The Prediction of Tropical Cyclone Wind Structure Using the Analog Ensemble Technique

The prediction of tropical cyclone (TC) wind structure is an important goal of operational forecasting centers since the structure and intensity of the wind field determines the aerial extent various TC impacts including damaging winds, storm surge, waves, and evacuation zones. The National Hurricane Center and Joint Typhoon Warning Center currently provide official forecasts of significant TC wind radii (i.e., 34-, 50-, and 64-kt wind radii, hereafter R34, R50, and R64) out to various lead-times. Both statistical models and dynamical models are used to estimate a size and wind structure of a tropical cyclone, but they often over-predict the swath of probabilities of significant wind radii, especially in the 5-7-day timeframe.

Here, we employ the analog ensemble (AnEn) method to the deterministic Hurricane Weather Research and Forecasting (HWRF) model (2017 version) to not only improve upon the HWRF prediction of storm structure, but furthermore, produce a computationally inexpensive and naturally calibrated ensemble prediction of significant storm radii. In the AnEn, a set of analog forecasts is created by searching archived HWRF forecasts that share key features with a current forecast from the same configuration of the HWRF. The meteorological variables used to identify a past forecast similar to the current one are called analog predictors. The actual observational estimates of the significant wind radii associated with the best analog forecasts are used in the ensemble forecast. The AnEn is developed here using HWRF reforecast data (2014-2016) and corresponding Best Track significant wind radii. AnEn models predicting significant wind radii are developed for both the Atlantic and Eastern Pacific Ocean basins. This presentation will overview the overall performance of the AnEn models, including deterministic and ensemble forecast skill metrics. We will also overview real-time performance of the AnEn for the 2017 hurricane season.