Space-borne radar and flood hydrology: flood of data · Space-borne radar and flood hydrology: flood of data Guy Schumann Hydrology Group, School of Geographical Sciences, University

Space-borne radar and flood hydrology: flood of data

Guy Schumann Hydrology Group, School of Geographical Sciences, University of Bristol

[email protected]

+44(0)117 928 8478

Why space-borne radar?• Aerial photographs: fly-over requires extensive

planning and is expensive

• Optical space-borne sensors: on very rare

occasions only when there is prolonged flooding,

most of the time flooding has receded before weather

conditions improve

• Field data: point measurements, difficult to collect

Flood monitoring with radar satellites• Independent of weather and day/night capability

• Spatially (and sometimes also temporally) distributed

• However, many uncertainties associated with data; these need to be accounted for

Rarely this high a spatial

resolution; we will see more of

this in the near future!

Rarely this clear!

Mapping flood extent

• Traditional wet/dry map derived

using various image processing

algorithms vs. observed flood

possibilities based on two

coinciding SAR images and a

multitude of selected image

processing techniques for a

prolonged December 2006 flood

on the River Dee (NE Wales, UK)

To be published in IEEE TGRS & JoH

Improving flood extent mapping with LiDAR

LiDAR

Flood edge based on SAR data only (a) and on SAR data while conditioned on LiDAR (b)

Mason et al. (2007)

Indirect measurements of water levels

• Fusion of flood extent with topographic data allows water levels to be derived at the

flood edges

• Including flood mapping uncertainties leads to variation in water levels at a given

locationFlood depth and extent map conditioned on multiple

cross-sectional water stages derived from a SAR

flood extent of an event that occurred on the Alzette

River (G.D. of Luxembourg) in early January 2003

Indirect measurements of water levels:

application

Approximation of a flood wave from space• Polynomial approximation to account for topographic & extent mapping uncertainties

reproduces a reliable water surface profile in near real-time

• River Po (Italy) flood 2008 example

Approximation of a flood wave from space:

benchmarking models

Tewkesbury project

Map showing heaviest rainfall (% > average) Source: Met Office

12 remotely sensed images over a single event hydrograph! This is a very unique opportunity to study

flood dynamics and urban flood modelling support based on remote sensing

• Data collection

Also:

• Two over-flights by MOD (25 & 26 July 2007)

• Wrack marks collected by EA & Infoterra UK (Jan 21 2008)

Urban flood mapping with TerraSAR-X

• Theoretically possible for the first time!

• David Mason has demonstrated that it really works and developed an

algorithm for urban flood mapping (Tewkesbury, July 2007 flood)

• Use this for urban flood model assessment and improvement of flood risk

estimates

Mason et al., 2007, DLR Proceedings

• Mapping flood dynamics

• Including flood edge positioning uncertainty allows distribution of likely

water levels and statistics to be derived

• Adequate data quality assessment

Input image

Pre-processing & geo-location assessment

Flood edge extraction & uncertainty

Selecting spot locations for maximum

extent water level extraction

Eliminating erroneous locations based on: (1) LiDAR vegetation

& building buffer; (2) topography, i.e.

skewness/normality test; (3) outlier test if needed

Filtered good

quality water level

data

Amanda’s PhD

http://bprc.osu.edu/water/index.php

In support of the upcoming SWOT mission