Modifying Mountains in Models Improves Pervasive Precipitation Bias in Western North America
Abstract
Most Global Climate Models (GCMs) overestimate precipitation in western North America, especially in winter, limiting confidence in the mechanisms controlling precipitation in this region and how its climate may change in the future. We assess the hypothesis that GCMs’ smoothing of the complex terrain of this region leads to this wet bias. To investigate the role of overly smooth mountains on precipitation biases, we use the GFDL CM2.5-FLOR model to simulate climate, and compare it to a version with topography raised to capture observed mountain peak heights (“HiTopo”) across the world. We find that this HiTopo simulation reduces winter precipitation biases in western North America by about 36% compared to the Control simulation. Experiments altering select regions of topography show that the vast majority of the bias improvement is due to mountains in North and Central America (as opposed to, e.g., Asian mountains). A moisture budget decomposition finds that the overall drying in western North America is due primarily to changes in the mean flow, in particular weakening of the westerlies along the west coast of North America that reduces moisture inflow from the Pacific Ocean. The weakened westerlies are shown to be associated with the enhanced orographic blocking. The bias reduction in precipitation is not accompanied by a bias reduction in the continental-scale mean flow at all levels, consistent with the expectation that wind biases (like precipitation biases) in GCMs have multiple causes. As a whole, these results imply orographic blocking remains insufficiently captured in GCMs, contributing to pervasive precipitation biases in western North America, and motivating future improvements to GCM representation of mountains’ resolved and subgrid-scale effects.
