Lake water extent variations and their response to a changing climate across mainland Southeast Asia | |
| Author | Tatsaneewan Phoesri |
| Call Number | AIT Diss. no.RS-25-04 |
| Subject(s) | Lake hydrology--Monitoring--Southeast Asia Water level--Data processing--Southeast Asia Climatic changes--Environmental aspects--Southeast Asia |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Remote Sensing and Geographic Information Systems |
| Publisher | Asian Institute of Technology |
| Abstract | Lake Water Extent (LWE) is a arucial indicator of hydrological and climatic changes, particularly across climate-vulnerableregions of Soutbesst Asia (SEA). This study ams to assess and analyze the spatial-temporal variation of bistorical and future LWE in response to climate dri vers acroSs SEA. A long-term LWE dataset was constructed to capture historical surface water variations 8Coss lakes in SEA from 1984 to 2019. This study assessed the spatial-temporal variation of LWE using monthly Global Surface Water (GSW) dataset. A semi-automated workflow was applied to generate three versions of the LWE dataset: a discontinuous monthly dataset (SEA-LWEun), a gap-filled continwous dataset (SEA-LWEgf), and a smooth, continuous dataset (SEA-LWEsm). These datasets enabled detailed analyses of LWE changes at monthly, seasonal, and anmual scales. Results revealed a net increase in sur face water ares, with 192 lakes exhibiting significant gains and 127 lakes showing long-term reductions. To simulate and predict LWE dynamics, data-driven models were developed using machine learning (ML) algorithms, including Extreme Gradient Boosting (XGB) Random Forest (RF), Support Vector Machine (SVM), and Artificial Neural Network (ANN). The models were trained in statistically selected climatic and runoff variables. The ensemble model was selected based on highest predictive performance for historical LWE reconstruction. The best-performing model of XGB was used to predict future LWE from 2015 to 2100 under SSP 1-2.6,SSP2-4.5 and SSP5-8.5, which bias-corrected CMIP6 climate projections. The results revealed spatially diverse trends in lake dynamics, with precipitation and runoff emerging as dominant predictors under future climate scenarios. This study contributes a robust, GIS-ready LWE dataset and a predictive modeling framework for assessing LWE responses to climate change in SEA, supporting long-term water resource ecosystems. |
| Year | 2025 |
| Type | Dissertation |
| School | School of Engineering and Technology |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Remote Sensing and Geographic Information Systems (RS) |
| Chairperson(s) | Virdis, Salvatore G.P.; |
| Examination Committee(s) | Tripathi, Nitin Kumar;Shrestha, Sangam; |
| Scholarship Donor(s) | Royal Thai Government;AIT Fellowship; |
| Degree | Thesis (Ph.D.) - Asian Institute of Technology, 2025 |