Statistical Problems in Crop Forecasting

Abstract

A reliable estimate of a crop yield well before harvest is of considerable importance in formulating policy. Usually, such preharvest estimates of the yield rate of a crop are obtained on the basis of visual observations of crop reporters, which is subjective. An objective method of preharvest forecasting, based on observations on biometrical characters (viz. plant population, plant height, number of leaves, etc.) as well as on weather parameters such as rainfall, temperature and humidity, is discussed in this paper. Several statistical problems are briefly examined, for example, the choice of appropriate variables, sampling design, and prediction model. Empirical studies have shown that the explanatory variables included in these models explain a large part of the variation in yield; however, for further improvement, it is necessary to use biometrical variables in conjunction with weather factors in prediction models. Further, research is needed on weather modeling, including weather forecasting and the use of satellite imagery, as well as on sample design for the collection of data for modeling purposes, and compartmental analysis of plant process models.