USING MACHINE LEARNING AS A POWERFUL TOOL FOR FLOOD FORECASTING & WARNING SYSTEM IN SOUTH SUDAN

Abstract

This dissertation presents a comprehensive review of flood forecasting and early warning systems developed for South Sudan, focusing on the Juba-Malakal areas. It aims to show that flood preparedness and response strategies have been improved by using machine learning models for accurate flood forecasting and the distribution of timely warnings for vulnerable communities. With the consideration for the use of historic rainfall and river discharge to assess the performance of machine learning models, (K-nearest neighbor (KNN), logistic regression (LR), Support Vector Machine (SVC), Disetion Tree (DT), and Random Forest (RF)). For the prediction of the accuracy of each model, we evaluated using performance metrics such as Mean Square Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). We will select the best model performance among those five models for future flood prediction in south sudan. We will use BCC data for the Juaba area and CM6 data for the Malakal area for future predictions. Through accurate evaluation and comparison of these models, the study aims to determine the most accurate and reliable model for predicting future flood appearance in the Juba and Malakal areas. The selected model will then serve as the cornerstone for designing and implementing an early warning system, which will leverage real-time data acquisition, advanced predictive analytics, and communication technologies to provide timely alerts to relevant stakeholders. Moreover, developing the early warning system will involve collaboration with local authorities, community leaders, and other stakeholders to ensure its effectiveness and relevance to the unique needs of the area’s communities. The system will be designed to be adaptable and scalable, capable of accommodating developed environmental conditions and emerging threats. Implementing a tailored flood forecasting and early warning system can significantly mitigate the impact of floods, improve preparation and response efforts, and enhance the resilience of communities in the Juba and Malakal areas of South Sudan.