Detect Changes in Marsh Plant Communities Based on Landsat Long Time Series Data and BFAST Model

Due to the combined effects of human activities and climate change, freshwater wetlands, especially in agricultural watersheds, face severe degradation threats. Therefore, it is necessary to explore in depth the changes in plant communities within these wetlands. This study investigates changes in wetland plant communities within these watersheds and assesses the feasibility of the Breaks for Additive Season and Trend (BFAST) method for detecting abrupt shifts in vegetation over long time series. Using long‐term Landsat imagery (1984–2016), annual maximum NDVI values were calculated for the Naolihe Basin Nature Reserve in Northeast China. The BFAST algorithm was then applied to detect NDVI changes in wetland plant communities, with results validated through field surveys. The results revealed four distinct NDVI change trends: no significant change, high‐to‐low shift, low‐to‐high shift, and continuous decline. NDVI deviations ranged from −0.85 to 0.94, with 1 to 5 abrupt changes mainly occurring between 1993 and 2006. The study confirms BFAST's effectiveness in detecting changes in wetland plant communities and, combined with field data, proposes a conceptual model to explain the degradation processes in freshwater wetlands. The model reveals the degradation process of different vegetation types under the influence of water competition and other factors, which contribute to a clearer understanding of vegetation change in freshwater wetlands and provide strong support for its sustainable conservation and management.