2026-11-03 –, Compagno
The SEA-LWE dataset provides 36-year monthly time series for 651 Southeast Asian lakes. Derived from GSW products via semi-automated mapping, it offers three versions: discontinuous (SEA-LWEun), gap-filled (SEA-LWEgf), and smoothed (SEA-LWEsm) hydrological records.
Lakes serve as critical sentinels of environmental health, reflecting complex interactions between climatic shifts and landscape change. In this session, we present a comprehensive study that quantifies Lake Surface Water Extent (LWE) changes across 651 lakes in mainland Southeast Asia a region uniquely vulnerable to the accelerating impacts of global climate change.
Leveraging a 36-year (1984–2019) time series derived from Global Surface Water (GSW) products, we developed a robust, reproducible workflow to delineate LWE dynamics. This talk will explore the transition from raw, discontinuous satellite observations to a refined, gap-filled, and smoothed hydrological dataset ready for large-scale analysis.
Technical Approach & Workflow:
The core of this project relied on a semi-automated processing pipeline designed to handle the challenges of long-term remote sensing data, specifically cloud cover and sensor discontinuities.
We developed three distinct versions of the LWE product to meet different analytical needs:
- SEA-LWEun: A raw, discontinuous monthly LWE dataset.
- SEA-LWEgf: A gap-filled, continuous monthly LWE dataset.
- SEA-LWEsm: A smoothed, continuous version optimized for trend analysis.
Our methodology emphasizes transparency and replicability, utilizing open-source geospatial frameworks to perform pixel-to-vector transformations and temporal interpolation. We will discuss the statistical validation process using RMSE, MAE, Bias, and NSE metrics to demonstrate how this approach improves upon existing global datasets by addressing regional data sparsity.
Key Findings:
Our analysis reveals a nuanced picture of hydrological change in Southeast Asia:
- Net Positive Growth: Despite regional variability, a net increase in total LWE was observed.
- Growth Trends: 192 lakes exhibited a significant average increase of 0.004 km²/year, resulting in a total gain of 24.9 km².
- Recession Trends: Conversely, 127 lakes showed a contraction, with an average decrease of −0.002 km²/year and a total loss of 10.8 km².
