The utility of very-high resolution unmanned aerial vehicles (UAV) imagery in monitoring the spatial and temporal variations in leaf moisture content of smallholder maize farming systems.

Masters Degree. University of KwaZulu-Natal, Pietermaritzburg.Maize moisture stress, resulting from rainfall variability, is a primary challenge in the production of rain-fed maize farming, especially in water-scarce regions such as southern Africa. Quantifying maize moisture variations throughout the growing season can support agricultural decision-making and prompt the rapid and robust detection of smallholder maize moisture stress. Unmanned Aerial Vehicles (UAVs), equipped with light-weight multispectral sensors, provide spatially explicit near real-time information for determining maize moisture content at farm scale. Therefore, this study evaluated the utility of UAV derived multispectral imagery in estimating maize leaf moisture content indicators on smallholder farming systems throughout the maize growing season. The first objective of the study was to conduct a comparative analysis in order to evaluate the performance of five regression techniques (support vector regression, random forest regression, decision trees regression, artificial neural network regression and the partial least squares regression) in predicting maize water content indicators (i.e. equivalent water thickness (EWT), fuel moisture content (FMC) and specific leaf area (SLA)), and determine the most suitable indicator of smallholder maize water content variability based on multispectral UAV data. The results illustrated that both NIR and red-edge derived spectral variables were critical in characterising maize moisture indicators on smallholder farms. Furthermore, the best models for estimating EWT, FMC and SLA were derived from the random forest regression algorithm with a relative root mean square error (rRMSE) of 3.13%, 1% and 3.48 %, respectively. Additionally, EWT and FMC yielded the highest predictive performance of maize leaf moisture and demonstrated the best correlation with remotely sensed data. The study’s second objective was to evaluate the utility of UAVderived multispectral imagery in estimating the temporal variability of smallholder maize moisture content across the maize growing season using the optimal maize moisture indicators. The findings illustrated that the NIR and red-edge wavelengths were influential in characterising maize moisture variability with the best models for estimating maize EWT and FMC resulting in a rRMSE of 2.27 % and 1%, respectively. Furthermore, the early reproductive stage was the most optimal for accurately estimating maize EWT and FMC using UAVproximal remote sensing. The findings of this study demonstrate the prospects of UAV- derived multispectral data for deriving insightful information on maize moisture availability and overall health conditions. This study serves as fundamental step towards the creation of an early maize moisture stress detection and warning systems, and contributes towards climate change adaptation and resilience of smallholder maize farming