An Evaluation of the NCEP RSM for Regional Climate Modeling

Abstract

In this study, the National Centers for Environmental Prediction (NCEP) Regional Spectral Model (RSM) has been evaluated as a means of enhancing the depiction of regional details beyond that which is capable in low-resolution global models. Three-month-long simulations driven by the NCEP?National Center for Atmospheric Research 40-yr reanalysis data are conducted with a horizontal resolution of about 50 km over the United States, for the two winters and summers. The selected winter cases are December?February (DJF) 1991/92 (warm eastern Pacific SST anomalies) and DJF 1992/93 (normal eastern Pacific SST anomalies). Summer cases are May?July (MJJ) 1988 (a drought in the Great Plains) and MJJ 1993 (a flooding). Overall, the results from the model are very satisfactory in terms of the precipitation distribution for different seasons as well as the representation of large-scale features. Evaluation of simulated large-scale features reveals that the model does not exhibit a discernible synoptic-scale drift during the 3-month integration period, irrespective of the seasons. Surprisingly, the model simulation is found to correct some biases in the large-scale fields that exist in the reanalysis data. This bias reduction is attributed to the improved depiction of physical processes within the RSM. This finding indicates that one should take special care in the interpretation and validation of simulated results against the analyzed data. Evaluation of the RSM simulated precipitation for the winter and summer cases generally agrees with results obtained from previous studies. For instance, the skill for simulated precipitation in the winter cases exceeds that of the summer cases by a factor of 2. Comparison of simulated precipitation with observations reveals the 3-month-long RSM simulated precipitation to be more skillful than that obtained from the reanalysis data (the 6-h forecast from the data assimilation system). In addition to seasonal variations in precipitation, daily variation in the simulated precipitation is quite good. However, detailed analysis points to the need for further RSM development, particularly in physics. In the summer cases the grid-resolvable precipitation physics simulate excessive precipitation over the northern United States. A more serious problem is found in the diurnal cycle of the simulation precipitation, in that the model initiates convection too early. Despite these deficiencies, it is concluded that the NCEP RSM is a very useful tool for regional climate studies.