I’ve been wanting to look into the feasibility of modelling the recent devastating floods in Malawi using just satellite captured data. So, whenever I’ve had a spare few minutes, I’ve been researching the region, trying to make connections and scraping together what data I can.

The first step is to build a Digital Elevation Model (DEM). This is a file which divides the land into squares, each of equal size, and gives a value of the average height in that square. The Shuttle Radar Topography Mission (SRTM) provides a good source of this data, and the latest product is at approximately 30 m resolution – so each one of those squares is 30 m by 30 m wide. This sounds pretty coarse and not very detailed, but for the area of interest I have chosen, the River Shire south of Lake Malombe, the grid contains more than 83 million squares!

This is far too many for a model to run in any feasible timescale, so I’ve had to merge these squares so they are 250 m x 250 m wide. This is still over 5 million squares, more than we would usually run in the model, but it loads and, with a test rainfall file, it runs.

Snap shot of the River Shire region of Malawi, running in CAESAR-Lisflood, built from SRTM elevation data

Satellite imagery of some of the recent flooding in Malawi 

The next stage will be to process this a bit more to allow water to flow more freely across the DEM. I will probably need to look at a smaller model domain, with a more detailed DEM, using data from the larger model to run it with, in order to capture the flood extents in a level of detail enough to compare with the sateliite data. More importantly, I need to get some rainfall data to use to drive it with!