Evaluation of the ocean component on different coupled hurricane forecasting models using upper-ocean metrics relevant to air-sea heat fluxes during Hurricane Dorian (2019)

Abstract

In August 2019 Hurricane Dorian traveled through the Caribbean Sea and Tropical Atlantic before devastating the Bahamas. The operational hurricane forecasting models under-predicted the intensity evolution of Dorian prior to the storm reaching its maximum strength. Research studies have shown that a more realistic upper-ocean characterization in coupled atmosphere-ocean models used to forecast hurricanes has the potential to lead to more accurate hurricane intensity forecasts. In this work, we evaluated four ocean products: the ocean component from one NOAA operational hurricane forecasting model that used ocean initial conditions from climatology, the ocean components from two NOAA experimental models using ocean initial conditions from a data-assimilative operational ocean model, and one US Navy data-assimilative operational ocean model for reference. The upper-ocean metrics used to evaluate the models include mixed layer temperature, mixed layer salinity, ocean heat content and depth-averaged temperature in the top 100 m. The observations used are temperature and salinity profiles from an array of six autonomous underwater gliders deployed in the Caribbean region during the 2019 hurricane season. We found that, even though the four models have good skill in predicting temperature and salinity over the whole observed water column, skill significantly deteriorates for the upper-ocean metrics. In particular, the models failed to capture the barrier layer that was present during the passage of Hurricane Dorian through the glider array. We also found that even small differences in the mixed layer temperature along the storm track on the hurricane models evaluated, led to noticeable differences in the total enthalpy fluxes delivered from the ocean to the atmosphere throughout the storm’s synoptic history. These findings highlight the need to improve the upper-ocean initial conditions and representation in coupled atmosphere-ocean models as part of the larger efforts to improve the various modeling aspects that control the hurricane intensity forecast.