Detect changes in marsh plant communities based on Landsat long time
series data and BFAST model
Abstract
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 contributes to a more accurate understanding of
vegetation change in freshwater wetlands and provides strong support for
its sustainable conservation and management.