Estimating relationships between snow water equivalent, snow covered area, and topography to extend the Airborne Snow Observatory dataset

Abstract. A new spatio-temporal dataset from the ongoing Airborne Snow Observatory (ASO) provides an unprecedented look at the spatial and temporal patterns of snow water equivalent (SWE) over multiple years in the Tuolumne Basin, California, USA. We found that fractional snow covered area (fSCA) significantly improves our ability to model the distribution of SWE based on relationships between SWE, fSCA, and topography. Further, the broad availability of satellite images of fSCA facilitates the transfer of these relationship to different years with minimal degradation in performance (r2 = 0.85, % MAE = 33 %, % Bias = 1 %) compared with models fit on the same day, by considering variations in SWE depth as expressed by differences in fSCA between years. The crux of this proposition is in selecting the model to transfer. We offer a method with which to select a model from another year based on the similarity in SWE distribution at existing snow pillows in the area. Comparison of the best transferred predictions and the selected predictions results in a mild decrease in r2 (0.02) and moderate increases in % MAE (15 %) and % Bias (10 %). The results motivate further refinement in the technique used to select the best model because if these dates can be identified then SWE can be modeled at accuracy levels equivalent to models generated from ASO data collected on the day of interest. Lastly, we found that models from ASO observations of anomalously low snowpacks in 2015 still transferred to other years, although the same cannot be said for the reverse. This research provides a first attempt at extending the value of ASO beyond the observations and we expect ASO will continue to provide insights for improving water resource management for years to come.