A179-0010
In Search of The Optimal Atmospheric River Index for Precipitation: Big Data Analysis of Index Ensembles over the North American West Coast and US Midwest

Tuesday, 15 December 2020
Poster
Chen Zhang, Purdue University, West Lafayette, IN, United States, Wen-wen Tung, Purdue Univ, West Lafayette, IN, United States and William S. Cleveland, Purdue University, Statistics and Computer Science, West Lafayette, IN, United States
Abstract:
Atmospheric rivers (ARs) significantly affect surface hydrometeorology in western North America and the Midwest. Here, we systematically sought optimal AR indices suited for expressing surface precipitation effects within the prevailing detection and tracking frameworks studied in the Atmospheric River Tracking Method Intercomparison Project (ARTMIP). We adopted a multifactorial ensemble approach. Four factors—moisture fields, climatological thresholds, shape criteria, and duration thresholds—collectively generated 81 West Coast AR indices and 81 Midwest indices.

Two moisture fields for identification were extracted from ARTMIP MERRA-2 Tier 1 catalogs: integrated water vapor transport (IVT) and integrated water vapor (IWV). Global Precipitation Climatology Project One-Degree Daily Precipitation data were used. Metrics for precipitation effects included two-way summary statistics relating the concurrence of AR and that of precipitation, per-event averaged precipitation rate, and per-event precipitation accumulation. The generation of AR index ensembles and detailed analysis were executed via distributed-parallel computing on a Hadoop cluster using R-based DeltaRho software.

We found that an optimal AR index depends on several factors: types of impact to be addressed, associated physical mechanisms in the affected regions, and timing and duration. In West Coast and Midwest, IWV-based AR indices identified the most abundant AR event time steps, most accurately associated AR to days with precipitation, and represented the gross presence of precipitation the best. Combined with a permissive climatological threshold (e.g., 75th percentile), they detected the most accumulated precipitation with the longest event duration. IWV-based indices are the overall choice for Midwest ARs. For the West Coast landfalling ARs, IVT-based indices suitably captured the accumulation of intense orographic precipitation. Combined IVT and IWV, with restrictive climatological thresholds, focused on extreme West-Coast AR precipitation but resulted in the fewest records and shortest duration. Our findings provide useful information for creators and users of AR indices who consider surface precipitation in their decision processes.