Assessing spectral indices for detecting vegetative overgrowth of reservoirs

Abstract

The main problem related to exploitation of reservoirs is the overgrowth of aquatic vegetation, which leads to the gradual disappearance of water bodies. Currently, satellite imagery data are an advantageous source for monitoring aquatic vegetation. The main goal of this study was to assess different spectral indices (ARVI, NDVI, NDCI, NDAVI, WAVI) for detection of the overgrowing process in reservoirs. Three reservoirs located in the western part of Poland were selected for analysis: Przebędowo, Jeżewo and Jezioro Kowalskie. The analysis was carried out on the basis of Sentinel-2 satellite imagery. All calculations were performed in ArcGIS 10.5 and Quantum GIS software. Results obtained for each of the spectral indices were verified on the basis of high-resolution color orthophotomaps. The results show that selected indices detect different overgrowth areas. The WAVI index shows the strongest agreement with reference data. The agreement between reference data for each pixel was calculated on the basis of the Kappa coefficient. Regardless of analyzed reservoir, the WAVI index has the highest value of the Kappa coefficient. Additionally, the analysis suggests that Sentinel-2 data can be used to identify emergent plant areas for reservoirs characterized by elongation ratio, width and inundation area. The highest uncertainty of results is shown by emergent plants characterized by small, dispersed areas and located near the banks of the water bodies.