Comparative Evaluation of Some Crop Yield Prediction Models using Tropical Cowpea Yield-Weather Data

Abstract

A computer program was adopted from the work of Hill et al. (1982) to calibrate and test three of the existing yield prediction models using tropical cowpea yieldÐweather data. The models tested were Hanks Model (first and second versions). Stewart Model (first and second versions) and HallÐButcher Model. Three sets of cowpea yield-water use and weather data were collected. The first one was used for calibration and the other two for validation. The results obtained from the models were compared with field values using correlation coefficient and mean error of deviation. Stewart first model had r-values of 0.93
and 0.74 and mean error of deviation of 0.1 and 0.15. The second model had r-values of 0.92 and 0.86 and mean error of deviation of 0.05 and 0.11. Hanks first model had r-values of 0.94 and 0.89 and average mean error of deviation of 0.03 and 0.10, the second model had r values of 0.93 and .096 and average mean error of deviation of 0.05 and 0.03. The r-values for Hall-Butcher model were 0.76 and 0.96. The mean errors of deviation were 0.14 and 0.03. Generally, Hanks model predicted better than the other models but may still need to be further modified to be able to predict well for tropical conditions.

Keywords: Prediction model, yield prediction, yield parameters, Hanks model, Stewart model, Hall Ð Butcher model, cowpea yield, tropical climatic conditions

Discovery and Innovation Vol. 19 (2) 2007: pp. 89-95