Deep-Learning-Based Snowpack Mapping and Forecasting with Ground Observations: A Case Study Using A Wireless-Sensor Network in the American Basin (Extended Abstract)
Title | Deep-Learning-Based Snowpack Mapping and Forecasting with Ground Observations: A Case Study Using A Wireless-Sensor Network in the American Basin (Extended Abstract) |
Publication Type | Conference Proceedings |
Year of Conference | 2021 |
Authors | Cui, Guotao, and Bales Roger |
Conference Name | 88th Annual Western Snow Conference |
Conference Location | Bozeman, MT |
Keywords | bias correction, deep learning, Snow mapping, Snow Water Equivalent (SWE), wireless-sensor network |
Abstract | Mountain snowpack in the Sierra Nevada is a major source of California’s water supply. Basin-scale snowpack is crucial hydrologic information for water-resources decision making (Bales et al., 2006), e.g. reservoir operation and flood control. In particular, during precipitation events associated with atmospheric rivers, rain-on-snow-melted snow can significantly augment basin runoff (Henn et al., 2020), highlighting the importance of near-real-time estimates and forecasts of snowpack for decision-making support. This study presents an approach based on a deep-learning Long Short-Term Memory (LSTM) model and a bias-correction method using ground snow measurements for snow mapping and forecasting. The approach is evaluated in the upper American River basin (elevation ≥1500 m, Figure 1). |
URL | /files/PDFs/2021Cui.pdf |