TOWARDS VALIDATED URBAN PHOTOVOLTAIC POTENTIAL AND SOLAR RADIATION MAPS BASED ON LIDAR MEASUREMENTS, GIS DATA, AND

HOURLY DAYSIM SIMULATIONS

J. Alstan Jakubiec1, alstan@jakubiec.net Christoph F. Reinhart1, tito_@mit.edu

1Building Technology Program, Massachusetts Institute of Technology, Cambridge, MA

ABSTRACT

We present a method for generating detailed geometric urban massing models combined with building footprint and material information from large GIS datasets and LiDAR elevation measurements. An example model for the city of Cambridge, Massachusetts, USA that contains over 17,000 buildings is used as input for annual solar radiation calculations using the RADIANCE / DAYSIM simulation engine. Based on hourly irradiance calculation results, we find it possible to make recommendations for PV placement on a building and to intelligently determine the total and useful roof area of buildings. Simulation results are compared to those typically used in practice to produce solar radiation maps of other US cities. It is found that the presented method yields better geometric accuracy and higher irradiation predictions compared to previous methods. This results in increased predicted PV energy production at lower installation costs and more accurate estimates of useful rooftop area.

INTRODUCTION

It has become increasingly popular for cities and municipalities to create solar potential maps with the intent of promoting renewable energy generation through photovoltaic (PV) panel installations within those jurisdictions. In the United States, large cities such as Boston, Los Angeles, New York City and Portland provide online maps which allow building owners to look up their address and view personalized predictions of,

electric production from a PV system (kWh)

energy savings from a SHW system (Therms)

resulting annual electricity savings (dollars)

carbon savings (lbs)

useful roof area (sq. ft.)

system payback period (years)

system costs (dollars)

local rebates and incentive programs

The objective of these maps and accompanying personalized property information is to reduce summer time peak loads, increase the environmental awareness of residents, reducing greenhouse gas emissions and to improve the sustainable image of a city through the expansion of solar energy technology.

While a number of cities have already generated such solar maps, to the authors’ knowledge, limited attention has been paid to the assumptions and calculation methods underlying these maps. The objectives of this paper are hence threefold.

We initially present a method of how a validated solar radiation calculation algorithm, thus far only been used at the individual building scale, can be applied to a city-sized model of Cambridge, Massachusetts, USA. The new method creates city-wide solar potential maps with a high degree of spatial and predictive accuracy based on the generation of a high resolution three-dimensional (3D) model sourced from available geographic information systems (GIS) data. Our model applies validated simulation methods which take into account detailed geometric data including shadowing from surrounding buildings, typical climate data, and reflections between buildings and the urban landscape. The results from this model are spatially and temporally rich; the variation of irradiation across a rooftop is displayed, and data is available at hourly time-steps for detailed peak load analysis and PV energy yield calculations. Secondly, we use the simulation results generated during the first step as a reference against which we compare results that one would obtain using the methods underlying other solar radiation maps. Finally, we discuss what relevance varying simulation results may have at both the individual building owner and city-wide policy level.

REVIEW OF CURRENT METHODS

Irradiation Calculations

In a review of eleven solar potential maps for North American cities (Table 1), we found that there are three typical predictive methodologies in place for calculating rooftop irradiation. Three (27%) of the surveyed maps used a constant assumption for solar

irradiation reaching a building rooftop. One (9%) reported using the National Renewable Energy Laboratory’s (NREL) PVWatts calculation module. Another five (45%) used the Solar Analyst plugin within Esri’s ArcGIS program. The remaining maps did not report their calculation methodology.

The use of a constant, global horizontal, solar radiation value across a rooftop will be inaccurate in many cases, for example buildings with peaked roofs. Such a constant value also does not consider local urban context such as trees and neighboring buildings which shade building rooftops. Also, the complex forms of individual roofs are ignored. Advocates of this approach determine the useful roof area for PV installation by using either a constant percentage (Boston, Portland) or based on orthophotographic image analysis techniques (San Francisco, Berkeley). The NREL PVWatts module is essentially a modified version of the constant approach (Marion, et al. 2001). Solar irradiation is distributed on a 40km square grid for the entire United States based on the typical meteorological year 2 (TMY), dataset. Local TMY2 irradiation data is used in combination with PV panel tilt, orientation, and urban temperature conditions to determine energy production. While roof shape is treated with greater detail than in a solar constant approach, shading from adjacent urban surfaces also cannot be modeled using the PVWatts module.

