Abstract
A global climate prediction system (PCCSM4) was developed based on the Community Climate System Model, version 4.0, developed by the National Center for Atmospheric Research (NCAR), and an initialization scheme was designed by our group. Thirty-year (1981–2010) one-month-lead retrospective summer climate ensemble predictions were carried out and analyzed. The results showed that PCCSM4 can efficiently capture the main characteristics of JJA mean sea surface temperature (SST), sea level pressure (SLP), and precipitation. The prediction skill for SST is high, especially over the central and eastern Pacific where the influence of El Niño-Southern Oscillation (ENSO) is dominant. Temporal correlation coefficients between the predicted Niño3.4 index and observed Niño3.4 index over the 30 years reach 0.7, exceeding the 99% statistical significance level. The prediction of 500-hPa geopotential height, 850-hPa zonal wind and SLP shows greater skill than for precipitation. Overall, the predictability in PCCSM4 is much higher in the tropics than in global terms, or over East Asia. Furthermore, PCCSM4 can simulate the summer climate in typical ENSO years and the interannual variability of the Asian summer monsoon well. These preliminary results suggest that PCCSM4 can be applied to real-time prediction after further testing and improvement.