Comparative evaluation of climatic data observation methods in Ibadan, Nigeria

Abstract

Sustainability of agriculture is dependent upon availability of reliable climatic data for planning. Daily and monthly data were obtained for Ibadan, Nigeria from International Institute of Tropical Agriculture (IITA) automatic weather station and Nigerian Meteorological Agency (NMA) manual weather station Two methods were employed to analyse the data. First method (M1) used daily data to generate sets of linear equations. Each equation represents linear relationship between the climate parameter measured at station X (manual) and that same measured at station Y (automatic}. Second method (M2) used monthly values to analyse the data. Accuracy of regression method was analysed by calculating Error Variance (EV) between manual and automatic stations. Errors associated with deviation-based statistics (RMSE) are generally higher than regression-based statistics (EV) for all climate parameters considered. Introduced deviation and correlation based statistics of Mean Squared Deviation (MSD) and its components do not explicitly eliminate error introduced from linearity assumption in regression analysis. Hence, discrepancies have not been adequately explained, but with r < 0.50 in all the climate parameters, the model is weakly correlated with measurement. Analyses have shown that manually observed climate data should not be substituted with automatically observed climate data without correcting data bias/errors prior to usage