Skip to main content

Miles Clement

Flood extent and volume from radar.

Project title

Flood extent and volume from multi-temporal satellite synthetic aperture radar using change detection and terrain model fusion


Areas of research

  1. NEOLab
  2. Geodesy

Project description

Earth observation data enables monitoring of the environment across wide geographical regions. Recently launched satellites provide imagery at an improved spatial and temporal resolution. These include images from the ESA Copernicus Sentinel programme. Of particular interest is Synthetic Aperture Radar (SAR). SAR has the ability to collect imagery independent of cloud cover.

We will develop a fully automated workflow for estimating flood extent, water depths and volume changes from SAR imagery.

We will determine flood extent. We will do this by using change detection, variable radiometric thresholding and region growing of SAR images. We will combine derived areas of inundation with high-resolution terrain models. This will enable us to estimate shoreline heights. From this, we will create a water surface using a grid-based approach. We can then calculate water depths and volumes, allowing for improved insight into the dynamics of a flood event.

The method has been successfully applied to the 2015-16 winter flood event across various catchments in the north of England. Global applicability has been showcased with a further case study in Australia. The increased temporal resolution of the satellite data has allowed for monitoring of the onset and retreat of flood waters over an extended time. The methods developed are globally applicable and fully automated. They highlight the potential for near real-time flood monitoring.

The Natural Environment Resources Council (NERC) Centre for Doctoral Training (CDT) in Data, Risk and Environmental Analytical Methods (DREAM) is funding the project.