Large-scale circulation patterns and related rainfall in the Amazon Basin: a neuronal networks approach

This study describes the main circulation patterns (CP) in the Amazonian Basin over the 1975-2002 period and their relationship with rainfall variability. CPs in the Amazonian Basin have been computed for each season from the ERA-40 daily 850 hPa winds using an approach combining artificial neural network (Self Organizing Maps) and Hierarchical Ascendant Classification. A 6 to 8 cluster solutions (depending on the season considered) is shown to yield an integrated view of the complex regional circulation variability. For austral fall, winter and spring the temporal evolution between the different CPs shows a clear tendency to describe a cycle, with southern wind anomalies and their convergence with the trade winds progressing northward from the La Plata Basin to the Amazon Basin. This sequence is strongly related to eastward moving extra tropical perturbations and their incursion toward low latitude that modulate the geopotential and winds over South America and its adjoining oceans. During Austral summer, CPs are less spatially and temporally organized compared to other seasons, principally due to weaker extra tropical perturbations and more frequent shallow low situations. Each of these CPs is shown to be associated with coherent northward moving regional rainfall patterns (both in in situ data and ERA-40 reanalysis) and convective activity. However, our results reveals that precipitation variability is better reproduced by ERA-40 in the southern part of the Amazonian Basin than in the northern part, where rainfall variability is likely to be more constrained by local and subdaily processes (e.g. squall lines) that could be misrepresented in the reanalysis dataset. This analysis clearly illustrates the existing connections between the southern and northern part of the Amazonian Basin in terms of regional circulation/rainfall patterns. The identification of these CPs provide useful information to understand local rainfall variability and could hence be used to better understand the influence of these CPs on the hydrological variability in the Amazonian Basin.