Deep Machine Learning-Based Water Level Prediction Model for Colombo Flood Detention Area

Machine learning has already been proven as a powerful state-of-the-art technique for many non-linear applications, including environmental changes and climate predictions. Wetlands are among some of the most challenging complex ecosystems water level Wetland prediction is vital, wetlands have their own permissible levels. Exceeding these levels can cause flooding other severe damage. On hand, biodiversity threatened by sudden fluctuation Hence, early benefits in mitigating such However, monitoring predicting worldwide limited owing to various constraints. This study presents first-ever application deep machine-learning techniques (deep neural networks) predict an urban wetland Sri Lanka located its capital. Moreover, first time prediction, it investigates two types relationships: traditional relationship between factors, temperature, humidity, wind speed, evaporation, temporal daily Two low load artificial networks (ANNs) were developed employed analyze relationships which feed forward (FFNN) long short-term memory (LSTM) networks, conduct comparison on unbiased common ground. The LSTM outperformed FFNN confirmed that much more robust than relationship. Further, identified interesting accuracy, data volume, ANN type, degree information extraction embedded data. (NN) achieved substantial performance, R2 0.8786, mean squared error (MSE) 0.0004, absolute (MAE) 0.0155 compared existing studies.