Browsing by Author "Kamteni, Yola"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Assessing the spatial variability of neglected and underutilized crop species (NUS) leaf and canopy chlorophyll content in KwaZulu-Natal smallholder farms using unmanned aerial Vehicle (UAV)-based high-throughput phenotyping
    (University of thw Western Cape, 2025) Kamteni, Yola
    Assessing the variability of crop chlorophyll content as an indicator of productivity is essential for optimising the production of Neglected and Underutilized crop Species (NUS) crops like as sweet potato and taro as well as establishing them among mainstream food crops. These NUS present a viable solution to address food and nutritional deficiencies in marginalised communities. Recent advancements in precision agriculture, particularly the use of drones outfitted with high-resolution sensors, have been demonstrated to offer near real-time, spatially explicit data that are invaluable for accurately monitoring and assessing crop growth dynamics at both farm and plot scales. The combined use of UAV-borne remote sensing techniques offers a platform for comprehensively understanding NUS crop productivity characteristics, which can guide operational decisions related to crop health, enabling timely remedial actions and optimising productivity. Hence, the purpose of this research was to assess the usefulness of data obtained from drones and remotely sensed data in mapping the leaf and canopy chlorophyll content of taro and sweet potato crops as a proxy for productivity. The first objective systematically reviewed existing literature on the use of earth observation data in characterising NUS productivity elements on smallholder croplands. The second objectivesought to predict the leaf chlorophyll content (LCC) of taro and sweet potato crops using UAV- derived data while comparatively assessing the accuracy of Random Forest (RF), Linear Regression, and Neural Network regressions in estimating chlorophyll across these NUS crops. The third objective sought to estimate the canopy chlorophyll content (CCC) of taro and sweet potato crops using UAV remotely sensed data. It also compared the prediction accuracies of LCC and CCC of the NUS based on RF. The findings of the review showed that very few published studies have focused on estimating and assessing both foliar and canopy chlorophyll content variability as an indicator of plant growth in NUS, particularly within smallholder contexts. Relative to the second objective, the findings revealed that the best machine algorithm for predicting the Leaf Chlorophyll Content (LCC) is the RF regression ensemble. Specifically, LCC prediction for sweet potato in the late vegetative growth phase showed limited accuracy, with an R2 of 0.23, a RMSE of 13.6 μmol m-2 and a RRMSE of11% based on SR1, CIRE ,The https://uwcscholar.uwc.ac.za/homeII chlorophyll content of the sweet potato crop was noted to be relatively higher than that of taro throughout the phenotyping stages. Using the RF algorithm, results showed that CCC can be accurately predicted during the mid-vegetative growth stage for both sweet potato and taro. For sweet potato, the prediction achieved an RMSE of 9.2% and R2 = 0.91 μmol m-2 with the most important variables being Cirededge, RED, NDV Irededge, Rededge, CIRE, and CI green. For taro the prediction yielded an RMSE of 14.5% and R2 = 0.96 μmol m-2 with CIrededge, GREEN, NIR, CIRE, and RGR as the most effective variables, ranked by significance. Additionally, the CCC estimation accuracies were significantly higher that the LCC estimation. This suggested that the CCC of NUS may be optimally estimated when compared to the LCC of sweet potato and taro throughout the development period. Overall, the results of this research imply that sweet potato and taro chlorophyll content can be optimally estimated using RF regression ensemble and UAV spectral variables across the growing season. This emphasizes how urgently UAV technology is needed to enhance the assessment of chlorophyll content and ultimately crop monitoring, offering insightful data for sustainable farming methods in food-insecure regions.