Attribution of the spring snow cover extent decline in the Northern Hemisphere, Eurasia and North America to anthropogenic influence

Nathan P Gillett,Mohammad Reza Najafi,Francis W Zwiers

doi:10.1007/s10584-016-1632-2

Abstract

While it is generally accepted that the observed reduction of the Northern Hemisphere spring snow cover extent (SCE) is linked to warming of the climate system caused by human induced greenhouse gas emissions, it has been difficult to robustly quantify the anthropogenic contribution to the observed change. This study addresses the challenge by undertaking a formal detection and attribution analysis of SCE changes based on several observational datasets with different structural characteristics, in order to account for the substantial observational uncertainty. The datasets considered include a blended in situ-satellite dataset extending from 1923 to 2012 (Brown), the National Oceanic and Atmospheric Administration (NOAA) snow chart Climate Data Record for 1968–2012, the Global Land Data Assimilation System version 2.0 (GLDAS-2 Noah) reanalysis for 1951–2010, and the NOAA 20th-century reanalysis, version 2 (20CR2) covering 1948–2012. We analyse observed early spring (March-April) and late spring (May-June) NH SCE extent changes in these datasets using climate simulations of the responses to anthropogenic and natural forcings combined (ALL) and to natural forcings alone (NAT) from the Coupled Model Intercomparison Project Phase 5 (CMIP5). The ALL-forcing response is detected in all of the observed records, indicating that observed changes are inconsistent with internal variability. The analysis also shows that the ALL-forcing simulations substantially underestimate the observed changes as recorded in the Brown and NOAA datasets, but that they are more consistent with changes seen in the GLDAS and 20CR2 reanalyses. A two-signal analysis of the GLDAS data is able to detect the influence of the anthropogenic component of the observed SCE changes separately from the effect of natural forcing. Despite dataset and modelling uncertainty, these results, together with the understanding of the causes of observed warming over the past century, provide substantial evidence of a human contribution to the observed decline in Northern Hemisphere spring snow cover extent.

Highlights

Northern Hemisphere (NH) snow cover extent (SCE) has declined over the past 90 years with the greatest reduction (~3.3 million km2) observed during the 1980s (Vaughan et al 2013)
The GLDAS and 20CR2 reanalyses show March-April rates of decline that are comparable to those seen in the Brown dataset for the same period, but exhibit lower rates of SCE decreases in May-June compared to March-April
The observed trends are regressed onto the simulated responses to ALL and natural forcings alone (NAT) forcing to determine whether the effects of anthropogenic forcings (ANT), natural forcings (NAT) as well as anthropogenic and natural forcing signals combined (ALL) are reflected in the observations

Summary

Introduction

Northern Hemisphere (NH) snow cover extent (SCE) has declined over the past 90 years with the greatest reduction (~3.3 million km2) observed during the 1980s (Vaughan et al 2013). Rupp et al (2013) detected the response to a combination of anthropogenic and natural forcings in observed SCE by comparing observed March and April SCE changes over the period 1922–2005 based on the dataset of Brown and Robinson (2011; see Brown, 2000) with an ensemble average of CMIP5 (Coupled Model Intercomparison Project Phase 5) simulations that were driven with historical natural and anthropogenic forcings combined (ALL). They showed that while SCE in the models declined, as expected, the model-simulated trends were only half as strong as observed. In several datasets starting before 1979 to determine whether changes are the consequence of external forcing

Data and methods

Observed and simulated changes

Detection and attribution analysis

Discussion and conclusions