Preharvest forecasting of castor yield on the basis of biometrical variables

Abstract

To suggest most suitable pre-harvest forecasting model for Banaskantha district of Gujarat state, two talukas (i.e., Palanpur and Vadgam) of Banaskantha district were selected randomly. Five villages were also randomly selected from each of the two selected talukas. The fields of ten castor grower farmers were randomly selected from each of the selected villages. Biometrical variables were recorded from each of the randomly selected 100 samples (fields of farmers) during the years 2007-08 and 2008-09 by specially trained 10 workers, who had recorded the data from 100 farmer's field. The step-wise regression procedure was employed by using total 21 biometrical variables and two dummy variables. It was revealed that for all the three models (120, 150 and 180 DAS) fitted in this approach, the periods (150 and 180 DAS) were found suitable for predicting castor yield in Banaskantha district. The model for 150 DAS was found more suitable because earliest forecasting is always better. Thus, precised castor yield forecast by applying 150 DAS model is possible two months before actual harvest by using biometrical variables.