Evidence for spring mountain snowpack retreat from a Landsat-derived snow cover climate data record

Abstract

Abstract. A Landsat snow cover climate data record (CDR) of visible mountain snow-covered area (SCA) across interior northwestern USA during spring was compared with ground-based snow telemetry (SNOTEL) snow-water-equivalent (SWE) measurements and mean surface temperature and total precipitation observations. Landsat spring SCA on 1 June was positively correlated with 15 May and 1 June SWE, negatively correlated with spring temperatures (April–June), and positively correlated with March precipitation. Using linear regression with predicted residual error sum-of-squares (PRESS) cross-validation, spring SCA was reconstructed (1901–2009) for the mountains of central Idaho and southwestern Montana using instrumental spring surface temperature records. The spring SCA reconstruction shows natural internal variability at interannual to decadal timescales including above average SCA in the 1900s, 1910s, 1940s-1970s, and below average SCA in the 1920s, 1930s, and since the mid 1980s. The reconstruction also reveals a~centennial trend towards decreasing spring SCA with estimated losses of −36.2 % since 1901. Based on the inferred thermal relationship between temperature and snow, strong evidence emerges for mountain snowpack retreat triggered by spring warming, a signal that includes both feedback and response mechanisms. Expanding snow cover CDRs to include additional operational satellite retrievals will add temporal SCA estimates during other snow accumulation and melt intervals for improved satellite-instrumental climate model calibration. Merging Landsat snow cover CDRs with instrumental climate records is a formidable method to monitor climate-driven changes in western US snowpack extent over 20th and 21st centuries.