Files
Abstract
Increasing tourism and the local populations have led to an increase in coastal development and expansion of infrastructure in Belize. This type of rapid land alteration directly affects the amount of eroded sediments and nutrients reaching coastal ecosystems. An increase in suspended sediments and nutrients in the coastal system will have negative effects on the coastal habitats by altering primary production processes due to eutrophication and acidification events.To assist in the mitigation of negative impacts on coastal habitats and understand how land-based changes impact the coastal system, it is crucial to be able to accurately monitor and predict water quality parameters such as Total Suspended Sediments (TSS) and Nitrogen levels. In this study, water quality models were created using Sentinel-2 satellite data combined with field data from Belize Barrier Reef System. Machine learning algorithms were implemented to account for the complexity of coastal ecosystems. An ensemble model consisting of Extreme Gradient Boosting (XGBoost), Random Forest (RF), and Neural Network algorithms was created to estimate suspended sediment concentration. The nutrient water quality model estimating nitrate levels was created using a Deep Neural Network (DNN) algorithm. The study found that out of the three models, neural network algorithms were able to estimate TSS concentrations with an R2 of 0.89. The nitrate algorithm was able to predict concentrations with an R2 of 0.68.