Challenge 2: 3D mapping for inundation
Challenger: 510 Red Cross


Amongst weather related disasters, floods are the most recurrent, and have considerable negative impact on communities. Local communities in flood prone areas are working with Red Cross & Crescent National Societies to better prepare and act on predicted floods. 

510 is an initiative of the Netherland Red Cross. Our purpose is to improve speed, quality & cost-effectiveness of humanitarian aid by using & creating data & digital products. 510 helps by  predicting flood extents as early and as accurately as possible so the communities and National societies can act in preparedness as early and as efficiently as possible. 

Not all flood types are the same and can be triggered by different phenomena, or found in different settings such as along a river, a lake or coastal areas. However, the impact a flood will have is closely related to the topography of the area, and more precisely, to the elevation above water bodies (sea, rivers or lake). It is therefore a priority for humanitarian actors to get access to highly accurate elevation data, in the form of a digital elevation model (DEM) representing the landscape as a 3D surface of which there are different resolutions. With high resolution DEM, 510 would be able to create more accurate flood risk maps, and calculate more precisely how communities will be affected based on their elevation/position.


The openly available global DEM dataset currently used for flood inundation mappings have a very low resolution and low vertical accuracy, for example  SRTM or ASTER DEM datasets expected vertical accuracy is about ±16 m absolute and ±6 m relative (Elkhrachy,2018), both datasets have horizontal resolution of 30m X 30m. There is a need for an algorithm based on high resolution satellite imagery that can refine the current topography (30m SRTM data). 

Several tech initiatives are jointly working on solving the digital elevation data accuracy problem As an example, Google on AI For Social Good, part of their work on flood forecasting, is able to generate a 1m DEM based on satellite images and satellite camera model corrections

Figure 1. A 30m SRTM based DEM of the Yamuna river compared to a Google generated 1m DEM of the same area. Source: Google AI Blog

Nowadays there is a high availability of high resolution satellite imagery. This optical imagery can be used to develop the high resolution DEM, which will result in a higher granularity of the model that considerably influences and improves the results of the flood impact prediction.


The Red Cross is involved in flood relief activities in the coastal areas of Dar es Salaam /Tanzania. With sea levels on the rise, and heavier storm occurrences, more and more the land-water interface will be in closer contact. Creating an algorithm that improves the DEM, of these sites, proves interesting for any type of disaster, whether it is coastal floods, or flood due storms and across any site (with its appropriate calibrations). Further, the selected site has a high accuracy topographical field measurement that serves to validate the refined DEM model.


Can you come up with a model that, with the use of high resolution satellite imagery, could refine the current topography 30m SRTM data, so that the Red Cross 510 team can improve their flood impact/intervention maps. Additionally, can you build a web platform where both DEMs are displayed to the public and where people can compare the flood map results derived from each one?


Copernicus Data.

Rules of Participation