Since Esri’s Solar Analyst plugin, based on the work of Fu and Rich (1999), currently constitutes the most widely used irradiation calculation method, it is discussed in greater detail. In this method, a sky mask is initially generated based on the surroundings of each pixel of a digital elevation model (DEM). A DEM is a geolocated raster image where the values of individual pixels correspond to elevation measurements. The direct and diffuse components of irradiation are calculated based on what amount of the sky can be seen from each pixel. Direct irradiation is calculated in accordance with the sun position, the slope of the DEM, a fixed transmissivity coefficient, and the

distance a solar ray must travel through the atmosphere. Diffuse irradiation is calculated in much the same way as the direct component, based on either a uniform sky model or a standard overcast model; however, no solar map reports on its website which sky model was used.

In Solar Analyst, sky transmissivity and the ratio between direct and diffuse insolation are fixed, constant values throughout the year. These assumptions can have a significant effect on the calculated annual radiation. For example, the Boston Logan TMY3 weather data illustrates a ratio between direct and diffuse irradiation which varies widely throughout the year (US Department of Energy 2012). The mean value of the hourly direct-to-total ratio of insolation is 64.2% for the 4,604 daylit hours in the Boston TMY3 weather file; however, the standard deviation from the mean is 31.3%. Further, the amount of cloud cover and thus atmospheric transmissivity varies throughout the year. Therefore, it is inaccurate to pick one value to adequately represent these factors. The extreme variance in direct and diffuse radiation throughout the year and cloud cover is shown in Figure 1 for the Boston climate TMY3 dataset.

As Solar Analyst uses only a sky mask based on a DEM, it has no capacity to model reflections between buildings, from surrounding trees or from the urban terrain. It has been proposed to assume a directional constant of reflected irradiation for obscured sky areas (Rich, et al. 1994), but it would be inadequate to consider complex reflections from surrounding buildings.

Table 1 Survey of Existing Solar Potential Mapping Methods in North America

CITY URL FLAT ROOF

CALCULATION METHOD

Anaheim Berkeley Boston Denver Los Angeles County Madison New York City Portland Salt Lake City San Diego San Francisco

http://anaheim.solarmap.org/ http://berkeley.solarmap.org/ http://gis.cityofboston.gov/SolarBoston/ http://solarmap.drcog.org/ http://solarmap.lacounty.gov/ http://solarmap.cityofmadison.com/madisun/ http://nycsolarmap.com/ http://oregon.cleanenergymap.com/ http://www.slcgovsolar.com/ http://sd.solarmap.org/solar/index.php http://sf.solarmap.org/

No Yes Yes

? No ?

No Yes No ?

Yes

Solar Analyst Constant Multiplied by Usable Roof Area Solar Analyst Unknown Unknown (Assumed, Solar Analyst) PVWatts with Sunlit Hours Graphic Solar Analyst Constant Assumption Solar Analyst Unknown Constant Multiplied by Usable Roof Area

Figure 1Cloud Cov

Observed

Of the 11 assumed thknown elerepresent ttheir assumassumptionroof is suirelied on locating Francisco)determinedthe number

None of considers r

METHOD

Our implembased on tvalidated Rdaylight siWalkenhor

LiDAR Da

The geomedigital mofunded 20And Rangisystem whbursts andwhile trac

1 Hourly Direcver from Bosto

Geometric an

solar potentiahat all buildingevation, four (3the roof, and tmptions. Of n, half assumedtable for PV (a proprietary rooftop ob

. With DEMsd by the predir of daylit hour

the surveyed reflections or b

DOLOGY

mentation of athe creation oRADIANCE/Dimulation engirst 2001).

ata and Geom

etric informatioodel of Cambr10 LiDAR suring, is an estab

herein a surveyd records the cking its loca

ct and Diffuse Ron Logan TMY3

nd Material A

al maps surveygs in the city ha36%) used a dthe remaining cities utilizingd that a fixed p(Boston, Portla

orthophotograstructions (Bs, useful rooficted rooftop irs.

cities use a building materi

an urban solar f a detailed 3

DAYSIM backine (Ward 199

metric Accurac

on used in crearidge comes frrvey. LiDAR, blished, accuraying aircraft etime until the

ation via Glo

Radiation and Y3 Weather Dat

ssumptions

yed, four (36%ad flat roofs at

detailed DEM 3 did not repog the flat ropercentage of thand). The otheaph method fBerkeley, Saf area is eithirradiation or b

method whicals.

potential map D model in th

kward-raytracin95, Reinhart an

cy

ating the detailefrom a publicl

Light Detectioate measuremeemits rapid laseir visual retuobal Positionin

ta

%) t a to

ort of he ers for an

her by

ch

is he ng nd

ed y-on ent ser urn ng

Systemreturn locatedthe urbthan 1selecteand trato be 0

Our pillustraOval (Figurein planwhere on the126,62which (18.5 kover aspacingmultipwhich were ddata spmuch g

