Day 4

Detailed paper information

Back to list

Paper title Flood Depth Estimation and Validation using the SAR derived Water Mask and the HAND Model
Authors
  1. MinJeong Jo UMBC/NASA Goddard Space Flight Center Speaker
  2. Batuhan Osmanoglu NASA Goddard Space Flight Center
  3. Franz J. Meyer University of Alaska Fairbanks
  4. Elodie Macorps NASA Goddard Space Flight Center
  5. Rustem A. Albayrak NASA Goddard Space Flight Center and University of Maryland
  6. Joseph H. Kennedy Alaska Satellite Facility
Form of presentation Poster
Topics
  • D1. Managing Risks
    • D1.01 Satellite EO for Geohazard Risks
Abstract text Floods are one of the most common disasters that can be triggered by hydro-meteorological hazards such as hurricanes, heavy rainfalls, rapid snowmelt, etc. With the recent proliferation of synthetic aperture radar (SAR) imagery for flood mapping due to its all-weather imaging capability, the opportunities to detect flood extents are growing compared to using only optical imagery. While flood extent mapping algorithms can be considered mature, flood depth mapping is still an active area of research even though water depth estimation is essential to assess the damage caused by a flood and its impact to infrastructure. In this regard, we have been working on development and validation of flood depths as a part of the HydroSAR project led by the University of Alaska Fairbanks. HydroSAR is a cloud-based SAR data processing service for rapid response and mapping of hydrological disasters. The water depth product at 30-meter resolution named WD30 is in preparation to be automatically generated leveraging Hybrid Pluggable Processing Pipeline (HyP3). To achieve that goal it has been validating for topographically different test sites.

To estimate water depth of a flooded area, the method utilizes a Height Above the Nearest Drainage (HAND) model and water masks generated from SAR amplitude imagery via the HydroSAR HYDRO30 algorithm. HAND is a terrain descriptor which is computed from a hydrologically coherent digital elevation model (DEM). The value of each pixel on a HAND layer represents the vertical distance between a location and its nearest drainage point. As the quality of a HAND model has a decisive impact on the calculated water depth estimates, a terrain model of high spatial resolution and accuracy is highly required to generate a reliable water depth map. In this study, to generate HAND we used the Copernicus GLO-30 DEM, which is a 30-meter global DEM released by the European Space Agency (ESA). The water height is adaptively calculated for each water basin by finding the best matching water extent given the water height and HAND.

In our presentation, we will show case studies of water depth mapping over Bangladesh related to a flooding event in 2020. SAR-based water masks from Sentinel-1 SAR amplitude imagery were generated through the HydroSAR implementation to be used as input. We generated WD30 products for the flood season in July using two different data sets. The accuracy of obtained water depth estimates were assessed by comparison with water level data from the Flood Forecasting and Warning Center-Bangladesh Water Development Board (BWDB). Before the statistical analysis of the comparison, adjustments for the different datums between WD30 estimates and reference water level have been carried out. R2 values for all dates from both data sets showed close to or larger than 0.8 with RMSE values of less than 2 m, which confirmed the flood depth estimates of WD30 were at the expected quality given the vertical accuracy of the input DEM.