Seasonal Climate Prediction over Northeast Brazil
Northeast Brazil experiences a rainy season that centers on the months of March through May. This seasonality can be explained by the southward migration across the Equator of an east-west band of tropical rainfall, called the Inter-Tropical Convergence Zone (ITCZ). During March and April the ITCZ reaches its southernmost latitude in response to the seasonal warming of the adjacent Atlantic Ocean. Heating of the lower atmosphere by the warmest ocean water initiates rising air currents that promote the formation of tall tropical clouds and heavy downpours. Accordingly, the southward movement of the ITCZ, and in turn the timing and intensity of precipitation over northeast Brazil during any given year, are profoundly affected by departures of Atlantic Ocean temperatures from their usual seasonal changes.
The strong link between northeast Brazil's seasonal precipitation and Atlantic Ocean water temperatures makes this region a promising candidate for seasonal climate prediction. Fortunately, vast pools of unusually warm or cold ocean temperatures (anomalies) often persist for months at a time. One way of predicting ocean temperatures during the coming season is therefore to assume that initial departures from the "normal" seasonal cycle will remain the same over the coming months.
The climatic consequences of unusually warm or cold ocean temperatures can be estimated by computer-based climate simulation models. The Regional Climate Model (RCM) currently in use at the NASA Goddard Institute and Columbia University Center for Climate Systems Research is particularly useful for studying climate anomalies over northeast Brazil because the spacing of its computational grid is only 50 km between adjacent elements. RCM simulations cover limited areas and consequently require a constant stream of climate information at the geographic edges of the domain. Our experimental seasonal predictions for northeast Brazil were based on simulations that covered most of South America and parts of the adjacent Atlantic and Pacific Oceans. In the prediction experiment for March-May 1997, these "lateral boundary data" were supplied from a global-domain climate simulation model which computes variables on a much coarser computational grid with 400 km spacing. In effect, the RCM interprets such coarser resolution predictions by giving more detailed structure to the spatial distributions of climate variables, such as the precipitation patterns associated with the ITCZ. Ocean surface temperatures used in the simulation were based on the assumption that February 1997 departures from their multi-year averages would remain the same during the subsequent three months.
The figure compares the RCM predicted March-May 1997 average precipitation rates (B) to our best estimate of actual rainfall during that period (A). The ITCZ is represented by the east-west swath of heavy rainfall that crosses the coast of Brazil near the Equator. Actual March-May 1997 average precipitation rates near the center of this swath were estimated at 6-12 mm/day while the RCM predicted rates that were slightly lower, but approximately aligned along the correct latitude. The corresponding prediction of the global model (C) features an ITCZ precipitation rate maximum that does not conform as well to its observed orientation and location.
Druyan, L., M. Fulakeza, P. Lonergan and M. Saloum 2001. A regional model study of synoptic features over West Africa. M. Weather Rev. 129, 1564-1577