Multiple seasons spatialy distributed maize yield and soil properties data for crop modeling applications in precision agriculture

This data was collected from a data-intensive farm management (DIFM) maize trial in Hennenman, Free State South Africa, from a private farm that allowed the research to be done and data used for academic purposes. The yield data were collected using a calibrated yield-monitoring system mounted on the combine harvester. Soil distribution data were collected by a commercial fertilizer company (Omnia) for soil analysis, which was conducted onsite before the start of each planting season. Normalised difference vegetation index (NDVI) values were calculated using Sentinel 2A images with a 10 m resolution in QGIS. Daily weather data [maximum and minimum temperature (°C), precipitation (mm), and solar radiation (MJ m−2)] were obtained from the Agricultural Research Council (ARC) Henneman automatic weather station (27° 54' 22.824" S, 27° 5' 4.848" E 1431 m.a.s.l), Free State, South Africa. Data on yields from many small plot treatment observations were linked to yield-influencing attributes, including seeding and urea rates, soil physical and chemical properties, and remotely sensed normalised difference vegetative index (NDVI), and were used to train and test four ML model [multiple linear regression(MLR), multilayer perceptron (MLP), decision tree (DT) and random forest (RF)] for spatial yield predictions. This data was also used with infield crop growth and soil moisture measurements of selected treatments in season 2021/22 to calibrate the Decision Support System for Agrotechnology Transfer (DSSAT) model for within-field yield predictions.