We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Exploring the Potential of Long Short‐Term Memory Networks for Improving Understanding of Continental‐ and Regional‐Scale Snowpack Dynamics.
- Authors
Wang, Yuan‐Heng; Gupta, Hoshin V.; Zeng, Xubin; Niu, Guo‐Yue
- Abstract
Accurate estimation of the spatio‐temporal distribution of snow water equivalent is essential given its global importance for understanding climate dynamics and climate change, and as a source of fresh water. Here, we explore the potential of using the Long Short‐Term Memory (LSTM) network for continental and regional scale modeling of daily snow accumulation and melt dynamics at 4‐km pixel resolution across the conterminous US (CONUS). To reduce training costs (data are available for ∼0.31 million snowy pixels), we combine spatial sampling with stagewise model development, whereby the network is first pretrained across the entire CONUS and then subjected to regional fine‐tuning. Accordingly, model evaluation is focused on out‐of‐sample predictive performance across space (analogous to the prediction in ungauged basins problem). We find that, given identical inputs (precipitation, temperature, and elevation), a single CONUS‐wide LSTM provides significantly better spatio‐temporal generalization than a regionally calibrated version of the physical‐conceptual temperature‐index‐based SNOW17 model. Adding more meteorological information (dew point temperature, vapor pressure deficit, longwave radiation, and shortwave radiation) further improves model performance, while rendering redundant the local information provided by elevation. Overall, the LSTM exhibits better transferability than SNOW17 to locations that were not included in the training data set, reinforcing the advantages of structure learning over parameter learning. Our results suggest that an LSTM‐based approach could be used to develop continental/global‐scale systems for modeling snow dynamics. Plain Language Summary: Understanding the spatio‐temporal distribution of water in the snowpack (known as snow water equivalent) is very important for understanding climate dynamics and climate change, and for forecasting and management of global water supplies. In this study, we use Deep Learning (DL) to model snow accumulation and melt at 4‐km pixel‐scale resolution across the conterminous US (CONUS). Long Short‐Term Memory (LSTM) networks are developed at both continental‐ and regional‐scale, by combining spatial pixel sampling and stagewise network pre‐training/fine‐tuning. We benchmark out‐of‐sample predictive performance against the physical‐conceptual temperature‐index‐based SNOW17 model, and find that LSTM networks significantly outperform calibrated versions of the SNOW17 model when given identical information. Further, when provided with additional meteorological information, performance of the LSTM is improved. The LSTM models also exhibits better transferability than the SNOW17, indicating the potential for future development of a DL‐based system for modeling continental/global‐scale snow dynamics. Key Points: A trained continental‐scale Long Short‐Term Memory (LSTM) network is capable of providing almost as good performance as a regionally trained oneThe continent‐scale LSTM outperforms a regionally trained SNOW17, and a SNOW17 model calibrated locally to each pixel across the domainThe LSTM exhibits better spatial transferability than SNOW17, and exhibits a trade‐off between transferability and model complexity
- Subjects
SNOW accumulation; WATER supply management; SNOWMELT; DEW point; WATER distribution; VAPOR pressure
- Publication
Water Resources Research, 2022, Vol 58, Issue 3, p1
- ISSN
0043-1397
- Publication type
Article
- DOI
10.1029/2021WR031033