Item: SWE ESTIMATION FOR HYDROLOGIC APPLICATIONS: COMBINING WEATHER RADAR AND MACHINE LEARNING
-
-
Title: SWE ESTIMATION FOR HYDROLOGIC APPLICATIONS: COMBINING WEATHER RADAR AND MACHINE LEARNING
Proceedings: International Snow Science Workshop 2024, Tromsø, Norway
Authors:
- Baxter E. Vieux [ Applied Research Team, Inc. ]
- Jean E. Vieux [ Applied Research Team, Inc. ]
- William E. Vieux [ Elevated Path, LLC ]
Date: 2024-09-23
Abstract: Snow water equivalent (SWE) estimation is critical for water resource management and forecasting in snow-dominated regions. Current SWE measurement technologies are limited in providing temporally and spatially continuous data across large watersheds. This study analyzes radar data from the 2022, 2023, and 2024 snow years, covering 1986 sq km in the Upper Conejos River basin, the headwaters of the Rio Grande River in Colorado’s Southern Rocky Mountains, USA. This research presents a novel approach integrating Airborne Snow Observatories, Inc. LiDAR data (ASO) with ground-based dual-polarization weather radar and machine learning techniques to enhance SWE estimation. SnowQ™ is a patent pending technology for estimating SWE contained in snowpacks based on radar data. The method is tailored for SWE estimation in areas with sparse direct measurements at watershed scales relevant to water resources management. We apply a Forward Generation approach, using parameters calibrated from one snow season to predict SWE in the other snow seasons. Results of 9 test cases show strong agreement between radar-based estimates and ASO measurements, with R² values consistently above 0.85 across multiple years. These results demonstrate the temporal transferability of calibrated parameters and the method's effectiveness at the sub-watershed level. The approach has significant implications for improving hydrological forecasting and water resource management in snow-dominated watersheds, offering the potential for more available, spatially comprehensive, and timely estimation of SWE across different snow seasons. The ability of weather radar to measure SWE directly makes it an important addition to the suite of current snow measurement tools, complementing existing technologies.
Object ID: ISSW2024_P9.12.pdf
Language of Article: English
Presenter(s): Baxter E. Vieux
Keywords: Dual Polarization, Weather Radar, LiDAR, Snow Water Equivalent, SWE, Airborne Snow Observatories, ASO, Machine Learning
Page Number(s): 1314 - 1319
-