CR 38:225-236 (2009)  -  DOI: https://doi.org/10.3354/cr00786

ABSTRACT: The statistical relationship between sea-surface temperature variability over tropical oceans and the 1952 to 2000 crops of maize and beans in the state of Ceará, Brazil, was investigated. This region mainly receives precipitation from February to May. The climatic indices used were the Niño-3 for the Pacific and the dipole index for the Atlantic. In a first series of diagnostic analyses, the climatic precursor variables were examined from November to January (Years 0 to 1) for the Niño-3 index and from February to April (Year 1) for the Atlantic dipole components. The agronomic variables were the planted area, annual production, yield (which showed a negative linear trend during the study period), price, and aggregate value (price/yield) for Year 1. A more robust statistical weight was obtained with the Atlantic dipole. The Pacific weight added to, or decreased, the Atlantic effect, depending on the sign of the index. Ocean climatic conditions that generate drought episodes in the Nordeste region of Brazil are generally associated with bad harvests. However, the reverse (i.e. wet weather episodes) does not necessarily produce good harvests. In a second series of prognostic analyses, a statistical model used the ocean climatic information available before the rainy season to forecast the yields of maize and beans. The model gave some relevant data in terms of yield tendencies, and was more sensitive when the climatic conditions were neutral or dry. Remarkably good yield forecasts were noted at the end of the study period. Harvests depend on complex combinations of multiple climatic criteria related, not only to quantitative, but also qualitative factors connected with the rainy season.

KEY WORDS: Climate variability · Northeast Brazil (‘Nordeste’) · Climatic impact · Rain-fed agriculture