Understanding of climate influence on crop yields can help in the design of policies to reduce climate-related vulnerability in many parts of the world, including the target of this case study-the state of Ceara, Brazil. The study has examined the relationships between climate variations and corn yields and, in addition, has estimated the potential predictability of corn yields in Ceara drawing on the now well-established seasonal predictability of the region's climate based on prevailing patterns of sea surface temperature (SST), especially in the tropical Atlantic and tropical Pacific Oceans. The relationships between corn yields and climate variables have been explored using observed data for the period of 1952-2001. A linear regression-based corn-yield model was evaluated by comparing the model-simulated yields with the observations using three goodness-of-fit measures: the coefficient of determination, the index of agreement, and the mean absolute error. A comparative performance analysis was carried out on several climate variables to determine the most appropriate climate index for simulating corn yields in Ceara. A weather index was defined to measure the severity of drought and flooding conditions in the growing season for corn. The analysis indicated that the weather index is the best climate parameter for simulating corn yields in Ceara. The observed weather index can explain 56.8% of the variance of the observed corn yields. High potential predictability of the weather index was revealed by the evaluation of an ensemble of 10 runs with the NCEP Regional Spectral Model nested into the ECHAM4.5 atmospheric general circulation model, driven with observed SSTs in each season for the period of 1971-2000. Whereas these runs are based on the actual observed SST pattern in each season, other studies have shown that persistence of SST over several months is sufficient for a true predictive capability. The aim here was to show that the SST-forced component of climate variation does translate into the weather features that are important for crop yields. Indeed, the results demonstrate the striking extent to which the year-to-year changes in SST force local climate characteristics that can specify the year-to-year variations in corn yields. The variance of corn yield explained by the SST-driven model was 49.5%.
141LZTimes Cited:0Cited References Count:58