Diagnosis of Track Forecast Errors for Hurricane Rita (2005) Using a High-Resolution Regional Reforecast Ensemble

Hurricane Rita was an intense tropical cyclone (TC) that developed over the western North Atlantic on 18 September 2005. Rita moved westward and intensified to Category-5 strength on the Saffir-Simpson scale over the central Gulf of Mexico on 22 September. Rita subsequently weakened to Category-3 as it turned northwestward and made landfall on the Texas-Louisiana border on 24 September. At its most intense phase near 0000 UTC 22 September, Rita deepened to 895 hPa, and operational track forecasts indicated a significant threat for Houston, Texas in the 24-72 hour time frame. All of the operational numerical guidance had a distinct left-of-track error, however, as the observed track of Rita fell along the extreme right side of the multi-model and NCEP Global Forecast System (GFS) ensemble of track forecasts. The aim of this presentation is to utilize an 11-member Advanced Weather Research and Forecasting (ARW) regional reforecast ensemble 72-h forecast to examine the factors that contributed to the left-of-track forecast errors for the 0000 UTC 22 September initialization.

Ensemble reforecasts are a long time series of ensemble forecasts from a stable data assimilation and forecast system. NOAA ESRL has recently generated a retrospective ensemble reforecast dataset using the 2012 version of the NCEP GFS ensemble (GEFS) forecast system. The reforecast dataset is comprised of an 11-member ensemble run once daily at 0000 UTC from 1985 to the present. The full resolution GEFS reforecast dataset was archived in mass storage to permit use as initial and lateral boundary conditions for high-resolution retrospective simulations with regional models such as ARW. The Hurricane Rita track forecasts initialized at 0000 UTC 22 September 2005 from the GEFS reforecast dataset were characterized by the same left-of-track error as the operational numerical model guidance. We will use the GEFS reforecast members as initial and boundary conditions to generate an 11-member high-resolution explicit ARW ensemble forecast that will be used to diagnose the Hurricane Rita track forecast errors. We will also examine the Rita track forecast errors in the context of the climatological track forecast errors for all TCs over the central and western Gulf of Mexico during the period of the GEFS reforecast dataset.

The results show that from the climatological perspective, 72–144 h track forecasts from the GEFS reforecast ensemble have a distinct left-of-track error over the central and western Gulf of Mexico. The left-of-track error occurs in conjunction with positive 500 hPa height errors north of the TC on the synoptic-scale that drives anomalous easterly flow – indicating that the subtropical ridge over the southeast U.S. is too strong in the GEFS reforecasts. The ARW regional reforecast for Rita revealed that the track forecasts were sensitive to the phase speed of midlatitude transients, where a more progressive pattern contributed to a more northward track and landfall near the Texas-Louisiana coastline. Conversely, a less progressive pattern and stronger subtropical ridge over the southeast U.S. contributed to a more westward track and landfall near Houston, Texas. Additionally, the vertical depth of the vortex also contributed to track errors, where the ensemble members with a deeper vortex moved on a more northward track compared to ensemble members with a shallower vortex.