SWE Reanalysis Model Temperature Threshold Sensitivity Analysis in the Olympic Mountains (full report pdf in presentation and report folder)
CEE 568: Snow Hydrology
March 19, 2021
We analyzed the ability of the Particle Batch Smoother (PBS) Reanalysis model to estimate snow water equivalent (SWE) over the Olympic Mountains and compared model outputs that used three different temperature thresholds for precipitation partitioning. The goal of this study was to determine model sensitivity to the temperature threshold parameter and to develop a recommendation for the use of regional as opposed to continental thresholds. We compared the PBS model SWE output to airborne lidar-derived SWE data over the same set of days, masked out glaciers, and differenced the SWE maps. We show that there is strong agreement between the PBS output and the ASO data at the 0℃ threshold. A regression analysis confirms the strong agreement between model and ground truth that increases in strength as we step from the 2℃ temperature threshold to 1℃ and 0℃. This change in precipitation temperature partitioning threshold correlates to an approximately 10% increase in accuracy from the original 2℃ threshold to the updated 0℃ threshold. Our results show that the PBS model is sensitive to the temperature threshold, and that threshold choices above 0℃ lead to an overestimation of regional SWE. Accordingly, we recommend that the PBS model should regionalize the temperature threshold for precipitation partitioning in order to improve SWE estimation.