Performance of decadal prediction in Coupled Model Intercomparisson Project Phase 5 (CMIP5) on projecting climate in tropical area

摘要

Skill of BCC-CSM1.1, MPI-ESM-LR and IPSL-CM5A-LR (climate models which are collected together in Coupled Model Intercomparison Project Phase 5) on projecting precipitation, air temperature and sea surface temperature (SST) in tropical region were analyzed during December-January-February (DJF) and June-July-August (JJA) using Empirical Orthogonal Functions (EOF), climatology, bias and Root Mean Square Error (RMSE) parameters. These models were compared with the reanalysis/observation data. The results showed that dominant modes (EOF1) of all models for air temperature were similar with the mode of reanalysis data. In general, EOF1 of all models for precipitation are similar with mode of the observation which is dominated by positive (negative) anomaly in northern (southern) equator. Climatology of precipitation, air temperature and sea surface temperature showed that high values were located in southern equator during DJF and shifted to the north during JJA. The highest air temperature bias was located in Africa (around 6°C during DJF and JJA) for MPI-ESM-LR. In addition all models captured the ITCZ signal. The highest RMSE and bias of precipitation (> 9 and 5 mm/day) were located in western Pacific in BCC-CSM1.1 during DJF and IPSL-CM5A-LR during JJA. BCC-CSM1.1 had higher RMSE (> 9 mm/day) in projecting precipitation in Papua and Borneo, MPI-ESM-LR detected higher bias in Sumatra, Java, Nusa Tenggara, while IPSL-CM5A-LR in Sumatra. The highest RMSE for SST (more than 2°C) was located in the eastern Pacific.