A Rainfall Evaluation of Deterministic and Ensemble Basin-Scale HWRF

Fresh water flooding associated with tropical cyclone (TC) rainfall can have catastrophic impacts on life and the economy. In 2017, three major hurricanes struck the U.S. – Hurricanes Harvey, Irma, and Maria. Hurricane Harvey produced exceptional rainfall amounts in Texas and Louisiana causing historical flooding, at least 68 deaths, and was the second costliest economic loss from a hurricane on record. Under the Hurricane Forecast Improvement Project (HFIP), the NOAA/AOML/Hurricane Research Division (HRD) developed and maintains an experimental version of the Hurricane Weather Research and Forecasting (HWRF) Model, called the “Basin-Scale HWRF” (HWRF-B). The 2017 version of this experimental model (HB17) produced low track errors compared to other operational models. Due to the high dependence of precipitation on TC track, HB17 is an excellent research tool to study TC rainfall. In 2018, HRD further upgraded HWRF-B to produce a 10-member ensemble of forecasts in order to explore the development of TC probabilistic hazard guidance. This study utilizes several rainfall evaluation tools developed in previous studies to: (1) evaluate HB17’s rainfall prediction performance for Hurricane Harvey; and (2) evaluate HB18 ensemble rainfall forecast performance. In both cases, HWRF-B precipitation estimates were compared with two observational datasets, the National Center for Environmental Prediction (NCEP) hourly national multi-sensor (Stage IV) and the Integrated Multi-satellitE Retrievals for GPM (IMERG) rainfall estimates. This evaluation validated rainfall performance in terms of peak and average rainfall amounts, rainfall distributions, and equitable threat scores. Moreover, this study provides a strategy for future TC rainfall evaluation using the probabilistic precipitation guidance based on the HWRF-B ensemble rainfall estimates.