Abstract
Maize (Zea mays) is one of the world’s most important grains that serve as staple food for billions of humans and feed for livestock. The cultivation of maize in South Africa contributes the most in the grain output of the African continent and one of the major contributors at the global level. The cultivation of maize is sensitive to climatic conditions which have focused its production around the famous “Maize Triangle” of South Africa which comprises of sections from three provinces namely Gauteng, North West and Free State with the Free State being the major producer. The production in this region like elsewhere in the world has been faced with challenges of various natures including socio-economic factors (such as high demand), agronomic (nutrients) and climatic (water availability) that does tamper with outputs. The socio-economic factors have sparked increased development of cultivars to withstand changing climatic conditions and maximizing yields. The estimation of yields well ahead of actual grain harvest is therefore relevant for a full comprehension of decision makers on import and export decisions. Monitoring of nutrient status such as nitrogen and water in the crops during the growth cycle is also a very important management aspect for intervention techniques to correct negative situations that could hinder productivity. All of these aspects are being studied in South Africa but through traditional or manual methods that entail extensive, labour-intensive, time-consuming and costly field visits and laboratory analyses. Remotely sensed data and their associated analytical techniques have offered complementary measures that are less costly, covering larger spatial extents at almost near real-time for identifying crop types in the field, estimating yields well in advance before actual harvest and monitoring nutrient status in crops generally. The continuous improvement in the technology and its techniques of information extraction have evolved over the years for both multispectral and hyperspectral data. Therefore, this study aimed to investigate the potential of using remotely sensed data from both multispectral and hyperspectral sensors at ground and space level to discriminate between cultivars, monitor nutrient concentrations and estimate grain yields in maize crops under field conditions.
The potential of hyperspectral data through in-situ measurements on maize leaves was tested in discriminating and classifying eight cultivars grown under field conditions, and monitoring water and nutrient (nitrogen) status. The data collected with the PSR-3500 series handheld spectroradiometer were processed successfully through the random forest...
Ph.D. (Environmental Management)