Climate Model Evaluation of Atmospheric Rivers over the Continental United States

Abstract:

Atmospheric rivers (ARs)—long corridors of intense atmospheric water vapor transport—significantly influence the hydrologic cycle and regional hydrometeorological extremes across the continental United States (CONUS). Ongoing and future climate change may alter AR characteristics and impacts, making confident climate model projections of future change, especially at regional scales, of critical importance. In order to better constrain uncertainty in such projections of future change, we perform a comprehensive climate model evaluation of AR climatology over the CONUS. Using an established AR detection algorithm, here we evaluate the representation of ARs in historical simulations (1984-2013) from a suite of models participating in the sixth phase of the Coupled Model Intercomparison Project (CMIP6). Models are evaluated against the Modern-Era Retrospective Analysis for Research and Applications, Version 2 (MERRA-2) reanalysis. Model performance for individual models and the multi-model mean is presented for AR frequency, intensity, geometry, and seasonality in order to highlight systematic biases. Results are summarized over the seven US National Climate Assessment regions. Results suggest that most CMIP6 models reproduce the reanalysis AR climatology and features with broadly reasonable fidelity, however some notable biases exist and some models provide more realistic representations than others. These results help inform future AR projection studies for the CONUS, identifying in which regions and variables we can place greatest confidence.