1649 20%

NORWOOD 335 2710 12%

RANDOLPH 304 2708 11%

MANCHESTER 235 2020 12%

BURLINGTON 229 3071 7%

MEDFORD 225 2196 10%

MAYNARD 194 1401 14%

SALEM 181 2191 8%

REVERE 168 1540 11%

144

MALDEN 157 1313 12%

BELMONT 156 1234 13%

CAMBRIDGE 147 1856 8%

CHELSEA 145 572 25%

MELROSE 117 1229 9%

STONEHAM 85 1726 5%

HULL 79 755 10%

MARBLEHEAD 73 1155 6%

LYNN 68 2965 2%

SOMERVILLE 63 1072 6%

SWAMPSCOTT 60 787 8%

ARLINGTON 36 1408 3%

WATERTOWN 23 1067 2%

NAHANT 6 296 2%

EVERETT 4 893 0%

Total 71,150 361,885 20%