Identification of dominant species of submerged vegetation based on Sentinel-2 red-edge band: A case study of Lake Erhai, China

• Indices (SVSR and CI edge ) were developed to enhance spectral differences for identifying submerged vegetation in Erhai Lake. • A decision tree model was created to identify dominant submerged vegetation species in Erhai Lake. • Long-term monitoring of submerged vegetation in Erhai Lake has been conducted. • The impact of various water quality parameters on aquatic vegetation was analyzed. Aquatic vegetation plays a crucial role as a primary producer and ecological regulator, serving as an indispensable purifier in inland lakes. However, weakened optical information caused by water absorption and reflection presents challenges in remote sensing identification and classification of submerged vegetation (SAV). In this study, Sentinel-2 satellite data was employed to examine and compare the spectral attributes of three preeminent species, Potamogeton maackianus , Vallisneria natans , and Ceratophyllum demersum , within the confines of the Lake Erhai study area. Using the disparities in the red-edge band, red-edge chlorophyll index (CI edge ), and red-edge normalized vegetation index (NDVI edge ), a carefully designed decision tree model was instituted for remote sensing-driven identification of SAV. The ensuing species identification exhibited a commendable overall accuracy rate of 96.8%, underscored by a Kappa coefficient of 0.95. Geospatial distribution revealed the prevalence of P. maackianus in the northern lake bay, counterbalanced by the dominance of V. natans in the southern lake bay. Furthermore, an incisive inquiry into the nexus between diverse water quality variables and aquatic vegetation was conducted. Notably, water temperature (WT) emerged as a the most influential factor, exhibiting a highly significant positive correlation with SAV. The correlation coefficient reached 0.98, indicating a pronounced influence, with WT contributing substantively to the observed variance in SAV, accounting for 63.3%. The proposed method highlights the potential of Sentinel-2 red-edge bands for ecological monitoring and management of SAV, with implications for broader environmental applications.