Tracking long-term wetland dynamics based on sample migration and two-stage hierarchical classification: A case study of Jiangsu Province

Wetlands play crucial ecological roles in carbon sequestration, flood control, and biodiversity preservation. The effective long-term monitoring of wetland dynamics is essential for sustainable management and conservation. However, obtaining wetland samples over multiple years and ensuring high classification accuracy pose unique challenges for precise extraction and spatiotemporal analyses. To address these issues, a sample migration algorithm was used to describe the interannual migration of wetland classification samples. The object-based hierarchical random forest (OBHRF) and object-based hierarchical decision tree (OBHDT) methods were integrated and combined with Landsat time-series data to map 11 wetland types and assess wetland dynamics in Jiangsu Province from 1985 to 2021. Change detection was used to investigate the land-use or landcover factors driving wetland dynamics, and a multivariate linear regression model was used to investigate the factors affecting interannual changes in wetland area. The optimal threshold for the cross-year migration of wetland samples occurred at a Euclidean distance ≤ 0.09, and a spectral angular distance ≥ 0.97. The overall accuracy of the wetland maps exceeded 79.36 %. Spatiotemporal dynamic analysis revealed a decline in the natural wetland area from 41.61 × 10³ km² to 39.12 × 10³ km², with over 68.5 % of the loss attributed to the expansion of aquaculture ponds and dry farmland from 1985 to 2021. Artificial wetlands increased from 31.71 × 10³ km² to 35.38 × 10³ km², with over 78.66 % of the gain attributed to the reduction in dry farmland. Policy-regulated economic development and population size were the main factors affecting changes in wetland areas. This study successfully mapped detailed long-term wetland maps and identified the drivers of wetland dynamics. The findings of this study provide an important reference for accurate wetland classification and contribute to the sustainable utilization and protection of wetlands in Jiangsu and across China.