Modeling the changing roles of snow and permafrost in mid- and high-latitude climate systems

Abstract

The land surface plays a key role in local and regional climates at mid- and high-latitudes as well as in the global climate system. Consequently, changes in snow and permafrost affect other parts of the climate system. In this dissertation, we explore the role of the land surface in the cryosphere, with a particular focus on high latitudes, using a hierarchy of standalone land surface models (LSMs), fully-coupled regional climate models (RCMs) and global climate models (GCMs). In Chapter 2, I describe simulated changes in snowpack and fire potential in the western US using the Variable Infiltration Capacity (VIC) hydrology model under future climate projections for an ensemble of GCMs from the Coupled Model Intercomparison Project (CMIP5) archive for two Representative Concentration Pathways (RCPs), RCP4.5 and RCP 8.5. Large losses of snowpack and increases in fire potential are projected to occur in the mountainous parts of the western US in the 21st century, whereas increases in fire potential are much more uncertain in lowland regions due to large uncertainty in precipitation projections. In Chapter 3, I draw on two modeling ensembles, the Community Earth System Large Ensemble (CESM-LE) and the CESM Low Warming Ensemble (CESM-LWE), to understand projected changes in snow and how these changes will affect soil thermal regimes and permafrost in the 21st century over the circumpolar Arctic for three levels of warming: 1.5°C, 2°C and RCP 8.5. Even for the lower emissions scenarios represented by the 1.5°C and 2°C global-mean warming pathways, the majority of the Arctic is projected to experience significant decreases in Snow Water Equivalent (SWE), while parts of Eurasia will experience substantial increases. Large losses of permafrost are projected due to a significant warming of the soil column by the end of the 21st century. Soil organic carbon (SOC) stocks are highly vulnerable and loss of permafrost could result in potentially large losses of carbon to the atmosphere. In Chapter 4, I describe the process of designing a new parameter set for application over a pan-Arctic domain in version 5 of the VIC hydrology model (VIC-5) and in the Regional Arctic System Model (RASM), a fully-coupled regional climate model. Simulated streamflow in RASM simulations is significantly higher than in standalone VIC-5 simulations and much more closely matches observations, while simulated permafrost in standalone VIC-5 simulations more closely approximates observed permafrost extent, illustrating the difficulties of designing land surface parameters for application in a land surface model that is used in both standalone and fully-coupled modeling contexts.