Framework for IOT Based Real-Time Monitoring System of Rainfall Water Level for Flood Prediction Using LSTM Network

• Published in: 2023 3rd International Conference on Pervasive Computing and Social Networking (ICPCSN) pp. 1321 - 1326
• Main Authors: William, P., Oyebode, Oluwadare Joshua, Ramu, Gandikota, Lakhanpal, Sorabh, Gupta, Keerat Kumar, Al-Jawahry, Hassan M.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2023
• Subjects: CNN; Data models; Internet of Things (IoT); Long Short-Term Memory (LSTM); Prediction algorithms; Predictive models; Sensor systems; Sensors; Social networking (online); Water level; Weather forecasting
• DOI: 10.1109/ICPCSN58827.2023.00222

Recent floods have consistently resulted in human casualties in addition to harm to the environment and the economy. People are less likely to be aware of oncoming floods since there is not a reliable early warning system. A deep learning algorithm is suggested in this study to forecast groundwater levels. Even with inadequate data, the model can nevertheless successfully fulfil the prediction objective. In order to determine the degree of temporal dependence that exists between groundwater level and meteorological indicators, a hybrid model called CNN-LSTM-ML is developed. This model makes use of a network structure that is a mix of CNN and LSTM networks. Long Short-Term Memory (LSTM) networks are used by the rainfall forecasting model in order to provide predictions about future rainfall and water level values, which may lead to floods. In order to conduct experiments on the principal finding, historical data from two sites with rainfall and one location with streamflow were used. Additional information was acquired from two water level stations and one rainfall station in order to confirm the basic findings. With assessment errors for historical data MAE, RMSE, and MSE of 0.93, 1.7, and 3.025 correspondingly, the forecasting technique that used LSTM demonstrated great "accuracy" of the outcome reaching more than 92% using IoT. According to these results, the system has the potential to be used as a cure that does not involve the construction of new structures in order to minimize the damage caused by urban floods.