An Update on COAMPS-TC Performance: New Validation Tools and Results for Recent Real-time and Retrospective Forecasts

In 2013, the Naval Research Laboratory continued the ongoing real-time forecast demonstration of COAMPS-TC for tropical cyclones worldwide. In addition, a large sample of retrospective COAMPS-TC forecasts was produced for Atlantic and Eastern North Pacific tropical cyclones from 2010-2012, as part of the Hurricane Forecast Improvement Program ‘Stream 1.5' retrospective test. These real-time and retrospective COAMPS-TC forecasts, all produced with the same version of the model, provide an excellent sample for in-depth validation of the model. Here, we focus on characterizing state-dependent and overall biases in track and intensity forecasts, in addition to reporting the standard summary accuracy measures. We also apply newly-developed surface wind radii verification tools to validate Atlantic basin COAMPS-TC forecasts against the best-track values, and compare the relative frequency distributions of COAMPS-TC forecast intensity and best-track intensity. Finally, we use validation of forecast fields against GFS analyses to examine select cases in which COAMPS-TC made a very poor track forecast. Our findings indicate that the most extreme track errors typically occur due to the nonlinear interaction of errors in the short-range prediction of tropical cyclone position and longer-range prediction of a mid-latitude trough, resulting in premature recurvature of the simulated tropical cyclone.