SMIP2 Experiment-based Analysis on the Simulation and Potential Predictability of Asian Summer Monsoon
SHI Hong-Bo,ZHOU Tian-Jun,WAN Hui,WANG Bin,YU Ru-Cong
2008-01-01
Abstract:The ability of simulating Asian summer monsoon(ASM) is evaluated and potential predictability of the ASM on seasonal time scale is analyzed by using the output of five atmosphere general circulation models participating in "Phase 2 of the Seasonal Prediction Model Intercomparison Project"(SMIP2).According to the requirements of SMIP2,seven-month ensemble integrations are performed from 6-10 consecutive days of initial condition at the beginning of February,May,August and November from 1978 to 1999,respectively.The observed sea surface temperature obtained from PCMDI(Program for Climate Model Diagnosis and Intercomparison) is used as boundary forcing.The reanalysis data or model climatologies from AMIP(Atmospheric Model Intercomparison Project) run is used as initial land conditions.In this paper,only summer mean data is used.Results show that these models can approximately reproduce the spatial pattern of Asian summer precipitation,except that the intensity of the precipitation is weakly simulated over the subtropical western Pacific,the marine continent to the south of the Philippines,and the equatorial Indian Ocean to the south of the Bay of Bengal.The empirical orthogonal function(EOF) analysis is used to extract the leading mode of summer precipitation anomaly over East Asia,South Asia and the western Pacific.It is found that the models are capable of simulating the spatial pattern of "north drought and south flood" in East Asia.But the extension and center of the simulated dry and wet regions are different from the observed ones.And there are very weak correlations between the time coefficient for the first EOF mode(EOF1) of the observed East Asian summer precipitation anomaly and the simulated ones except MGO model.All models are able to simulate the positive anomaly of summer precipitation in the south of the India peninsula,while the negative anomaly in the Bay of Bengal is neglected.The time coefficients of the simulated South Asian summer precipitation anomaly EOF1 are well correlated with the observed one.At the same time,the models can well simulate the space-time pattern in the western Pacific except that the anomaly centers simulated by individual models are a little different from the observed.And yet the correlation coefficients are very large between the simulated and observed time coefficients,which exceed 0.8 and pass the significance level of 0.01.The analysis of variance is used to investigate the potential predictability of the ASM.It is showed that the potential predictability of the ASM is very high in the tropics.The prediction skill in the Northern subtropical region is good but lower than that in the tropical region.There is very low predictability in the Southern extratropical region.The prediction skill is lower over the land than over the ocean.Furthermore,the potential predictability is model-dependent.For example,LASG,MGO and MRI models display high potential predictability over various regions for 500-hPa geopotential height.External variability contributes much to the high potential predictability over the tropical region and the Northern subtropical region.The potential predictability of the simulated summer mean zonal wind at 850-hPa exhibits distinct interannual variability.The models have very high prediction skill during strong El Nino years and La Nina years over East Asia,South Asia and the western Pacific.In order to further study the influence of El Nino(La Nina) events on potential predictability,an average over four selected years is made to measure the potential predictability for the El Nino,La Nina and normal cases,respectively.The results reveal that potential predictability is higher for the El Nino,La Nina cases than for the normal case.The extension of high potential predictability is larger in El Nino years than in La Nina years.And the potential predictability center is more westward in El Nino years.Moreover,most models have good prediction skill in El Nino years than in La Nina years over South Asia and the Indian Ocean.