Contrasting spatial and temporal analogs for evaluating climate sensitivity of western U.S. snowpacks using a large snow monitoring network

Empirical sensitivity analyses are important for evaluation of the effects of a changing climate on water resources and ecosystems. Although mechanistic models are commonly applied for evaluation of climate effects for relatively simple processes, like snowmelt, empirical relationships provide a first-order validation of the various postulates required for their implementation. Recent studies of empirical sensitivity for snow in the western United States have focused on April 1 snow water equivalent (SWE), and were developed by regressing interannual variations in SWE to interannual variations in winter precipitation and temperature. This logic offers a temporal analog for climate change impacts, positing that a warmer future looks like warmer years do now. Spatial analogs hypothesize that a warmer future may look like warm places, and are frequently applied alternatives for complex processes, states, or metrics that show little interannual variability (e.g. forest cover), or data with short records. In this presentation, we contrast spatial and temporal analogs for sensitivity of April 1 SWE and the mean residence time of snow using data from 524 Snowpack Telemetry (SNOTEL) stations across the western US. Like previous studies, we found the poorest temporal analog relationships in areas showing the highest sensitivity to warming, and temporal relationships for mean residence time were generally poorer than those for April 1 SWE. In contrast the spatial analog sensitivities showed strong correlations (Nash Sutcliffe R2=0.87 for April 1 SWE and 0.81 for mean residence time) using just winter (Nov-Mar) precipitation and average temperature. Partial derivatives of the spatial relationships were steeper, and showed a substantially heightened sensitivity in warmer areas than did the temporal analogs. High elevation stations also showed greater vulnerability considering a spatial analog than has been shown in previous modeling and sensitivity studies. The spatial analog models provide a simple perspective to evaluate potential futures and may be useful in further evaluation of snowpack futures.