Performance and Evaluation of the MRI Regional Climate Model with the Spectral Boundary Coupling Method

Abstract

A limited area model has to be integrated over a long period to simulate a regional climate. The spectral boundary coupling (SBC) method is used for a regional climate model to adjust the phase shift of disturbances between the outer coarse mesh model and the inner fine mesh model, and to suppress the noise which is inevitably developing near the lateral boundary in our model.
In this study, the MRI regional climate model (MRI/RCM) with a horizontal transformed grid size of 40km (at 60°) is nested in the Global Analysis data set (GANL), which was produced by the Numerical Prediction Division of the Japan Meteorological Agency (NPD/JMA). One-month long integrations of the model are performed for January and August 1994 with the following four different boundary conditions; (i) the conventional lateral boundary condition, namely, the boundary relaxation method without using the SBC method (CASE A), (ii) the SBC method used for all layers and without the boundary relaxation method (CASE B), (iii) the SBC method used only above the σ=0.5 level and without the boundary relaxation method (CASE C), and (iv) the boundary relaxation method at all layers and the SBC method used only above the σ=0.5 level (CASE D). The simulated results are verified in comparison with observations using some statistical methods. The results of the statistical evaluation show that when the SBC method is not used, the Root Mean Square Error (RMSE) of the height at the 850hPa level near the lateral boundary is smaller than that around the center of the calculation domain. In contrast, when the SBC method is used, the RMSE of it around the center is smaller than that near the lateral boundary. In the summer case, the height at the 850hPa level is improved by using the SBC method. However, using the SBC method for all layers is not good for the scores of surface air temperature and precipitation. Almost all the scores are improved by using the SBC method only above the σ=0.5 level. In the winter case, the boundary relaxation method suppresses noise effectively, while the SBC method shows a little improvement. The best score, however, is obtained from the combination of both the methods. Almost all the statistical values show that the SBC method for only above the σ=0.5 is useful for the regional climate model throughout both the seasons, particularly in the summer case.