Meteorological Applications · 2019 · 49 citations · 61 references
Natural SciencesEarth SciencesOceanographyOcean MixingPhysical OceanographyWeather ForecastingData AssimilationMixed Layer DynamicsNumerical Weather Prediction Data AssimilationOcean ForecastingOcean-atmosphere CouplingTropical Cyclone ForecastingAir-sea Interaction ModelingLandfall ImpactsOcean Boundary Layer ModelingTropical Cyclone Landfall PredictionTropical Cyclone Rapid IntensificationTropical Cyclone TracksTropical Cyclone-monsoon InteractionsTropical Cyclone-ocean Interaction
Abstract The sensitivity of the simulated tropical cyclone (TC) intensity and tracks to the different ocean mixed‐layer depth (MLD) initializations is studied using coupled weather research and forecasting (WRF) and ocean mixed‐layer (OML) models. Four sets of numerical experiments are conducted for two TCs formed during the pre‐ and post‐monsoon. In the control run (CONTROL), the WRF model is initialized without coupling. In the second experiment, the WRF‐OML model is initialized by prescribing the MLD as a constant depth of 50 m (MLD‐CONST). In the third experiment, the spatial varying MLD obtained from the formulation of depth of the isothermal layer (MLD‐TEMP) is used. For the fourth experiment (MLD‐DENS), the model is initialized with the density‐based MLD obtained from ARMOR‐3D data. The results indicate that the CONTROL exhibits an early intensification phase with a faster translation movement, leading to early landfall and the production of large track deviations. The coupled OML simulations captured the deepening phase close to the observed estimates, resulting in the reduction of errors in both the vector and along the tracks of the storm. The initialization of the different estimates of the MLD in the WRF‐OML shows that the TC intensity and translation speed are sensitive to the initial representation of the MLD for the post‐monsoon storm. The gradual improvements in the intensity and translation speed of the storm with the realistic representation of the OML are mainly due to the storm‐induced cooling, which in turn alters the simulated enthalpy fluxes supplied to the TC, leading to the better representation of secondary circulation and the rapid intensification of the storm.
61