THE PERFORMANCE OF LONG-TERM INTEGRATIONS OF A LIMITED-AREA MODEL WITH THE SPECTRAL BOUNDARY COUPLING METHOD

Some numerical experiments are conducted in order to investigate the performance of the SBC (spectral boundary coupling) method proposed by Kida et al. (1991). The SBC is a new method for long-term integrations of a limited area model for regional climate simulation studies. The modified Version of the FLM (the fine-mesh limited area model) which was operationally employed for short-period weather forecasting in the JMA (Japan Meteorological Agency) from 1982 to 1988 is used in present study. The model is nested in the GANL (the JMA operational global analysis data sets) using the one-way nesting method. One-month integrations are performed using (a) the SBC method and (b) a conventional lateral boundary condition (LBC), and these results are compared. In addition to those methods, the boundary relaxation method is used in order to suppress noise development near the lateral boundary. The simulated phase of synoptic-scale disturbances using the conventional LBC method tends to deviate from that of the GANL. However, when the SBC method with a spectral boundary condition of N-c = 3 is employed, the phase deviation from the GANL is small during the one-month integration. Two experiments with the spectral boundary conditions of N-c = 3 and N-c = 1 in the SBC method are compared in order to examine the dependence of the model performance on the specification of the spectral boundary wave number. The phase in the case of N-c = 1 deviates from that of the GANL as compared with the N-c = 3 case. Statistical verification analyses of the precipitation which are simulated using (a) the SBC method with N-c = 3 and (b) the conventional LBC method are conducted. The results show that the SBC method provides improved performance of precipitation in one-month integrations of the regional model.