Estimating maize grain yield from crop growth stages using remote sensing and GIS in the Free State Province, South Africa

Early yield prediction of a maize crop is important for planning and policy decisions. Many countries, including South Africa use the conventional techniques of data collection for maize crop monitoring and yield estimation which are based on ground-based visits and reports. These methods are subjective, very costly and time consuming. Empirical models have been developed using weather data. These are also associated with a number of problems due to the limited spatial distribution of weather stations. Efforts are being made to improve the accuracy and timeliness of yield prediction methods. With the launching of satellites, satellite data are being used for maize crop monitoring and yield prediction. Many studies have revealed that there is a correlation between remotely sensed data (vegetation indices) and crop yields. The satellite based approaches are less expensive, save time, data acquisition covers large areas and can be used to estimate maize grain yields before harvest. This study applied Landsat 8 satellite based vegetation indices, Normalized Difference Vegetation Index (NDVI), Soil Adjusted Vegetation Index (SAVI) and Moisture Stress Index (MSI) to predict maize crop yield. These vegetation indices were derived at different growth stages. The investigation was carried out in the Kopanong Local Municipality of the Free State Province, South Africa. Ground-based data (actual harvested maize yields) was collected from Department of Agriculture, Forestry and Fisheries (DAFF). Satellite images were acquired from Geoterra Image (Pty) Ltd and weather data was from the South African Weather Service (SAWS). Multilinear regression approaches were used to relate yields to the remotely sensed indices and meteorological data was used during the development of yield estimation models. The results showed that there are significant correlations between remotely sensed vegetation indices and maize grain yield; up to 63 percent maize yield was predicted from vegetation indices. The study also revealed that NDVI and SAVI are better yield predictors at reproductive growth stages of maize and MSI is a better index to estimate maize yield at both vegetative and reproductive growth stages. The results obtained in this study indicated that maize grain yields can be estimated using satellite indices at different maize growth stages