Analysis of a long term Regional Model Simulation

The use of regional climate models in studying atmospheric dynamics is on the increase in southern Africa, where the MM5 regional model has been used for a range of experiments. Such work has led to the start of a 15-year climatology simulation for the model so that more accurate evaluation of the model can be made. This paper will look at some of the results obtained from this experiment. The increasing use of model output necessitates a deeper analysis of the data than is conventionally undertaken. Therefore in addition to comparing time averaged climatic fields from the model to observed station data, the output has been compared to other statistics such as total number of rain days, and indices of wet and dry spell duration. Initial analysis of one year of the long term simulation has shown that the model produces a large positive rainfall anomaly along the eastern escarpment of South Africa. Although positive rainfall anomalies are also present in the interior of the region, these are not as high. The comparison of the number of rain days above 20mm with those of observed data show that although the model produces more rain events, the magnitudes of these events are not high enough to change the actual 80th percentile of the model output drastically. This initial finding has been encouraging and the analysis of the complete long term simulation using these different statistics will be useful in assessing the biases of the model as well as the errors that exist in it. Such analysis will also be of great relevance to the end users of model output.