Integrating Remote Sensing and Machine Learning for High-Resolution Vulnerability Mapping of Seasonal Wetland Shrinkage under Climate Change in Southern Malawi

Malawi's Elephant Marsh, a RAMSAR site, is under intense pressure from climate change and human activities. This study utilized remote sensing and machine learning to analyze land use and land cover changes in the Elephant Marsh from 2009 to 2024, revealing a dynamic and surprising ecological trajectory. Contrary to dominant narratives of continuous degradation, our analysis shows two distinct phases. Phase 1 (2009–2018) saw clear environmental decline: water depletion, bareland expansion, and vegetation loss, largely due to unsustainable practices and drought. Yet, Phase 2 (2018–2024) demonstrated a dramatic recovery, characterized by robust water regeneration and vegetation increase, primarily driven by significant cyclonic events from 2021 onwards. This research fundamentally challenges the idea of linear wetland degradation, highlighting how high-magnitude, episodic weather events can temporarily override persistent anthropogenic pressures and catalyze rapid ecosystem recovery. While this short-term recovery is notable, it also contrasts with regional patterns of long-term wetland shrinkage, suggesting a nuanced relationship between immediate climatic impacts and broader environmental trends. Our findings advocate for adaptive environmental management that recognizes the episodic climatic events’ due role in both degrading and restoring these vital ecosystems.