Characterizing Biases in Mountain Snow Accumulation From Global Data Sets

Abstract

AbstractMountain snow has a fundamental role in regional water budgets through its seasonal accumulation, storage, and melt. However, characterizing snow accumulation over large regions remains difficult because of limited observational networks and the inability of available satellite instruments to remotely sense snow depth or water equivalent in mountains. Models offer some ability to estimate snow water storage (SWS) on mountain range to continental scales. Here we compare four commonly used global data sets to understand whether there is a consensus regarding mountain SWS estimates among them. The data sets—European Centre for Medium‐Range Weather Forecasts Reanalysis‐Interim, Global Land Data Assimilation System, Modern‐Era Retrospective Analysis for Research and Applications version 2, and Variable Infiltration Capacity—agree to within ±36% of the four–data set average for total global SWS. When mountain areas are extracted using a new seasonal mountain snow classification data set, the four data products have more agreement, where all are within ±21% of the seasonal SWS for mountain regions. However, when compared to high‐resolution (9 km) simulations of SWS from the Weather Research and Forecasting (WRF) regional model, the four global products differ from WRF‐estimated North American mountain snow accumulation by 40–66%, with a negative bias up to 651 km3, comparable to the annual streamflow of the Mississippi River. If we extend the North America SWS bias to global mountains, the global data sets may miss as much as 1,500 km3 of SWS, equivalent to 4% of the flow in all the world's rivers. The potential difference of SWS suggests more work must be done to characterize water resources in snow‐dominated regions, particularly in mountains.