Anzali Wetland Surface Area Evaluation Based on Landsat Time Series Data and NDWI Indices

Wetland habitats are one of the most important natural ecosystems in the world. Evaluating and managing these valuable ecosystems require accurate and up-to-date data that remote sensing makes it possible. In this study, the changes of Anzali International Wetland in Gilan province, Iran, were investigated using Landsat satellite images and the Modified Nomalized Diffrence water index (MNDWI) in Google Earth Engine (GEE) platform during the years 1986 to 2020. To monitor waterbodies changes, two classes of water and non-water area were classifeied by Support Vector Machine (SVM) algorythm and MNDWI index was used to distinct the water surface areas. On the other hand, climate data including TRMM satellite data and PDSI index from TerraClimate data and Caspian Sea water level data were used to determine their effects on water level fluctuation of the wetland. The maps of SVM classification had overall accuracy more than 87% and Kappa coefficient was more than 88%. The wetland water body loss has decreased by 20% in its area according to MNDWI index maps, it has reached from 5926 hectares to 954 hectares, so that initially (until 2000) there was an upward trend and then a downward trend in the wetland water level. Also, the water level of Anzali wetland have been affected more by the sea level than the climatic factors. The results show that water indices and Google Earth Engine are efficient tools to identify the trends of water level changes of wetlands, and could provide more detailed scientific guidance to protect and manage natural resources in the studied areas.