A Markov chain model for cotton yield forecasting

Abstract

Using the Markov chain theory, the cotton growth characteristics at a given period were partitioned into states; these crop states were related to yield at harvest period. The Markov chain model enables one to predict the expected distribution of cotton yields from the state of the crop at an early growth period. The predictive properties of the Markov chain model were compared with the multiple regression model, which is more commonly used for yield forecasting. In addition to yield forecasting, a Markov chain model was constructed to forecast the number of open bolls per area on the basis of the sequence in cotton fruiting from squares to open bolls.