Background
The authors write that "underestimated rainfall over Amazonia was a common problem for the Coupled Model Intercomparison Project phase 3 (CMIP3) models," and, therefore, they state that in their current study they "investigate whether it still exists in the CMIP phase 5 (CMIP5) models."

What was done
Yin et al. evaluated the performance of "eleven CMIP5 models for historical rainfall seasonality over Amazonia by comparing them to the GPCP [Global Precipitation Climatology Project] and CMAP [CPC merged analysis of precipitation, where CPC is rainfall datasets], the ERA-Interim reanalysis product and NOAA/NCDC sea surface temperatures."

What was learned
The three U.S. researchers report that (1) "some models still underestimate rainfall over Amazonia," that (2) "during the dry season, both convective and large-scale precipitation is underestimated in most models," that (3) "during the wet season, large-scale precipitation is still underestimated in most models," that (4) "in some models, overestimates of rainfall are associated with the adjacent tropical and eastern Pacific ITCZs [Intertropical Convergence Zones]," that (5) "during the transition season,[i] low pre-seasonal latent heat, [ii] high sensible flux, and [iii] a weaker influence of cold air incursions contribute to the dry bias," that (6) "about half the models can capture, but overestimate, the influences of teleconnection," that (7) "the majority of the models either overestimate the Atlantic ITCZ or the eastern Pacific ITCZ or both," that (8) "no models realistically represent the observed distribution pattern of rain rates," that (9) the models "generally underestimate large-scale rainfall in all seasons and all four regions " that they studied, that (10) "almost all the models overestimate surface net solar radiation," which leads to (11) "a high bias in surface net radiation," that (12) during Dec, Jan and Feb "the weak westerly wind area, representing the anti-cyclonic center, is overestimated over most of the models," and that (13) "dry biases during the dry and transition seasons still exist in the majority of the models."

What it means
In comparing the CMIP5 models with the CMIP3 models, some progress is noted; but there is still a long, long way to go before climate reality is replicated by computers.