The sdividedGIS dand gr2004). trianguresultinthe en16,547

Hourly

The trithe RRADIAcustomassumewith agroundAnnuabuildin(1.5x1approxnormal

SimulaRADIAapproaet al. irradiaDAYSto a sk

ms (GPS). Thdata is late

d point data. Tban context of1m root meaned validation taditional GPS 0.062m (Allian

process of crated with an at the Massae 2a). As LiDAn, it creates andifferent poin

e airplane pat24,764 points

has a total akm2). We unifa plan grid ofg, taking the m

ple points exisdid not vary b

discarded. Thispace to a mergeometric reso

simplified LiDd into two ca

datasets from tround scape As a last s

ulated using a ng in a highly ntire City o

7,790 triangula

y Simulations

iangulated surfRADIANCE bANCE each su

mized, optical sed that buildina 35% reflecd were assignal irradiation ng roof surfac.5m). Simula

ximately 1/64”l direction of th

ations are perfANCE plug-inach and the Pe1993) to pred

ation (ReinhaSIM works by ky dome consi

he collected er processed The vertical accf Cambridge wn squared errotests the RMSmeasurement

nce for Sustaina

reating a detaexample surro

achusetts InstitAR data is non awkward datnt densities areth of flight. I

spread acrossarea of approxformly resampf approximatelmean of the firsted. Resultingby greater thans resulted in a re 9,403,750 polution.

DAR-derived ategories usingthe City of C(Figure 2c) (

step the two a Delaunay alg

accurate and dof Cambridgear surfaces.

s with Radianc

face model wasbackward rayurface may hasurface proteri

ng walls were Lctance while ned a diffuse r

was then cce at a grid

ation sensor ” (0.4mm) abohe roof surface

formed with Dn which uses aerez all weathedict annual poart and Wperforming onsting of 145 d

location and into geographcuracy of the dwas bounded tor (RMSE), aSE between Lmethods was able Energy 20

ailed 3D moounding the Ktute of Technt uniformly sata space (Figue present depeInitially, theres all of Cam

ximately 4,500pled the LiDARly 4’x4’ (1.2xrst return data g neighboring n one foot versimplification

points without

points wereg publicly ava

Cambridge: bui(City of Cam

point groupsgorithm (Figurdetailed 3D moe that consis

ce/DAYSIM

s then converteytracer formaave different, ies. In our modLambertian difbuilding roofreflectance of calculated on resolution of points are lve and facing e.

DAYSIM, a vala daylight coefer sky model (int illuminatio

Walkenhorst ne raytrace opediffuse sky segm

timed hically data in to less and in LiDAR shown

010).

del is Kresge nology ampled ure 2b) ending e were

mbridge 0 acres R data x1.2m) where points

rtically of the losing

then ailable ildings

mbridge were re 2d), odel of sts of

ed into at. In highly del we ffusers fs and f 20%.

each 5’x5’

ocated in the

lidated fficient (Perez, on and 2001). eration ments,

3 ground approximadistributedfrom the sithen weighin the scsimulated astep withouraytrace ca

In order simulation the unusuErrors in thacceptable larger. As in plan anbeyond treasonablesurfaces spthe ambien(26,526.5 resolution the DAYS

(a) Bir

(c) R

segments and ately 65 dired along the solimulation senshted relative tocene. In this across an entirut running thou

alculations for

to ensure parameters w

ally large sizhe ambient cal

for surfaces our model wa

nd our simulatthis threshold

e. According topaced closed tnt resolution.”

ft.) divided of approximatIM simulation

rds-eye image o

esampled and

Figur

a second raytct solar posar path. By tra

sor points, each its contributiomanner, irra

re year in any iusands of sepaeach time step

accuracy, thwere considereze of the Camlculation were

spaced four as resampled ation sensor pod, the assu

o Ward, error whan the scene Thus the Radby four giv

tely 6,750. Tabparameters us

of the Kresge O

categorized Li

re 2 Process Im

tracing run wiitions that aacing backwarh sky segment ons to each poiadiation can bincremental timarate and length.

he RADIANCed in relation mbridge modecalibrated to bfeet apart an

at this resolutiooints are spaceumption seemwill “increase o

size divided bdiance scene sizves an ambieble 2 documened.

