EGU25-7201, updated on 14 Mar 2025
https://doi.org/10.5194/egusphere-egu25-7201
EGU General Assembly 2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
Poster | Tuesday, 29 Apr, 14:00–15:45 (CEST), Display time Tuesday, 29 Apr, 08:30–18:00
 
vPoster spot A, vPA.14
Risk Mapping and Adaptation Strategies: Enhancing SWE Predictions with an LSTM Model for Snowmelt-Dependent Regions
Pei Ju Tsang and Wen Ping Tsai
Pei Ju Tsang and Wen Ping Tsai
  • National Cheng Kung University, National Cheng Kung University Hospital (Tainan City, Taiwan), Hydraulic & Ocean Engineering, Taiwan, Province of China (n86121145@gs.ncku.edu.tw)

Accurately predicting Snow Water Equivalent (SWE) has become increasingly crucial. It holds particular significance for managing water resources in regions heavily reliant on snowmelt. The present study introduces an integrated Long Short-Term Memory (LSTM) model that incorporates extreme heat events and diverse climate change projections to generate detailed SWE distribution maps and long-term trend analyses. By including lagged SWE observations and climate indicators, the model captures the intricate temporal dynamics of snowfall accumulation and melt processes, thereby improving forecast accuracy and stability.

Previous studies indicate that areas dependent on seasonal snowpack face accelerated snowmelt timing and reduced water availability under rising temperatures. These shifts can exert critical impacts on agricultural irrigation, ecosystem habitats, and water allocation strategies, highlighting the importance of robust forecasting tools for proactive resource management. Furthermore, the development of comprehensive risk maps pinpoints high-risk hotspots where anticipated temperature increases coincide with substantial changes in SWE and snowmelt patterns. These zones are prime candidates for early adaptation measures, including infrastructure upgrades and policy interventions aimed at mitigating potential water shortages.

As global warming persists, this modeling framework provides stakeholders, policymakers, and local communities with valuable insights into emerging water resource risks. The integration of climate change scenarios into the LSTM model underscores the necessity of forward-looking research that can inform both short-term operations and long-term planning. Ultimately, this approach lays the groundwork for crafting sustainable adaptation strategies, preserving agricultural output, protecting ecosystems, and ensuring water security in regions where snowmelt is pivotal to resource availability.

How to cite: Tsang, P. J. and Tsai, W. P.: Risk Mapping and Adaptation Strategies: Enhancing SWE Predictions with an LSTM Model for Snowmelt-Dependent Regions, EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-7201, https://doi.org/10.5194/egusphere-egu25-7201, 2025.