Summary: | This paper presents a contribution on the climate modelling studies with emphasis on seasonal rainfall variability in eastern Amazonia, during the austral summer and autumn seasons (DJF and MAM). Based on RegCM3 regional climate simulations for a 26 years period (1982/83 to 2007/08) using high resolution domain scale (30 km) and two different convection schemes (Grell and MIT), it was investigated the model performance to simulate the regional pluviometric distribution in eastern Amazon, with reference to a new observational data base containing regional aspects extracted from a dense rain gauge station network. The quantitative analysis showed that RegCM3 presents systematic errors, especially those related to the dry bias in the Amapá and north/northeast of Pará using both schemes Grell and MIT, which indicate that the model does not reproduce ITCZ characteristics over equatorial Atlantic. The simulations using MIT also indicated wet bias in the southwest/south/southeast of Pará and north of Tocantins. Moreover, through composites technique, it was also investigated RegCM3 response to reproduce the anomalous spatial rainfall patterns in association with ENSO episodes and interhemispheric SSTa gradient phases across the intertropical Atlantic. The results showed that the model represented realistically well the spatial pattern related to the rainfall anomalies above (below) than normal in most of eastern Amazonia, during the known favourable scenarios, i.e., La Niña and south Atlantic SSTa gradient (unfavourable, i.e., El Niño and north Atlantic SSTa gradient). O presente trabalho apresenta uma contribuição aos estudos de modelagem climática com ênfase na variabilidade pluviométrica sazonal da Amazônia oriental, durante as estações de verão e outono (DJF e MAM). Baseado nos resultados das simulações regionais do RegCM3 para um período de 26 anos (1982/83 a 2007/08) e usando domínio em alta resolução espacial (30 Km) e dois diferentes esquemas de convecção (Grell e MIT), foi investigado o desempenho do modelo ...
|