Oval (Google)

iDAR points

mages of 3D Mo

ith are ds is

int be

me hy

CE to el. be nd on ed ms on by ze

ent nts

Table 2

PARA

Calcul

As preis directhe deradiancKnowisimulaclimatethe per

A direbased oAssumthe aremethod

(b) In

odel Generatio

2 Key Radianc

AMETER ab ad as ar aa

amamamamam

lation of Phot

eviously discusct access to ho

etailed Perez sce distributioning in additioated point ane, a reasonablrformance of a

ection vector ison the roof sur

ming that the roea the point d of calculating

Initial, nonunifo

(d) Re

on from LiDAR

ce/DAYSIM Sim

DESCRIPTImbient bounces mbient divisions mbient super-sammbient resolutionmbient accuracy

tovoltaic Poten

ssed, a key beourly simulatedsky model thatns for each timon the explicitnd informationle approximatia PV panel in an

s assigned to erface normal imoof is planar arepresents, ~2g the area is sh

formly dense Li

esulting 3D mo

R and GIS Data

mulation Param

ION VA

mples n

2 204816 67500.1

ntial

enefit of this md irradiation dat mimics actume step in thet area beneathn about the on can be man urban contex

ach simulationmmediately beand unvarying 25ft2 in our chown in Equati

iDAR point dat

odel

a

meters

ALUE

8

0

method ata and ual sky e year. h each

urban ade for xt.

n point low it. below

case, a ion 1,

ta

0 0 1 ∙

(1)

where is the unitized roof surface normal vector.

We further realize that photovoltaic performance is dependant on many factors which are unknown at the time of making a conceptual irradiation map such as module efficiency, panel orientation and maintainance. However, it is known that temperature and radiation heating up the panel will have an adverse effect on its production. The urban ambient temperature (Tamb, ° ) and the incident irradiation (E, Wm-2) at each timestep can be used to estimate the temperature of the photovoltaic panel (Tc), shown in Equation 2, by relying upon knowledge of the nominal operating cell temperature at ideal conditions (T0) (Luque and Hegedus 2011) Further, the photovoltaic maximum power at ideal conditions ( , ) can be derated based on a temperature correction factor ( ) equal to 0.0038° (Equation 3) (Marion, et al. 2001).

20° /800 (2)

∗ 1 ∗ (3)

Equations 2 and 3 above are used as a first-order approximation in derating panel efficiency based on temperature and point irradiation at each hourly timestep.

Determination of Useful Rooftop Area

Useful rooftop area in our model is calculated based on predicted economic feasibility of panels installed at a location. Further, any roof surface sloping greater than 60 degrees (67%) was discarded as it is approaching being considered a vertical surface or wall.

On average, PV installations cost approximately $5.67 per watt in Cambridge in 2011 (MassCEC, 2012). We assume that a panel is rated at 17.2 W/ft2 (185W/m2) (Sunpower E18/230W 2012), installation cost is $97.52/ft2 ($1049.70/m2), and the cost of electricity is $0.15/kWh. If we consider a 10 year investment period with a 10% discount rate per year, 115.7 kWh/ft2/yr (1244.9 kWh/m2/yr) must be generated to have a net present value (NPV) in which the investment breaks even (NPV equals zero). In ideal circumstances this would require a panel efficiency of nearly 80% in Cambridge! To achieve a simple payback over the same 10 year period, only 65kWh/ft2/yr (699.7 kWh/m2/yr) must be generated. Still, this requires a panel efficiency of approximately 50%.

However, there are national and state rebate programs that dramatically improve the feasibility of the installation of PV for residential properties. The federal government offers a 30% tax rebate on the cost of a PV installation up to a maximum of $2,000 (Energy

Improvement and Extension Act 2008). Further, Massachusetts offers a 15% rebate up to a maximum of $1,000 that can be carried over for three years (Residential Renewable Energy Income Tax Credit 1979). The Massachusetts Clean Energy Center offers a minimum $0.40/W rebate on new PV systems (MassCEC 2012). Finally, Massachusetts offers a 100% protection from increased property taxes due to PV installations for a 20 year period (Renewable Energy Property Tax Exemption 1975). Factoring these rebates into the previous NPV calculation, it is possible to have a break even point for an unshaded panel at ~24% efficiency, which is still not an ideal financial incentive, but things are markedly improved. Looking at simple payback over a period of 10 years for our example Sunpower panel, any point with over 56.6 kWh/ft2 (609 kWh/m2) irradiation per year is likely to recoup its value while providing additional savings after the initial 10 year period as the effective lifetime of a PV system is known to be typically greater than 30 years. Thus, points with greater than 609 kWh/m2 and their associated roof areas are considered to be useful to install PV panels. As such sensor points are displayed spatially across the roof (see results section), it is possible to determine optimal placement locations for PV panels.

