Evaluation of the Large-Scale Environment in NOAA Experimental Models

The simulated large-scale environment may have a significant impact on the accuracy of the track prediction of high-resolution tropical cyclone forecasting models. With increased demand for a greater reduction in track forecast error at ever increasing lead times, attention must be paid to the degree to which variations in the simulated large-scale environment impact the accuracy of track forecasts. NOAA’s current operational hurricane prediction system, HWRF, as well as many of its experimental hurricane prediction systems, for example, the ‘basin-scale HWRF’ that is being developed at the NOAA’s Hurricane Research Division of the Atlantic Oceanographic and Meteorological Laboratory, rely on initial and boundary conditions as provided by a global forecasting system, the Global Spectral Model. With the selection of the global FV3 model, developed by NOAA’s Geophysical Fluid Dynamics Laboratory, to replace the Global Spectral Model as the dynamical core for NOAA’s future Global Forecasting System (fvGFS), it is worthwhile to investigate the large-scale environment it produces as well.

With support from the Hurricane Forecast Improvement Project (HFIP), this presentation seeks to compare the large-scale environment, as generated from one or more of NOAA’s operational and/or experimental models to that of the current global standard, the Global Spectral Model, over portions of the 2017 Atlantic Hurricane season. A method following that of Riehl 1965, is applied to employ the 500mb geopotential height as a proxy for the overall large-scale environment. An area-averaged root mean squared error estimate is used to quantitatively compare the results of the various models. Emphasis is given to select cases having significant, highly-adverse impact on large populations of the coastal United States, including Hurricanes Harvey, Irma, and/or Maria.