Mudança climática no nordeste do Brasil, Amazônia e Bacia do Prata: avaliação dos modelos do IPCC e cenários para o século XXI

Abstract

The global models of CMPI3 to the fourth report of Intergovernmental Panel on Climate Change (IPCC-AR4) are evaluated for the Northeast Brazilian region, Amazon region and La Plata basin for the representation of seasonality and multiannual variability of precipitation for the period 1901 to 1999. This evaluation is performed using data from the Climatic Research Unit (CRU) and the 20th Century Reanalysis V2 do National Oceanic and Atmospheric Administration (NOAA). For the seasonality study is proposed to create a performance index based on measures of correlation and mean square error to the evaluation and model classification. For the multiannual variability, is applied the wavelet transform to the observed data, and then models are evaluated and ranked based on measures of spectral correlation and Euclidean distance between the observed variance and models variance in each band. Then, models are classified following the general classification, for the Northeast the CSIRO.MK3.0.RUN2 model; for La Plata basin the CCCMA.CGCM3.1.RUN2 model; for the Amazon region the GISS.MODEL.E.R.RUN6 model those which best represented the rainfall in the twentieth century. At A1B scenario, is made an analysis of seasonal forecast and multiannual trends of better placed models in the seasonal and general classification. For the seasonal projection, analysis consist in the anomalies precipitation calculations on all models for the periods 2010-2039, 2040-2069 and 2070-2099 during the period 1901- 1999. For the multiannual trends for the period 2010 to 2099, annual observed data are standardized, based on the parameters of the scenario 20C3M. Then is done an analysis with average moving, median moving of 10 years, regression linear, Mann Kendall Sen and Wavelet transform. In wavelets trends the MRI.CGCM2.3.2A.RUN3 model does not indicate trend, the GISS.MODEL.E.R.RUN2 model indicates an increasing trend from 2030, the UKMO.HADCM3.RUN1 model shows a declining trend by 2050 in the Northeast. In La Plata basin, the CCCMA.CGCM3.1.RUN2 indicates increasing trend throughout the twenty first century, the CCCMA.CGCM3.1.RUN1 shows an increase trend from 2030 and the NCAR.PCM1.RUN1 model shows no trend. In the Amazon region, the CSIRO.MK3.0.RUN1, CCCMA.CGCM3.1.RUN1, INGV.ECHAM4.RUN1 models do not indicate any trend over the twenty-first century.