Using Citizen Science Snow Depth Measurements in Snowpack Modeling

After snow depth measurements are recorded by CSO participants, we’ve developed a way to integrate those observations into the process of snowpack modeling. Snow models use data from weather stations and landscape characteristics to build a snowpack during the winter and melt it away when the weather gets warmer in spring and summer. But our ability to accurately predict snow depths is dependent upon accurate measurements of snowpack conditions as they change throughout the year. Since most mountainous areas are difficult to access on a daily or weekly basis, scientists are hoping that CSO participants will fill the data gap. Citizen science based measurements will allow us to monitor snowpack conditions more effectively, and possibly improve our snow modeling capacity.

Better results from snow models will inform multiple interest groups. Avalanche prediction relies on snow models to produce accurate snow depths, elevations, and aspects for wind loading. Flood prediction related to rain-on-snow events depend on snow models to accurately predict the density and water content of a snowpack. Water resource managers use on snow models to predict the amount of water that will end up in our reservoirs and river systems when the snow melts in springtime.

The initial results from CSO project study area in Alaska are promising. We chose an area near Valdez, AK along the Richardson Hwy that is often visited by winter recreationalists for snowmobiling and backcountry skiing. Last year we received hundreds of snow depth measurements from CSO participants within the study area at Thompson Pass. Here is a map of the Thompson Pass along with markers of the locations of snow observations.

We ran the model hundreds of times to make sure we have it “tuned in” to local environmental conditions. Then, we selected the best model run to integrate the measurements from CSO participants. The takeaway: the model results using citizen scientist observations of snow depth greatly improved the accuracy of the snow model when compared to a highly-trusted dataset called SNOTEL, which includes precise measurements of actual snow depths.

Going forward, we are developing methods to quickly and accurately integrate CSO participants’ snow depth observations into the modeling process and we hope to apply these methods to additional study sites in locations around the globe.