Advancing the Classification and Attribution Method for Alpine Wetlands: A Case Study of the Source Region of Three Rivers, Tibetan Plateau

Alpine wetlands are highly vulnerable to changes caused by global warming. Rapidly and accurately mapping alpine wetlands and analyzing the driving factors of their spatiotemporal changes are crucial for protecting and managing these resources. However, few studies have investigated classification methods and attribution analyses for alpine wetlands. To address this gap, a novel classification method has been developed, integrating the Google Earth Engine, alpine wetland features, and a random forest classifier, named GAWRF, to delineate wetlands in alpine regions. Additionally, an improved Partial Least Squares Structural Equation Model (PLS-SEM) was utilized to explore the mechanisms of spatiotemporal changes in wetlands of the Source Region of Three Rivers (SRTR) from 1990 to 2020. The results indicate (1) the high accuracy of the SRTR land cover maps from 1990 to 2020, with an overall accuracy of above 92.48% and a Kappa coefficient of over 0.91, satisfying the subsequent analysis of wetland spatiotemporal changes; (2) a net loss of 3.8% in the SRTR alpine wetlands, with a notable 7.9% net loss in marsh wetlands and nearly 32,010 km2 lost by 2015; and (3) topography and permafrost change as key drivers (as identified by the PLS-SEM), with permafrost contributing 52% to the significant marsh wetland loss from 2010 to 2015. This study aims to provide fundamental information that is essential for the monitoring and conservation of alpine wetlands.