High-resolution statistically downscaled climate and hydrology projections over Alaska

Ensemble, large domain, climate projections and hydrologic modeling at high resolution has the potential to provide better understanding of potential future shifts in regional water budgets due to a changing climate at various spatiotemporal scales. Very few century-long, high-resolution assessments over Alaska have been performed to date compared to the large number of hydrologic projection assessments over the Contiguous United States (CONUS). Here we perform multiple off-line hydrologic model simulations forced by statistically downscaled climate model outputs to obtain hydrologic simulations from 1950 to 2099 over the entire Alaska domain (except the very western-most Aleutian Islands).

We take an approach similar to our recent CONUS hydrologic assessments: 1) select 25 climate model outputs produced based on Representative Concentration Pathway 8.5; 2) perform statistical downscaling to generate high-resolution climate input data for hydrologic models (12 km grid-spacing here); and 3) perform off-line hydrologic model simulations, in this case using the Variable Infiltration Capacity (VIC) model using the simplified daily time-stepping water-balance configuration. We evaluate the spatiotemporal characteristics of differences among the downscaled climate model ensembles in retrospective simulations (1977-2006) across: 1) the downscaled climate data (precipitation and temperature); and 2) hydrologic model outputs (runoff, evapotranspiration, snow, and soil moisture, and streamflow). We also examine the hydroclimate shift of these hydro-climate variables from the retrospective period to the end of the 21^st century (2071-2100).

In addition to those water-balance-only simulations, we perform VIC model simulations using the full water and energy balance model to simulate frozen soils. We discuss differences across the simulations and challenges, including:1) a lack of long-term, comprehensive time series of observations at sites from differing hydro-climates across the domain which hinders model calibration and evaluation; and 2) the need for inclusion of several additional complex hydrologic processes prevalent in Alaska (glaciers, permafrost, river ice), yet not extensively evaluated in the VIC model in the context of climate projection simulations.