How Mountains Impact Precipitation Simulation in the Western U.S.

A common bias in comprehensive Global Climate Models (GCMs) is overestimation of precipitation in the western U.S., which limits our confidence in how the climate of this region may change in the future. Prior work has suggested this bias is partially due to SST biases in atmosphere-ocean coupled GCMs. In this work we explore the alternative possibility that GCMs' portrayal of the complex terrain of this region, namely the high elevation Rockies, Cascades, and Sierra Nevada, leads to this wet bias. In GCMs, placing observed topography on a model grid smooths out observed mountain peak heights, which may reduce their impacts on circulation. To investigate the role of mountain peak height on the precipitation biases, we use the GFDL CM2.5-FLOR model (~0.5° res. atmosphere, ~ 1° res. ocean) to simulate climate with altered topography. Our simulation, called HiTopo, changes the surface height in every grid cell from the average observed elevation in that cell to the maximum. We find that the HiTopo simulation does a noticeably better job than the Control (standard surface height) simulation at simulating precipitation patterns over the Western U.S., especially during northern hemisphere winter (DJF). In our search to determine what causes this improvement, we find the stationary wave pattern associated with the raised surface height causes a weaker westerly flow from the Pacific Ocean, and a southward shift in the jet stream. This in turn reduces moisture coming into the western U.S. and precipitation in this region. This analysis is supported by a moisture budget decomposition for the region and an examination of whether the forcing is driven primarily by local mountains (the Rockies) or by distant ones (the Tibetan Plateau and Himalayas). Possible implications for future projection of western U.S. precipitation and snowpack will be discussed.