Assessment of moisture transport and convergence fields in CMIP3 and CMIP5 Global Climate Models in South America

GULIZIA, CARLA; CAMILLONI, INÉS

Dubrovnik

Workshop; Workshop on CMIP5 Model Analysis and Scientific Plans for CMIP6; 2015

EMBRACE/WCRP

In a previous study we have seen that during australsummer though WCRP/CMIP5 Global Climate Models (GCMs) still underestimateprecipitation mainly over southeastern South America, they did so to a lesser extent thanthose of CMIP3, and the dispersion among the new generation of models was smaller than in theprevious one. Two sets of 5 GCMs from each intercomparison project have been previouslyselected based on their reasonable representation of precipitation over South America. Theaim of this study is to assess if the improvements found in CMIP5 models to simulate summerprecipitation are due to a better representation of moisture transport and convergenceover South America. On the other hand, as CMIP5 models still present some difficulties insimulating summer precipitation, another objective is to evaluate if the deficienciesin simulating precipitation can be at least partially explain by an inadequate representation ofthe atmospheric circulation patterns. Consequently, principal component analysis is appliedto austral summer monthly fields for the 1960-1999 period in order to identify theprincipal patterns of observed vertically integrated moisture transport and evaluate the ability of bothsets of GCMs, individually and/or the ensembles, to represent them. Observedspatial patterns are calculated based on NCEP/NCAR reanalysis data. Moreover, we analyzed ifthe GCMs are able to represent the accumulated precipitation anomalies fields and theconvergence and moisture transport associated only with the second and third principalmodes. The comparison of both ensembles' performance showedno significant changes in the skill of the art of CMIP5 regarding CMIP3 models inrepresenting the main moisture transport patterns. Therefore, the better representation ofsummer precipitation in South America should be due to an improvement in simulating otherprocesses/mechanisms not related to moisture transport and convergence. Besides, asmoisture transport from tropics to extratropics is a key climate feature in South America and stronglydetermines both the spatial pattern and the sign of rainfall anomalies, itsrelatively erroneous representation explains at least partially the deficiencies of GCMs to estimateregional precipitation accurately. Thus, this analysis suggests a possible pathway to improvemodel rainfall representation in South America. Finally, we aim to advance in the understanding of therole of summer moisture convergence as a possible mechanism to explain precipitationprojections. For this purpose, the same two subsets of 5 GCMs each were selected. The A1B scenario(for CMIP3) and the medium-low RCP4.5 (for CMIP5) were used to prepare climate changescenarios for the time slices 2020-2029, 2040-2049 and 2070-2079 with respect 1990-1999.Summer precipitation changes and future convergence projections were analyzed for threeparticular sub-regions within the continent. Results indicate that projected changes in moisture convergencecould be attributed as one of the main mechanisms to explain precipitationfuture scenarios, especially in the last generation of models. In this sense, althoughthis study did not assess the uncertainties associated with climate projections, it suggests a firstattempt to understand the role of one of the mechanisms that explainpart of the precipitation, such as moisture convergence.