Diagnosis And Research Of Real Predictability Of Model T213 Based On The Activity Of Subtropicalanticyclone In East Asia

In order to checkout the performance of model T213 and the quality of its products, and understand its predictability in East Asia, it calculates standard error of the given five fields in five different layers with the model's historical data from July to September in 2004. It also initially discusses the characteristic and rule of the forecast skill change. In addition, it analyses the synoptic situation of the unstable prediction cases that 24 hours' forecast skill is high but 48 hours' is quite low in the place of all cases, and researches their reasons. At last, it analyses the relationship between the changes of standard error and the movements of subtropicalanticyclone.Results show that: (1) The standard error of different elements in different layers becomes worse along with forecast time lengthening. And the changing scope of standard error of 48 hours' prediction is larger than 24 hours' prediction, showing 48 hours' forecast skill is more unstable. But in every layer of every element, the standard error curves of 24 hours' prediction and 48 hours' prediction have similar varying trend from July to September in 2004. And majority of the cases display that its 24 hours' forecast skill and 48 hours' are consistent in the place of all cases. (2) There are still minority cases that 24 hours' prediction are satisfying but 48 hours' prediction are not very ideal. The main reason of the above instance is that 24 hours' synoptic situation change of these cases are slight, but 48 hours' are large even sudden. (3) In 500hPa height field, the peak values and vale values of the standard error curve of 24 hours' prediction and the movement of subtropicalanticyclone have some relevance from July to September in 2004.