Using Dynamical Adjustment to Estimate the Anthropogenically-forced Response of Surface Temperature and Precipitation within a High-resolution Regional Climate Model: A Case Study of the Pacific Northwest

Abstract:

One of the greatest challenges in regional climate prediction is distinguishing the anthropogenically-forced response from low-frequency internal variability. In a large ensemble, the forced response is well approximated by the mean trend of the ensemble members. However, in mountainous regions like the Pacific Northwest, very high model resolution is required to accurately represent the terrain, making large ensembles prohibitively expensive. Here we take a different approach, employing a statistical technique called "dynamical adjustment" to estimate the forced response of wintertime (DJF) surface temperature and precipitation within two high-resolution simulations of the 21st-century climate in the Pacific Northwest. The simulations were performed at 12-km resolution using the Weather Research and Forecasting Model, downscaled from global CCSM3 and ECHAM5 simulations under an A1B emissions scenario. While the raw simulations exhibit large differences in the magnitude and spatial structure of precipitation and surface temperature trends, dynamical adjustment results in much better agreement between the simulations, especially with regard to projected surface warming. These results suggest that dynamical adjustment of a small number of high-resolution simulations can provide much of the benefit of a large ensemble, but at far less computational expense.