Previous studies have considered the long term variability of precipitation in Amazonia, either based on the records at individual stations or with several stations with short time series, as well as discharges of the Amazon rivers and its tributaries. In principle, the discharges of the Amazon river, together with the Negro Solimões, Madeiras and Purus rivers, should show the influence of the rainfall accros de basin. In the case of the Solimoes River, the largest affluent of the Amazon, the large volumes of this river may upset the gauging values in lower courses of tributaries, producing with is known as the backwater effect. This indicates that for some basins, the assumption that the discharges/levels measured on large basins represent the hydrometeorological conditions at those sites is no longer safe. In addition, the availablity of worlwide gridded reanalysis (NCEP, ECMWF, NASA-DAO) have allowed assessments of the water vapor transport into Amazonia, and in fact some recent works have shown contradictory results in terms of trends in input moistyure into the Amazon basin. . Thus, rainfall observations are seen as perhaps the only way to infer on the components of the hydrogical cycle of the region, and to validate both studies based on global reanalysis and model results.
Indices of rainfall in northern Amazonia developed by the author in 1991 have been updated and extended in time and spatial coverage. Based on these rainfall indices since 1928 it was observed that: a) interannual variations show that ENSO influences are significant more notably the in the northern drainage basin of the Amazon basin, drier during El Niño than la Niña years, b) interdecadal variations indicate that both the central and western part of the northern basin, show positive trends in rainfall between 1940-62, and a negative trend between 1962-82. A wet period was observed between 1950-76 [wettest in 1960-74], and also coincides with cool sea surface temperatures (SST) in the equatorial Pacific. Dried periods were observed in 1928-42, 1978-84, and 1990-94 [driest in 1979-83]. These interdecadal variations seem more related to an interdecadal mode of variability of the meridional SST gradient (dipole) and atmopsheric circulation over the tropical Atlantic, which also seems applicable to river discharge in the whole basin, as well as rainfall in Northeast Brazil. On the other hand, southern Amazonia shows a short wet period in 1930-40, and upwards trends since 1975, while a drier period was found in 1960-72. This variability is independent of the Pacific SST or ENSO influences, and does not follow variations if rainfall in northern Amazonia..