Flood Damage Detection Algorithm Using Sentinel-2 Images (Case study; Golestan Province, March 2019)

Remote sensing data can be used with reasonable accuracy today for areas without statistics to investigate natural events. One of these sudden and damaging phenomena is flooding. April flood in Golestan province has caused many damages in various areas of housing, agriculture, forestry and so on. In the present study, using satellite data and presenting flood damage detection algorithm, it is investigated. The highest difference in the NDVI index derived from Sentinel-2 data was calculated in the study period of March 2019 and late April 2019 in the study area. These results show that some areas have been severely damaged by floods, in which severity and duration of rainfall have had a significant role. These areas, which are estimated at an area of 70908 hectares, include agricultural, forest, residential, etc. Among the advantages of the proposed algorithm are the ease and speed of calculation, the capability to apply in urban areas, the high accuracy with reduced impact of other phenomena. The overall accuracy of the present algorithm is estimated to be 93.5% using ground control points. The accuracy of the results shows that the flood damage algorithm has a good accuracy to distinguish the areas under flood damage from the safe areas of danger.