Assessing cyanobacteria in a small reservoir using unmanned aerial vehicle systems (UAVs): a case study of High Flight Farm Dam.

Abstract

Monitoring water quality, particularly chlorophyll-a (chl-a) concentrations, is critical for managing irrigation water, as excessive chl-a can degrade aquatic ecosystems and reduce water availability. While multispectral satellite-based remote sensing is widely used, its spatial resolution is inadequate for small water bodies, which are crucial to smallholder farmers. Unmanned Aerial Vehicles (UAVs) offer high-resolution, near-real-time data, presenting a promising solution. This thesis investigates UAV-based multispectral imaging for chl-a estimation in small reservoirs through an empirical study in South Africa, supported by a background systematic review of existing literature. The empirical study integrates UAV-based multispectral data from April, June, and July 2024 with in-situ measurements of chl-a, total nitrogen (TN), total phosphorus (TP), and dissolved oxygen (DO). The machine learning models tested include Artificial Neural Networks (ANN), Random Forest (RF), Support Vector Machine (SVM), and Extreme Gradient Boost (XGBoost), K-Nearest Neighbours (KNN), with ANN consistently outperforming the others and achieving the highest R² values across the three sampling periods: 0.949 (April), 0.991 (June), and 0.734 (July). The green, red, and red-edge bands were the most sensitive for chl-a estimation. Seasonal patterns emerged, with high chl-a concentrations in April and June, followed by a decline in July due to reduced water levels. Strong correlations were found between chl-a and nutrient parameters, particularly TP (R² = 0.879) and TN (R² = 0.711) in July. This study highlights the potential of UAV-based remote sensing for high-resolution chl-a monitoring in small water bodies. This study demonstrates the potential of UAV-based remote sensing for accurate, localized, and detailed chl-a monitoring in small water bodies, offering valuable insights for water resource management in smallholder agricultural systems worldwide.