Inspection of the bias plot does show some important differences between the simulated summer precipitation field and observations. CONTROL-EC4 underestimates precipitation over the central region of Amazonia in northern and eastern Brazil, by as much as 6mm/day, while CONTROL-E40 shows more limited decreases there. Similar differences have been detected dasatinib IC50 in various RCMs/GCMs, such as NASA GISS, CPTEC/COLA, MM5, and HadRM3P (AM10 and references therein), among others. The negative bias in AM10 (their Figure 2) spans an even larger area over Brazil and part of northern Bolivia. Marengo et al. [23, 24] suggested that the radiation parameterization and/or land-surface processes could be associated with such underestimates, and possible effects of local dynamics forcings such as dry or wet soil may be dominant over the large-scale SST forcing.

Given that climate, soil, and vegetation interact on many different temporal and spatial scales, it is not simple to reproduce such mechanisms and the resulting errors will remain present both in GCMs and RCMs. Another bias during summer, also present in other GCMs and RCMs is a precipitation overestimate of about 4mmday?1 in Paraguayan and Bolivian Chaco as well as neighboring parts of Brazil. Even larger differences are observed over the Andes, in particular over their eastern slopes, with overestimates up to 6mmday?1. A similar behavior is found for CONTROL-E40 with enhanced precipitation over southern Brazil. Comparison with AM10 shows that the current simulation has a somewhat larger overestimate in the Paraguayan and Bolivian Chaco, but excess precipitation over the Andes does not extend as far south.

There is no excess precipitation over Uruguay and Argentina’s Mesopotamia (between the Paran�� and Uruguay Rivers) and Chaco.It could be that the inadequate circulation simulation along the Altiplano and the Andes, with a tendency to a more perpendicular flow towards the orographic barriers, together with a more northeasterly flow, also displaced closer to the Andes, can enhance moisture advection there and lead to such precipitation enhancements, in agreement with Da Rocha et al. [12].During the winter (JJA, Figures 4(c) and 5(c)) dry season, CONTROL-EC4 and CONTROL-E40 runs appear to reproduce all the main features of CRU. Main differences in CONTROL-EC4 arise in southern Brazil and Uruguay, where precipitation is underestimated. Solman et al. [25] obtained a similar result with RCM MM5. Larger precipitation underestimates are also observed along the northwestern edge of the domain in the Peruvian Amazonia and along the northern edge, over Brazil, in agreement with AM10. Similar results, Anacetrapib if somewhat less extended, are obtained in CONTROL-E40.