Probabilistic Analysis of the 2017 Atlantic Hurricane Season from a Continuously Cycled Basin-Scale HWRF Data Assimilation and Prediction System

The 2017 Atlantic hurricane season was among the most active periods ever recorded for this basin, featuring several intense, long-lived storms, and multiple hurricane landfalls. This season also coincided with the development of a new testbed for model development, observing system design, ensemble forecasting, and satellite data assimilation research at the NOAA Atlantic Oceanographic and Meteorological Laboratory (AOML) Hurricane Research Division (HRD). The new testbed adopts the operational Hurricane Weather Research and Forecasting (HWRF) model framework but uses an extensive fixed domain similar to an experimental “basin-scale HWRF” model run in real time at AOML/HRD. This configuration allows for the continuous assimilation of conventional and satellite measurements over a large 18-km grid spacing domain, which is performed using the community Development Testbed Center (DTC) ensemble Kalman filter (EnKF) and a 60-member HWRF ensemble. Unlike past applications of the HWRF model, the current system runs non-stop through hurricane seasons, using the NCEP GFS model only to supply boundary conditions. The resulting strategy supports a systematic season-long evaluation of the HWRF model and delivers a valuable set of analyses for hurricane modeling research at AOML/HRD. This presentation provides an overview of how HWRF represents dynamical features governing the development and maintenance of major tropical cyclones over an active three-month period from this past season. Special attention will be placed on the probabilistic representation of storm-environment interactions from ensemble analyses and forecasts and model biases revealed from continuously cycling model states during data assimilation.