Abstract
The availability of quantitative means of probing anticipated rainfall is essential for the purpose of planning and policy formulation everywhere in the world. However reliable prediction of rainfall remains a challenge over West Africa and Nigeria in particular. This paper predicts sea-sonal rainfall in Nigeria with the hope of increasing the reliability of predicted seasonal rainfall. Seasonal rainfall data (June-September) from 21 Nigerian stations and monthly Extended Re-constructed Sea Surface Temperature (ERSST) of the tropical Oceans spanning 1960-2012 con-stituted the input data for the formulation of prediction equations using regression analysis. Cross-validation was applied with training period of 1960-1985, verification period of 1986-2012 and vice versa. The correlations between the predicted and the observed seasonal rainfall of the training and verification periods of 1960-1985 were all significant at 5 % level except for one station. While for the training and verification periods of 1986-2012, correlations are not gener-ally significant although some were found to be above the significant level of 5%. For the recon-structed seasonal precipitation, correlations of ten stations were found to be significant at 5 %, but at 10 % level, their number increased to 13. The t-test analysis revealed that there is no sig-nificant difference between the means of predicted and observed seasonal rainfall amount for all stations except for only three stations (Osogbo, Calabar and Enugu). This suggests some degree of skill which is also an indication that Sea Surface Temperature (SST) is a good predictor of June, July, August and September (JJAS) seasonal rainfall in Nigeria. Keywords : Rainfall, Prediction, Tropical Ocean, Cross-validation, Sea Surface Temperature
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have