Authors: Adam Cossey, Dane Liljestrand, Ryan Johnson – University of Utah
Title: Optimizing Snow Monitoring in Complex Terrain with Low-Cost Sensor Systems and Machine Learning
Presentation Type: Poster
Abstract: Current NRCS SNOTEL in situ snow monitoring systems provide valuable information for hydrological forecasting but typically only provide snow information for a single point in a watershed. This data does not capture the snowpack distribution in complex terrain, specifically snow depth and snow-water-equivalent (SWE). Other snow monitoring methods include snow surveys and remote sensing products (e.g., ASO inc). However, these methods are temporally discontinuous and can be prohibitively expensive. Our project aims to use a probabilistic machine learning algorithm to enhance SNOTEL and snow surveys with low-cost, low-power snow monitoring systems spatially optimized within a watershed based on slope, vegetation, aspect, and elevation. We anticipate the combination of snow sensing location optimization and the expansion of temporally continuous snow observations to better characterize snow accumulating and melting throughout the water year. We are enthusiastic about sharing the progress on our project, including the key milestones, sensor tower construction, low-cost alternative sensors, deployment locations in the Parley’s Watershed area based on engagement with Salt Lake City Department of Public Utilities (i.e., end user), and highlighting the challenges during the deployment phase. We will also be sharing initial data