Geolocation of Data From GIS to Radiance Models

All GIS models including the LiDAR data and building footprints were constructed in the projected North American Datum 1983, Massachusetts State Plane Mainland coordinates system (Schwarz and Wade 1990). This is a serendipitous choice as distances and areas can still be measured without necessitating corrections. Thus, the Radiance simulation model was built in an identical coordinate system. The Massachusetts State Plane system also has a known relationship between X and Y coordinates and latitude and longitude global coordinates. It is possible to translate easily between the two coordinate systems by use of an Inverse Lambert Conformal Conic Projection with proper parameters.

RESULTS In this section, the results from methodologies discussed in the “Review of Current Methods” section are compared with our own detailed method. Ten buildings are used for the purposes of the comparison from the over 17,000 in the Cambridge dataset. Of these ten buildings, five can be described as having flat roofs; however, they often have HVAC equipment present on the roof such that it is not truly flat. The other five have roofs of some complexity with at least one ridge line. These test buildings are shown in Figure 3. For our 5’x5’ grid of analysis points across each building, annual and monthly irradiation data was used

Table 3Compa

METH

Solar A

Flat RConstant

(a) ComComplex

F

3 Calculated Rared to Detailed

HOD RM(TO

nalyst Roof t Value

334853

mparison of DRoofs and Sol

Figure 4 Comp

Figur

RMSE of Varioud 3D DAYSIM

MSE OTAL)

RMSE(FLAT

39.2 89.9 32.1

365.8456.5488.5

Figure 3

etailed DAYSIMlar Analyst with

parison of Glob

re 5 Resulting P

us AssumptionsM Model Method

E T)

RMSE(COMPLE

8 5 5

274.1554.2614.7

Ten Test Build

IM Model with h Detailed DEM

bal Irradiation

Point Irradiatio

s d

EX)

Table 4 A

METH

DetaiSolar A

Flat RConstant

dings Used in C

M (b) C

n Calculation U

on Maps For

Annual PV ProDetermined b

HOD PV (MW

iled Analyst Roof t Value

2233

Comparing Res

Comparison ofComplex Ro

Using Varied M

Varying Calcu

oduction, Roofby Various Ass

PROD. Wh/yr)

ARE(ft

2,445 2,148 3,329 3,473

113,112,143,143,

sults

f Detailed DAYoofs and with F

Methods for 10

ulation Method

f Area and Cossumptions

EA t2)

INSTACOST

863 179 525 548

$11,104$10,940$13,997$13,999

YSIM Model wiFlat Roofs

Buildings

ds

sts as

ALL ($)

4,375 0,144 7,132 9,375

ith

as a basis oof data tha

Table 3 scompared should notmethods wRMSE hermodel. It roof geomSolar Anawith compexpected tconsideratiof a consta

Comparis

When cosimulationannual radcumulativeacross the climactic mAnalyst (Yclustered aof variance

It is expeaverage prground, tconsideredto be zerosuggests soexplained models atphotographirradiation which is tprocess ofbuilding poimprove thAs Solar Anot differecalculated to stratifieFig. 4(a) aAnalyst). Sof building

It is usefulyields. Uirradiation area, the calculated.2,445MWhcombined Comparati2,148MWhroof area.

of comparison t Solar Analys

shows the Rto the detaile

t be interpretewill necessarire represents dcan be seen t

metry is importaalyst has the splex roofs at athe flat roof ion of climate,ant solar irradia

on with Solar

ompared to , Solar Analydiation for eae annual irradi

ten buildings model (X-Axi

Y-Axis). It canabout the idente; the RMSE is

ected that a rredict higher vtrees and add where Solar o. However, thome other effeby the geome

t building edhs and planar for a single bu

typical across f creating a 3Dolyline from Ghe model resolAnalyst works aentiate betweeslope at edge

ed errors as seand the upperSuch an error g perimeter to a

to further quaUsing a simp

multiplied bytotal electrica Using the deh/yr was predi113,863 ft2 (1vely, the Solah/yr across 112These differen

because the sht can return is m

MSE of the ed climate-based as absoluteily introduce deviation fromthat consideraant to accuratesmallest RMSan error of 274

method, even, performs pooation performs

Analyst

a climate-bayst appears to

ach point. Figiation for each

as calculated s) and as calc

n also be seen tity line, there is 339.2 kWh/m

raytracing simvalues as refledjacent buildin

Analyst assumhe high varianect is at work, etric quality ofdges. Figure 5

projections ouilding using dthe ten test b

D model of a GIS to create ex

ution at the edacross a pure Den building apixels is extrem

een in the extredge pixels i

will increase warea increases.

