Methodology for Mapping Flood Extent on Estonian Floodplains

Estonia is known for its large floodplains that are annually covered by water. A methodology based on Sentinel-1 and -2 imagery was developed for operational monitoring of floodplains in Estonia. Filed campaign for validation/calibration of flood extent estimation methodology was performed using UAV in March 2019 over the Soomaa floodplain. Sensitivity analysis of Sentine1-1 data indicated greatest backscattering ( σ
⁰
) difference between land and water for polarizations VH (IW mode) and HV (EW mode) at incidence angle 39-47°. For water mapping, incidence angel normalization was applied using linear relationship between backscattering ( σ
⁰
) and local incidence angle established from our dataset. Water was mapped from Sentinel-2 data using MNDWI index. Flooded area mapped from Sentinel-l was compared with corresponding flooded area mapped from Sentinel-2. The kappa coefficient between these maps was 0.86 indicating good coincidence. Established methodology was tested to map floods at Keila test site for springs in 2018 and 2019. Difference in flood extent and duration at Keila test site in 2018 and 2019 was 6.8 km
₂
and 9 days.