Forecasting Short-Term Changes in Snowmelt due to Dust Impacts on Snow Albedo

Deposition of dust on Colorado snowpack has been shown to accelerate snowmelt due to reductions in snow albedo. Since melt timing directly affects downstream water availability, the ability to forecast short-term changes in melt rate can improve water management decisions. This project combines four sources of variable inputs in a system of energy balance calculations to forecast short-term changes in snowmelt rate for a small study catchment in the Southern Rocky Mountains. Two sources are historical (2003-present) datasets from Center for Snow and Avalanche Studies (CSAS). CSAS collects hourly meteorological data (including P, T, RH, wind, and radiation balance) from two automated sensor towers within a small mountain catchment. CSAS also collects roughly biweekly manual assessments of snowpack characteristics and dust layer depth at a network of mountain sites. The third source of variables is the NOAA Automated Surface Observation System (NOAA ASOS) historical (2006-present) dataset of automated hourly meteorological data. The fourth source of inputs are NOAA NWS weather forecasts, archived daily for the 2017-2018 snow season. Using the three historical datasets, we calculated the energy balance for the snowpack at the two towers within the CSAS study catchment, and evaluated the spatial variability of snow albedo due to the presence of dust between the two sites. We combined CSAS manual assessments of dust within the snow profile with snow albedo calculated from automated radiation data to determine rates of dust emergence with given meteorological forcings. To use this simple physical model to forecast potential changes in melt rate, we determined possible energy balance inputs from the NOAA NWS weather forecasts and applied those bounds to the measured snowpack. This forecast method was calibrated at one CSAS tower site and evaluated at the other site for the 2017-2018 snow season. We will evaluate the forecast method in real-time during the 2018-2019 snow season. Since this method uses manual assessments of dust layer location and rate of emergence within the entire snowpack, it does not have the remote-sensing limitation of only capturing snow surface characteristics. These improvements to short-term melt rate forecasts can better streamflow forecasting during snowmelt.