Maize crop yield estimation with remote sensing and empirical models

Crop yield estimation are topics of interest in Latin-American countries, for farmers and government officers responsible of managing agricultural national policies. Besides, modern remote sensing methodologies to obtain these predictions represent important steps towards attaining the goals of precision agriculture for the 21st century. Digital data from satellite images analyzed jointly with crop modelling parameters provide information that enables crop yield estimation. The objective of this study was the estimation of yield and total volume of maize production using Spot-5 satellite images and empirical models. These models expressed a) yield (Y) as a function of LAI, and b) yield as a function of NDVI. To determine the efficiency degree of the calculated predictions at the flowering stage of the crop, yield sampling was done at the physiological maturity stage in pilot plots. Regarding yield prediction in the flowering stage, the models Y = f (LAI) reported a value of 5.96 ton.ha-1 and the model Y = f (NDVI) a value of 5.04 ton.ha-1 was obtained. These data represent 114% and 97% respectively of the true yield recorded on the field. The models are specific to the maize crop and the cultivated plots location, and that the forecasts can be acceptably accurate provided the sown areas are precisely determined.