In this study, we evaluate the performance of global climate models for reproducing the present SLP (sea level pressure) field in the CO
2 transient run, which will be used for the prediction of regional climate change. The outputs calculated using the model are projected in ‘PC-space’, with the component scores from the principal components (or EOF) of observed data as the axes, and their biases are evaluated quantitatively.
This evaluation method was applied to the previous runs (CSM125 and ACACIA) using NCAR-CSM. In the ACACIA run, the January SLP field is reproduced realistically and is improved from CSM125, although the winter monsoon is still weak. The reproduced pattern resembled that of December. On the other hand, the July SLP field is not reproduced well; the intensity of the North Pacific High is over estimated and its ridge is shifted to the north, which are both beyond the observed range. Through these analyses, the difference of bias pattern for each year between two runs which have similar mean distribution pattern is evaluated quantitatively.
Then the performances of the models are compared quantitatively to the other global climate models in the world through the application of this method. All the models tend to reproduce the Siberian High weakly, which is a common feature of the present models. For the summer SLP field, all the models show poor performance as well as two NCAR runs. A series of analyses revealed that, outputs from two NCAR runs in winter seem to be usable, with some care, for the prediction of regional climate changes. However, summer results are insuflicient for such use, determined from the viewpoint of the performance of reproducing the present SLP field.
抄録全体を表示