THE RELATIONSHIPS BETWEEN TROPICAL PACIFIC AND ATLANTIC SST AND NORTHEAST BRAZIL MONTHLY PRECIPITATION

The monthly patterns of northeast Brazil (NEB) precipitation are analy zed in relation to sea surface temperature (SST) in the tropical Pacif ic and Atlantic Oceans, using singular value decomposition. It is foun d that the relationships between precipitation and SST in both basins vary considerably throughout the rainy season (February-May). In Janua ry, equatorial Pacific SST is weakly correlated with precipitation in small areas of southern NEB, but Atlantic SST shows no significant cor relation with regional precipitation. In February, Pacific SST is not well related to precipitation, but south equatorial Atlantic SST is po sitively correlated with precipitation over the northern Nordeste, the latter most likely reflecting an anomalously early (or late) southwar d migration of the ITCZ precipitation zone. During March, equatorial P acific SST is negatively correlated with Nordeste precipitation, but n o consistent relationship between precipitation and Atlantic SST is fo und. Atlantic SST-precipitation correlations for April and May are the strongest found among all months or either ocean. Precipitation in th e Nordeste is positively correlated with SST in the south tropical Atl antic and negatively correlated with SST in the north tropical Atlanti c. These relationships are strong enough to determine the structure of the seasonal mean SST-precipitation correlations, even though the cor responding patterns for the earlier months of the season are quite dif ferent. Pacific SST-precipitation correlations for April and May are s imilar to those for March. Extreme wet (dry) years for the Nordeste oc cur when both Pacific and Atlantic SST patterns for April and May occu r simultaneously. A separate analysis reinforces previous findings in showing that SST in the tropical Pacific and the northern tropical Atl antic are positively correlated and that tropical Pacific-south Atlant ic correlations are negligible. Time-lagged analyses show the potentia l for forecasting either seasonal mean or monthly precipitation patter ns with some degree of skill. In some instances, individual monthly me an SST versus seasonal mean (February-May) precipitation relationships differ considerably from the corresponding monthly SST versus monthly precipitation relationships. It is argued that the seasonal mean rela tionships result from the relatively strong monthly relationships towa rd the end of the season, combined with the considerable persistence o f SST in both oceans.