Abstract:
The objective of this research was to analyze flood trends in the lower Chao Phraya River in 10 provinces, Phra Nakhon Si Ayutthaya, Nonthaburi, Pathum Thani, Lop Buri, Chai Nat, Sing Buri, Uthai Thani, Saraburi, Suphan Buri, and Ang Thong. The flooding problem in these areas was a catastrophic impact that causes damage to the country, human life, property, agriculture, etc. Therefore, we proposed the flood forecasting system which applied Geographic Information Systems (GIS) methods, including Machine Learning algorithms in the prediction process and used the ArcGIS Pro for visualizing the results in mobile applications. We conducted the experiments by using six factors: elevation, water content in dams, rainfall, land use, soil type, and flood history. We compared the prediction accuracy of three algorithms: Model Logistic Regression (0.95 accuracy), Model Random Forest (0.99 accuracy) and Model Long Short-Term Memory (0.42 accuracy). According to the results in the lower Chao Phraya River flood forecasting system, we can predict the flood in the next three days by the Model Random Forest to analyze flooding in our system with high accuracy and help reduce the damage that will occur to people in those areas.