Twentieth century ENSO-related precipitation mean states in twentieth century reanalysis, reconstructed precipitation and CMIP5 models

摘要

El Nio-Southern Oscillation (ENSO)-related precipitation during the entire twentieth century is compared among the twentieth century reanalysis (20CR), a statistically reconstructed precipitation dataset (REC) and 30 Coupled Model Intercomparison Project Phase 5 (CMIP5) models. Empirical orthogonal functions, ENSO-related precipitation composites based on sea surface temperature (SST)-constructed ENSO index and singular value decomposition (SVD) are employed to extract ENSO-related precipitation/SST signals in each dataset. With the background trend being removed in all of the data, our results show that the REC and the 20CR resemble both in their precipitation climatology and ENSO-related precipitation results. The biases in the CMIP5 models precipitation climatology such as dry equator over the Pacific Ocean, "double-intertropical convergence zones (ITCZs)" and overly zonal Southern Pacific convergence zone (SPCZ) are major reasons for lowering spatial correlations with the REC and the 20CR precipitation climatology. Two groups of CMIP5 models are built based on severity of these biases in their precipitation background and the spatial correlations of ENSO-related precipitation with the observations. Compared with the group with more severe biases in its precipitation climatology, the group with smaller biases tends to produce more ENSO-like precipitation patterns, simulate more realistic mean magnitude and seasonal variability of ENSO precipitation signals, as well as generating better ENSO-related SST/precipitation correlation patterns produced in its SVD analysis. The ENSO-related precipitation biases in the CMIP5 models over the western Pacific and Indian Ocean, as well as the equatorial Pacific, are strongly related with their precipitation climatology biases over these regions. The ENSO-related precipitation biases over the off-equator eastern Pacific Ocean are associated with both the "double-ITCZs" biases in the precipitation climatology and the ENSO-related SST biases in the models.