alify this data inplified equatiy panel efficienal yield of setailed climatected across all0,578 m2) of uar Analyst m2,179 ft2 (10,42nces may not s

hortest timescamonthly.

other methosed model. The error, as bo

errors. Insteam a best-practiation of detailee calculations E for buildin4.1 kWh/m2. An with detaile

orly. Assumptioworst.

ased raytracino estimate le

gure 4(a) showh sensor locatio

by the detaileculated by Solthat while data is a large degr

m2.

mulation will oections from thngs are bein

mes such effecnce of the daand this can b

f the simulatio5 shows aeriof the predicteifferent metho

buildings. In thcity we use th

xtra points whicdge of buildingDEM which doand ground, thme and can leareme outliers in Fig. 5 (Solwith as the rat

nto PV electricion of annuncy (0.185) ansystems can be-based methol rooftops withuseful roof are

method predicte21 m2) of usef

seem large whe

ale

ds his oth ad ce ed as gs As ed on

ng ess ws on ed lar is ee

on he ng cts ata be on ial ed ds he he ch gs. es he ad in

lar tio

cal ual nd be

od, h a ea. ed ful en

lookinghowevcosts revealspredictAnalysdifferedocum

To attthe detwe chothe moshownmonthlHere climatedoes Srelativeradiatioit mayperiodparame

FigBui

Comp

A secsimulaforms derived

g only at predver, putting th

(dollars) pers a differentions lead to st predictions ence of 10.8

mented in Table

tempt to quantailed climate-ose a test buildonthly total i

n in Figure 6ly global horizagain we obse-based model

Solar Analyst, fe error is greon. As these m

y be that the es with less eters of Solar A

gure 6 Monthlyilding Colored

parison with F

ond comparisations for the

and with flat d by discard

dicted energy phe figures in t

predicted ennt story. Tha cost of $4.5lead to a cos8 percent. e 4.

ntify temporal -based method

ding with little irradiation at where points

zontal radiationserve that forl calculates grefor the same reeater in monthmonths have lerror will be grdirect radiatio

Analyst.

y Point Irradiad by Monthly G

lat Roof Assu

son was run uten buildings roofs. The fla

ding points w

production in Mterms of instanergy yield he detailed 54/kWh whilet of $5.09/kWThese results

differences bed and Solar Anvariation and peach point. Tare colored b

n in the TMY3r most montheater irradiatioeasons; howevhs with less oss direct compreater in climaon for the d

ation for a SingGlobal Irradiati

umptions

using climate-with complex

at roof heightswith a z-coor

MWh; allation (kWh) model Solar

Wh – a s are

etween nalyst, plotted This is by the 3 data. hs the n than

ver, the overall ponent, ates or default

gle ion

-based x roof s were rdinate

greater thausing the m

Figure 7 shcontext wiroof; eaccalculationmethod (Ralone, it isgoing to cris because radiation asky. Figurwith pointcomplex roroof assuassumptionmajority oline. The surfaces thsurfaces thlook straigbuilding. equipment of the roofour ten testinter-buildrise buildin

Comparis

If for everyfrom the T532.1 kWhwith detailconstant amethod ofconsider building sbuildings.

F

an one standardmedian value o

hows an irradiith a detailed rch image dns of the areRobinson and S

clearly visiblereate extreme roof slope wil

as well as the re 4(b) illustrat irradiation froof forms on mption on tn overestimateof data points

reason for thhat slope to that face surrought up towarFurther, it or roof projec

f. The effects ot cases, are acting shading fongs such as Ca

on with Const

y point the meaTMY3 data ish/m2 comparedled roof formsassumed valuf solar irradiatclimate-specifishading nor r

(a) City with D

Figure 7 Irradia

d deviation frof the remaining

ation map of aroof and with

displays detaiea using the Stone 2004). Fe that a flat roodifferences in ll change the ipercent and po

ates a scatter prom the detailthe X axis an

the Y axis. es available rare to the left

his is that thethe north, norunding buildinrds the open

is impossiblctions to shadeof roof orientatually higher thor a city of predmbridge.

tant Solar Ins

an global horizs assumed, thed to the clima. The large err

ue is the wotion calculatio

fic data, roofreflections fro

Detailed Roof

ation Maps (Cu

om the mean ang points.

a test building an assumed fl

iled irradiatiocumulative sk

From this imagof assumption the results. Thncident angle ortion of visibplot comparisoled model usinnd using the fl

