SnowView: A Satellite Data and Model Driven Decision Support Tool for Water Resource Management

Snowmelt from mountain forests is critically important for water resources and hydropower generation because streams and rivers often originate from such environments. It is therefore important to provide water managers with high quality, real-time, spatially distributed estimates of snow in these areas to support their operational decision-making.

In this research, we present a novel web-enabled snow modeling and visualization system called “SnowView”, which was built for the Salt River Project (SRP), a provider of water supply and hydropower to millions of customers in Phoenix, AZ. At its core is a snow modelling system based on a robust method of interpolation of Snow Water Equivalent (SWE) and snow depth data from NRCS SNOTEL and NWS COOP stations along with machine learning of physiographic attributes to generate high-spatial-resolution daily estimates of SWE and snow depth spanning > 35 years (1982-present). This gridded dataset significantly improves SRP seasonal streamflow forecasts over the prior methodology of only using snow data from SNOTEL stations. Maps of key snow variables from this dataset, along with other snow products useful for snow monitoring, are delivered to SRP in near real-time through a web-based decision support tool. This tool is valuable for SRP operations because it allows for easy comparison of different gridded data sources, and it provides watershed analytics to compare the current year’s data with those for previous years. The ability to find similar water years in the historical record is important for water managers such as SRP, because it gives an idea of potential seasonal inflows to their managed reservoirs.

We are currently working to expand SnowView to cover more watersheds across the US. In addition, we are working with SRP water managers to improve SnowView’s functionality, and are advancing the accuracy of the underlying snow data with field data that we have been collecting along Arizona’s Mogollon Rim, at the headwaters of SRP’s managed watersheds. This spatially and temporally expansive dataset includes field surveys, automatic remote snow cameras, airborne lidar, and Structure from Motion (SfM) reconstruction of multi-angle aerial photography from unmanned aircraft, and has allowed us, in particular, to improve our representation of how forest canopy impacts snowpack in the region.