Abstract
Numerical experiments of tropical cyclone motion are performed with the use of a triply-nested grid model which has been developed primarily for the study of tropical cyclones. One of the features of the model is that the mixing ratios of cloud water and rainwater are taken to be predicted variables because evaporation of rainwater and convective downdraft are considered to be important. Cumulus parameterization and other aspects of the model are basically the same as those described in Yamasaki (1986, 1987) except for the use of the longitude-latitude coordinates, inclusion of topography, finer vertical resolution (ten-layer model) and so on. The grid sizes of the coarse, intermediate and fine grid areas are taken to be 15/4, 5/4 and 5/12 degrees, respectively.
Several typhoons in August and September 1990 (SPECTRUM period) are chosen for the numerical experiments. The initial condition is taken from the objectively analyzed global data of JMA. An axially symmetric vortex is superimposed on this data. The objective of this study is to see to what degree the model can predict typhoon motion and what are the primary causes of prediction errors. It is found that while the model appears to show somewhat good performance for the prediction of typhoon motion, the primary causes of prediction errors are poor prediction of the behavior of the subtropical high and the initial fields of wind and conditional instability.