The flat roradiation as thft of the identiere are no ror are there anngs; all surfac

sky above thle for HVA

e another portioation, in most han the effects dominantly low

olation Metho

zontal irradiatioen the RMSE ate-based modror shows thatorst performinons. It does nf shape, inteom neighborin

f Models

umulative Sky M

nd

in lat on ky ge is

his of

ble on ng lat of he ity of ny es he

AC on of of w-

od

on is

del t a ng

not er-ng

DISCAdmittsignificachievthe flcombinshouldthis ne

First, tcannotopportsubsetsfurtherwho mmodel walls; buildinhowevirradiavalidatin the based Using calculadetaileThe simcalculainformarea anenergy

Spatia

To comnecessthe undrooftopto rankinstallethe pre

Method) For V

CUSSION tedly, the detacantly more

ve when complat roof assumned with a de

d ask, what benew method?

the value of ht be underesttunities to invs of the modelr analysis by d

make policy. Fiin analyzing

It can be seenngs in the arever, in many loation >609 kted DAYSIM s simulation rweather inforDAYSIM a

ated irradiationed equations ofmulations can ation speed.

mation is availand the total incy generation eq

al Display of P

mmunicate phosary to providederstanding ofp irradiation fok the relative ed at that poineviously calcul

(b) C

Varying Degree

ailed climate-btime and prared to generamption or us

etailed DEM. Tnefits can be e

having a full 3timated. Such vestigate wall l can easily be esign teams origure 7 illustrag the solar pon that the wallea is less thanocations walls ckWh/m2. Secsoftware proviresults, considrmation and ralso provides n data which ff PV yield and

also be run inFinally, deta

able for quanticident irradiatioquations.

Photovoltaic P

otovoltaic potee useful visualf the data. We or Cambridge predicted perfnt. The threshlated 609 kWh

City with Flat R

es of Geometri

based methodrocessing powating a model sing Solar ATherefore the expected from

D model of tha model prmounted PV

extracted to sur government eates the utility otential of bul irradiation of

n rooftop insocan support PVcondly, usingides extra confiers typical clradiative reflec

access to hfacilitates the peak load redu

n parallel to inailed rooftopifying useful ron (kWh) used

Potential

ntial effectivell output that adivide the siminto four bins formance of aholds were bash/m2 cutoff val

Roofs

ic Accuracy

d takes wer to

using Analyst

reader m using

he city ovides

V, and upport

entities of this uilding f most lation;

V with g the fidence limate-ctions. hourly use of uction.

ncrease area ooftop

d in the

ly, it is aids in

mulated meant

a panel sed on lue for

Massachusthe Cambr1000 kWhthresholds

CONCLU

We partn(http://wwwresults on Google Mapproach tthe irradiatclimate. Inengage witheir roof produces installationsimulation which is imand to incr

To use vala new effowill suppoto predict h

Figure 8

setts resident’sidge simulatioh/m2 and 140are then used i

USION AND

nered with Mw.modestudiowtop of a search

Maps API (Fitaken by the LAtion thresholdsn this way, homith the map thspecifically an

varied suns. Essentially,

results are pmportant to prorease interest in

lidated irradiatort. We believeort policy decishourly peak-lo

8 Google Maps

s payback and n data. and are00kWh/m2. Thin the display o

D OUTLOO

Modern Develweb.com/) tohable map docigure 8). ThiA county solars are tailored tomeowners andhrough the abnd notice how uitability for, users of the mpersonalized tooduce confidenn the goals of th

tion models at that in the futsions as they aoad reduction a

s API Mockup

observations e at 600kWh/mhese irradiatioof data.

OK

lopment Studo display ocument using this is the samr map; howeveo the Cambridgd businesses cability to identi

its unique forr photovoltamap feel like tho their buildinnce in the resulhe map.

the city-scale ture such modeallow the abiliat an urban sca

of Spatial Roo

of m2, on

dio ur he

me er, ge an ify rm aic he ng lts

is els ity ale

or ammethodmodel least puser Furthewith th10.8 pthan th

Limita

Beyonseveralmodelefound flat roo45 degbrackeadditiopotentiSouth producnoted method

oftop Photovol

mong a groupds have not quality and ce

partially) visuaengagement

er, our results she city and higpercent cost rehe most popula

ations of Deta

nd what was dil limitations. ed as parallelfrom discussio

ofed buildings gree tilt towardets. Future veonal layer atoial photovolta

in this mances reasonably

that LiDAR d still introduc

ltaic Potential

p of buildingshad this bene

ertainty about rally assessed,

with sustainshowed good ggher predictioneduction for tear method.

ailed Climate-B

iscussed previoCurrently a

with the roofons with local often have PVds the South u

ersions of the p flat-roofed

aic installationnner. Further,y accurate roof

data and ouce errors in som

By Modern De

s whereas prefit. With incresults that canwe aim to in

nable technolgeometric agrens of PV yield en typical bui

Based Method

ously, the modall PV panelf; however, we PV contracto

V panels installusing standard

map will habuildings anas tilted toward while our

f forms, it shouur point-simplme buildings.

evelopment Stu

evious creased n be (at ncrease logies.

eement with a ildings

d

del has ls are e have rs that ed at a

d angle ave an alyzing ds the model uld be lifying

udio

ACKNOWLEDGEMENT This work was supported by the National Science Foundation project in the Office of Emerging Frontiers in Research and Innovation (award number: EFRI-1038264).

