Revealing Causes of a Surprising Correlation: Snow Water Equivalent and Spatial Statistics from Calibrated Enhanced-Resolution Brightness Temperatures (CETB) using Interpretable Machine Learning and SHAP Analysis

Mahboubeh Boueshagh ^1* Joan M Ramage ¹ Mary J Brodzik ² David G Long ³ Molly A Hardman ² Hans-Peter Marshall ⁴

• ¹ Lehigh University, Bethlehem, United States
• ² National Snow and Ice Data Center, University of Colorado Boulder, Boulder, Colorado, United States
• ³ Brigham Young University, Provo, Utah, United States
• ⁴ Boise State University, Boise, Idaho, United States

The seasonal snowpack serves as a crucial water source, and accurate Snow Water Equivalent (SWE) estimation is vital for water resource management and environmental assessment. This study introduces a novel approach to Passive Microwave (PMW) SWE estimation, exploiting the strong, unexpected correlation between SWE and the Spatial Standard Deviation (SSD) of PMW Calibrated Enhanced-Resolution Brightness Temperatures (CETB). By integrating spatial statistics, linear and machine learning methods, and SHapley Additive exPlanations analysis, this research aims to assess CETB SSD as a key feature for improving SWE estimations and other environmental retrievals, enhancing the representation of land surface and snowpack variability. Analysis at three diverse sites—Monument Creek, AK, Mud Flat, ID, and Jones Pass, CO—reveals site-specific drivers of SSD variability, with SWE as the dominant factor at Monument Creek and Mud Flat, while soil moisture plays a leading role at Jones Pass. In snow-dominated areas with less surface heterogeneity, such as Monument Creek, SSDs can improve SWE estimations by capturing spatial variability. In more complex environments like Jones Pass, SSDs can aid SWE estimations by accounting for additional variables, such as soil moisture, that significantly impact snowpack dynamics. Another theory is that SSDs can capture a range of environmental variations, which may help better represent the complexities involved in SWE estimations across diverse terrains. The findings demonstrate the potential of CETB SSD to enhance remote sensing capabilities for snow research across varied environments, offering benefits for hydrological modeling and water resource management.