River flow forecasting is required to supply basic information on a wide range of problems associated with the planning and operation of river systems. While traditional models are important for understanding hydrological processes, still accurate predictions are required at some specific locations. In this context ANN (Artificial Neural Network) black box in nature, is emerging as most successful machine learning techniques with flexible mathematical structure, capable of showing non-linear association between input and output data. Therefore,in current years, ANN models are used frequently for forecasting of flood. Most of the previous work of ANN techniques in forecasting has been done by taking rainfall as input and runoff data as output. These forecasting has not considered other flood causative factors. Therefore, the main goal of this study is to develop a flood model by considering many factors which are responsible for occurrence of flood. Also, it has been observed that space and air-based observations of earth gives a unique way to monitor and assess the floods. So, this study has tried to take an advantage of both ANN and GIS by integrating these techniques together. Therefore, ANN and GIS are used for modeling and simulations of flood prone area in Yamuna Nagar district of Haryana. The model was implemented in MATLAB. Maps related to flood causative factors such as rainfall, slope, elevation, flow accumulation, soil, land use and geology are prepared with the help of GIS, Remote Sensing data and field surveys. Water level is produced by ANN model and then flood map showing flow accumulation is generated using GIS.
The model performance is measured in terms of coefficient of determination (R2), sum squared error(SSE), the mean squared error(MSE) and the root mean squared error(RMSE).
Key words: Artificial Neural Network (ANN), Geographic Information System (AAN), MATLAB, Coefficient of determination (R2), Sum Squared Error (SSE), Mean Squared Error (MSE) and Root Mean Squared Error (RMSE).
|
