Numerical problems associated with tropical cyclone intensity prediction using a sophisticated coupled typhoon-ocean model

Abstract

When tropical cyclone (TC) intensity is predicted using a sophisticated numerical model under a given initial condition with a typhoon bogus, numerical problems, particularly associated with the tendency errors, arise as well as problems related to the intensity limitation due to relatively coarse horizontal resolution. In order to investigate the problems, numerical experiments are performed for nine TCs in the western North Pacific from 2000 to 2002 using the typhoon model (TYM) and typhoon-ocean coupled model (CTYM). CTYM reduces the overdevelopment that occurs in the later integration predicted by TYM due to local sea surface cooling caused by the passage of TCs. However, CTYM hardly improves the tendency errors that occur in the early integration. The errors are found in all predictions using CTYM with each of the three physical schemes and have different characteristics for typhoons Bilis (2000), Wutip (2001), and Phanfone (2002). The TC thermodynamic structures of Wutip also differ from the three predictions using CTYM with each of the three schemes even at almost the same integration time and central pressure. Under the steady-state assumption, we estimate the maximum potential intensity (MPI) for three schemes and two TCs from the two-dimensional axisymmetrical mean structure. Assuming that the MPI is estimated from the net gain energy through the isothermal process scale-analyzed from the Colioris force and adiabatic process from the centrifugal force, it is found that predominance of convective available potential energy due to the adiabatic process leads to the underdevelopment of TCs. The improvement of physical schemes in CTYM is planned for the underdevelopment: revising a surface boundary formulation or introducing a sophisticated planetary boundary formulation contributes to increasing the energy caused by the isothermal process, while revising the precipitation or cumulus parameterization contributes to reducing the energy caused by the adiabatic process. © 2007 by the Japan Meteorological Agency/Meteorological Research Institute.