Time-lagged correlations of pre-monsoon precipitation in the Indochina Peninsula confirmed in a large ensemble simulation dataset

Statistically significant time-lagged relationship of pre-monsoon precipitation in Indochina Peninsula (ICP) with large-scale feature over the Pacific and Indian Oceans, which was recently found in observational datasets by the authors, is investigated with a large-ensemble (100 ensembles) simulation database called "Database for Policy Decision-Making for Future Climate Change" (d4PDF). Two different strategies were taken to perform Empirical Orthogonal Function (EOF) analysis on the large-ensemble dataset, namely statistics of the analysis results for each ensemble member separately and analysis on the 6,000-year data (100 times 60 years). d4PDF can reproduce well the climatological characteristics of the observed pre-monsoon precipitation. Significant time-lagged correlations of several climate indices related to sea surface temperature (SST) with the pre-monsoon precipitation in ICP as obtained in the observation are confirmed in d4PDF for both strategies of EOF analyses on each ensemble member and on 6,000-year data. The climate simulation in d4PDF can capture the lag-to-lag variations of the time-lagged correlation patterns very well, though it is weaker. The time-lagged regressions of SST and zonal wind at 850 hPa upon the first principal component (PC1) time-series of the EOF analysis of the pre-monsoon precipitation over ICP are investigated in low latitudes over the Pacific and Indian Oceans. The observed time-lagged regression features are also confirmed in d4PDF dataset; the significant regression areas expand larger and significant time lags become longer than the observation, especially in the case of 6,000-year data, due to enough number of samples. Cluster analysis on the regression maps shows that the obtained groups with larger numbers of ensemble members are closer to the observation than the other groups. The randomness of time-lagged regression feature among the 100 ensemble members is not directly related to the randomness of the SST perturbation introduced in the lower boundary condition for the ensemble simulation of d4PDF.