Constraining Future Projections of Atmospheric Rivers Using a Multiobjective Model Evaluation Framework

Atmospheric rivers (ARs) are narrow, elongated, synoptic jets of water vapor that play important roles in the global water cycle, local weather, and regional hydrology. A recent study (Guan and Waliser 2017) ¹ developed and applied 17 AR performance metrics (including frequency, intensity, geometry, and seasonality, among others) to a suite of global climate models, and found a range of performance fidelity (relative to reanalysis) across the models considered. In addition, a more recent study (Espinoza et al 2018) ² quantified the change in global AR characteristics between future climate projections and historic simulations in a suite of CMIP5 models. In their case, a subset of the above AR metrics and simple multi-model ensemble averages were used as the primary basis for comparison. Here, we utilize and build on information from these two studies, including consideration of all 17 AR metrics for the CMIP5 models, to constrain end-of-century projections of global AR dynamics based on a multi-variate, observation-based model evaluation framework. We reduce uncertainties in the projections by eliminating less accurate or redundant simulations and focusing on the simulations that have high fidelity relative to reanalysis products (ERA-Interim and MERRA2). These constraints allow us to gain a higher confidence in the projected changes in AR activity over the next century.

¹ Guan, Bin, and Duane E. Waliser. "Atmospheric rivers in 20 year weather and climate simulations: A multimodel, global evaluation." Journal of Geophysical Research: Atmospheres122.11 (2017): 5556-5581.

² Espinoza, Vicky, et al. "Global Analysis of Climate Change Projection Effects on Atmospheric Rivers." Geophysical Research Letters 45.9 (2018): 4299-4308.