Geoforum 2014

Using EDINA datasets in a hydrology project

About me

Background in environmental science

MSc in GIS, 2013

Now Spatial Data Analyst at GeoPlace (National Address Gazetteer)

About GeoPlace

Owned by Local Government Association / Ordnance Survey

Definitive national address & street gazetteers (datasets)

380 local authorities working together across GB – includes One Scotland Gazetteer (OSG) managed by ThinkWhere

AddressBase products used by emergency services, government agencies, insurance companies

AddressBase features

Point data for every address in Great Britain

Full postal address and many other attributes

Unique Property Reference Number (UPRN) for every property

Property lifecycle – includes historical & in planning

Detailed property classification

Cross references to other OS products e.g. MasterMap, ITN

Address data in Flood Management

Properties located in different flood zones Allows emergency services target resources in flood event

Importance of address data...

Ofwat: Thames Water misreported the number of properties at high risk of sewer flooding between 2005 and 2010

"This meant that more properties were recorded at higher risk of sewer flooding than there was evidence to support“

While this fine relates to sewers it is crucial that accurate property data is available for these types of analyses

Project overview

Hydrology example (mainly maps)

EDINA datasets (via. Digimap)o Land Cover Map 2007o Digital Terrain Model (DTM) – OS Profileo OS Strategi

Other datasets o Catchment shapefiles (Centre for Ecology & Hydrology / CEH)o River flows data (HiFlows / Environment Agency)

Project aims

Aim of project was to calculate a regression-based equation for PROPWET (soil moisture value)

Currently engineers use constant value for catchment PROPWET (Flood Estimation Handbook)

Test & develop a regression-based equation using GIS, following initial Halcrow analysis

Particular relevance to natural flood management

Natural Flood Management (NFM)

NFM involves mitigating flooding using nature, currently difficult to analyse

Increasing flooding (climate change) and increasing costs

Regression-based PROPWET allows simulations of different ground conditions e.g. increased vegetation

SEPA projects eg Eddleston Water, Scottish Borders

NFM example: Eddleston Water

69km2 waterway, part of the River Tweed Catchment

Straightened to improve land drainage leading to habitat loss and increased flood risk, 589 properties at risk of flooding (SEPA Flood Map)

Actions to date: Riparian planting, re-meandering, removal of manmade embankments

PROPWET (soil moisture)

Soil moisture important for flood estimates and the PROPortion of time soils are WET (PROPWET) is an important catchment descriptor

PROPWET range 0-1, e.g. wet catchment 0.8, dry catchment 0.1

PROPWET value currently taken from Flood Estimation Handbook (FEH) (Centre for Ecology & Hydrology, 1999) based on database query

FEH PROPWET is a single value for a catchment and based on when Soil Moisture Deficit (SMD) ≤6mm during long term average (1961-90)

FEH PROPWET doesn’t consider land use or climate factors

Monitoring stationsRiver network

Halcrow analysis

Initial regression-based equation proposed based on 43 catchments (Halcrow, 2012)

PROPWET = 0.5365Log10[SAAR – (PE x LAI4.70 x r 0.79 x 2.003E - 5)] - 0.998

Project workflow

Literature review

Obtain datasets

Generate river catchments

Zonal statistics, ArcGIS models to iterate – breakdown into different datasets required

Update regression equation

Land Cover Map 2007 (LCM2007)1km raster

Land Cover Map 2007 (LCM2007)

LCM2007 is a parcel-based classification of 23 land cover classes based on the UK’s terrestrial Broad Habitats

Classifying summer-winter composite images captured by satellite sensors with 20-30m pixels

LCM represents land cover type and not land use, although the two are often synonymous in practise

Land Cover Map 2007 (LCM2007)

Obtained vector & 1km raster (aggregated vector, dominant class per 1x1km pixel)

Vector dataset better metadata but long geo-processing times for country wide queries, vector manipulated using ogr2ogr

Vector would have been fine in my desktop GIS if had a smaller study area

Raster clipped against river catchments after resampling to 100x100m to obtain better fit, each catchment containing an average of 9 different land cover types

Catchment delineation

Possible to generate river catchments from terrain (DTM) and ESRI’s Hydrology toolset

1. Digital Terrain Model (DTM)2. Reclassify noData values3. Fill ‘holes’ in DTM4. Flow direction 5. Flow accumulation6. Stream network definition7. Catchment extraction

Catchment delineation - results

Geoprocessing

Used ArcGIS python model builder to iterate zonal statistics tool Standard Average Annual Rainfall (SAAR), Air Temperature (Ta), Extra-terrestrial

Radiation (Re) per catchment

Geoprocessing -> Zonal Stats Maps

Standard Average Annual Rainfall (SAAR)

Extra-terrestrial radiation (Re)Air temperature (Ta)

Potential Evapo-transpiration (PE)

Leaf Area Index (LAI), Surface resistance (Rsc)

Leaf Area Index: amount vegetation cover Surface resistance: friction, affects water flow Assigned classifications using research paper (Hough & Jones, 1997), applied Zonal

Statistics by table tool

Results

OS MasterMap Networks - Water Layer

New OS Water Layer (Beta) created with Environment Agency, SEPA, INSPIRE

Mainly derived from OSMM Topo but first time provides dedicated OS Water product with many new useful attributes for flood modelling / mapping

Features: WatercourseLink (inc Flow Direction), HydroNode, WatercourseLinkSet, WatercourseSeperatedCrossing, WatercourseInteraction

Thank you...

Any questions?

Extraterrestrial radiation (Re)