REFERENCES Berkeley Solar Map. Retrieved on 3/5/2012.

http://berkeley.solarmap.org.

City of Anaheim Solar Map. Retrieved on 3/5/2012. http://anaheim.solarmap.org.

City of Cambridge GIS Buildings Layer. 2004. Retrieved on 1/15/2012. http://gis.cambridgema.gov/DataDictionary/Datapages/Buildings.html.

Denver Regional Solar Map. Retrieved on 3/5/2012. http://solarmap.drcog.org.

Energy Improvement and Extension Act. Public Law 110-343. 122 Stat. 3807. 2008.

Fu, P and Rich, PM. 1999. Design and implementation of the Solar Analyst: an ArcView extension for modeling solar radiation at landscape scales. Proceedings of the Nineteenth Annual ESRI User Conference.

LA county Solar Map. Retrieved on 3/5/2012. http://solarmap.lacounty.gov.

LiDAR Campaign (Boston) Report of Survey. 2010. Alliance for Sustainable Energy, LLC.

Luque, A and Hegedus, S. 2011. Handbook of photovoltaic science and engineering, 2nd edition. John Wiley & Sons, Ltd.

MadiSUN. Retrieved on 3/5/2012. http://solarmap.cityofmadison.com/madisun.

Marion, B, et al. 2001. PVWATS version 2 -- Enhanced spatial resolution for calculating grid-connected PV performance. National Renewable Energy Laboratory publication NREL/CP-560-30941.

Massachusetts Clean Energy Center. Installers, Costs, Etc. Retrieved on 5/07/2012. http://masscec.com/masscec/file/Installers,%20Costs,%20Etc%20for%20Website%203-16-2012.xls.

Massachusetts Clean Energy Center. Commonweath Solar II. Retrieved on 5/07/2012. http://masscec.com/index.cfm/page/Commonwealth-Solar-II/cdid/11241/pid/11159.

NYC Solar Map. Retrieved on 3/5/2012. http://nycsolarmap.com.

Oregon Clean Energy Map. Retrieved on 3/5/2012. http://oregon.cleanenergymap.com.

Perez, R, Seals, R and Michalsky, J. 1993. All-weather model for sky luminance distribution—Preliminary configuration and validation. Solar Energy 50 (3): 235-45.

Reinhart, CF and Walkenhorst, O. 2001. Validation of dynamic RADIANCE-based daylight simulations for a test office with external blinds. Energy and Buildings 33 (7): 683-97.

Renew Boston Solar. Retrieved on 3/5/2012. http://gis.cityofboston.gov/SolarBoston/.

Renewable Energy Property Tax Exemption. Massachusetts General Law. Ch. 59 §5 (45, 45A). 1975.

Residential Renewable Energy Income Tax Credit. Massachusetts General Law. Ch. 62. §6(d). 1979.

Rich, PM, Dubayah, R, Hetrick, WA and Saving, SC. 1994. Using viewshed models to calculate intercepted solar radiation: applications in ecology. American Society for Photogrammetry and Remote Sensing Technical Papers. 524-9.

Robinson, D and Stone, A. 2004. Irradiation modelling made simple: the cumulative sky approach and its applications. Plea 2004 – The 21st Conference on Passive and Low Energy Architecture, Eindhoven.

Salt Lake City Solar. Retrieved on 3/5/2012. http://slcgovsolar.com.

San Diego Solar Map. Retrieved on 3/5/2012. http://sd.solarmap.org/solar/index.php.

San Francisco Solar Map. Retrieved on 3/5/2012. http://sf.solarmap.org.

Schwarz, C and Wade, E. 1990. The North American datum of 1983: Project methodology and execution. Journal of Geodesy 64 (1): 28-62.

Sunpower E18/230 Solar Panel Technical Sheet. Retrieved on 3/1/2012. Link.

United States Department of Energy. EnergyPlus Weather Data. Retrieved on 2/1/2012. http://apps1.eere.energy.gov/buildings/energyplus/cfm/weather_data.cfm.