MPhil : Improving flood extent mapping in urban areas
Graduate Position 2 years Leicester, UK
Uploaded 29 Aug 2020
Around 1 billion people globally are exposed to flooding. This can devastate homes and businesses and lead to significant loss of life. It is therefore critical that timely information on flooding events is made rapidly available enabling emergency responders to move or rescue vulnerable populations.
Satellite imagery provides an effective way of monitoring flood events by mapping the extent and evolution of flood waters. In particular, radar imagery which can penetrate cloud cover and can see at night is especially valuable in these scenarios. However, radar imagery is acquired obliquely and since the greatest risk to life generally occurs in urban areas, shadows from tall buildings, urban vegetation and underlying terrain may obscure the visibility of the flood waters. This often leads to incorrect classification of water in these areas.
Hence, there is an urgent need to improve flood mapping in urban environments and provide enhanced reliability in the flood extent retrieval algorithms.
This project seeks to enhance the existing flood extent retrieval approaches by including knowledge of the satellite viewing parameters along with knowledge of topography and lidar-derived building height into existing flood mapping retrieval algorithms to significantly enhance the accuracy of urban flood maps. By modelling the relationship between the satellite viewing parameters and the structure and layout of the urban environment, areas of low confidence can be identified and removed from the prediction.
The project also provides an opportunity to explore and characterise how the algorithm is affected by the height, separation, and structure of local buildings, under different topographic scenarios and in urban areas in different countries with significantly different structural forms.
The student will work closely with flood modelling specialists at Previsico Ltd, to support improvement of existing flood mapping approaches, leading to enhanced validation of their flood models.
Applicants will be expected to have a good numerate first degree with excellent programming skills (ideally Python but R might be considered). Some experience with two or more of the following is required: image processing, remote sensing, geographic information systems (GIS), radar. We are looking for someone with excellent team working skills who is happy to collaborate with partners in industry as well as academia.