Validation of COAMPS-TC Forecasts of Hurricane Joaquin (2015) using TCI Observations

Real-time deterministic dynamical model forecasts of Hurricane Joaquin’s track had unusually large errors, particularly the GFS and GFS-based regional TC prediction models. CTCX, the real-time GFS-based run of COAMPS-TC, was no exception with forecasts initialized early in the storm’s life consistently erring north and west of the observed track.

CTCX is a nested-grid model, with a fixed outer domain at 45 km horizontal resolution, and storm-following inner nests at 15 km and 5 km horizontal resolution (as configured in 2015). We have retrospectively run the new Basin-scale COAMPS-TC configuration for Joaquin, utilizing the same initial and boundary conditions as CTCX. Basin-scale COAMPS-TC employs a very large fixed 5 km grid covering the same region as the CTCX outer domain, in essence utilizing the CTCX innermost grid “everywhere”. The Basin-scale COAMPS-TC track forecasts for Joaquin are far closer to the observed track than the real-time CTCX forecasts, capturing the unusual southwestward motion much better than CTCX and the other GFS-based models.

Here we assess why the Basin-scale COAMPS-TC track forecasts for Joaquin are superior to those of CTCX, validating the simulation of the vortex and the near-storm environment using HDSS dropsonde observations taken during TCI, as well as using ECMWF analyses. We also examine the track performance for Basin-scale COAMPS-TC in the context of a large sample of recent forecast cases in the Atlantic and Western North Pacific to ascertain whether the Joaquin results have some general implications for Basin-scale vs. nested-grid track prediction.