A model of estimation maize yield based on weather, agronomical and satellite data

Abstract

A timely and reliable system of maize yield forecasting well in advance is prime emphasis to farmers and other people who are dependent on cereal crop. The best model was generated using maize field experiment trial which was conducted at Dorbasta union of Gobindagonj upazila, Gaibandha during two consecutive Rabi crops growing season 2018-19 and 2019-20. Randomized Complete Block Design (RCBD) along with five treatments (or varieties) and three replications were considered for maize yield performance. The agronomical and weather parameters and also, satellite data (Landsat 8 OLI) were used for the required maize field experiment. We found that Normalized Difference Vegetation Index (NDVI) was strongly positively correlated with the weather variables in this study. Stepwise regression method was applied for generating best estimated model. Best estimated model (Backward elimination) showed that only five controlled variables which were variety 5 (BHM 13), 1000 grain weight, diameter of cob, plant height and NDVI that were factors to the yield of maize.  The developed maize yield forecast model (ideal model) including agronomical, weather and satellite data give the better results of yield estimation at regional level on the basis of best model criterion. Therefore, the ideal model used in specific region including all types of data that gives more precise result on maize yield or production that should be more significant and reliable in national level. So, the researcher, policymaker can use this maize yield prediction model forty to fifty days earlier of harvesting time.

Bangladesh J. Agril. Res. 48(4): 433-449, December 2023