Evaluating the Representations of Atmospheric Rivers and Their Associated Precipitation in Reanalyses With Satellite Observations

Abstract

Atmospheric rivers (ARs) are filaments of enhanced horizontal moisture transport in the atmosphere. Due to their prominent role in the meridional moisture transport and regional weather extremes, ARs have been studied extensively in recent years. Yet, the representations of ARs and their associated precipitation on a global scale remains largely unknown. In this study, we developed an AR detection algorithm specifically for satellite observations using moisture and the geostrophic winds derived from 3D geopotential height field from the combined retrievals of the Atmospheric Infrared Sounder and the Advanced Microwave Sounding Unit on NASA Aqua satellite. This algorithm enables us to develop the first global AR catalog based solely on satellite observations. The satellite‐based AR catalog is then combined with the satellite‐based precipitation (Integrated Muti‐SatellitE Retrievals for GPM) to evaluate the representations of ARs and AR‐induced precipitation in reanalysis products. Our results show that the spreads in AR frequency and AR length distribution are generally small across data sets, while the spread in AR width is relatively larger. Reanalysis products are found to consistently underestimate both mean and extreme AR‐related precipitation. However, all reanalyses tend to precipitate too often under AR conditions, especially over low latitude regions. This finding is consistent with the “drizzling” bias which has plagued generations of climate models. Overall, the findings of this study can help to improve the representations of ARs and associated precipitation in reanalyses and climate models.

Assessment of Storm‐Associated Precipitation and Its Extremes Using Observational Data Sets and Climate Model Short‐Range Hindcasts

“…The overestimation of light precipitation is a long-standing "frequent drizzle problem" for many GCMs (e.g., Chen et al, 2021), though it had been improved from E3SM v1 to v2 by revised convective triggering function (Golaz et al, 2022;. In a recent study (W. Ma et al, 2023), the drizzle problem also been found in reanalysis for AR precipitation, which is consistent with our results. The PDFs also shown that modeled precipitation rarely exceeds 10 mm hr 1 .…”