Rainfall-Runoff Modelling Using HEC-HMS And ANN Models

In this modern world,floods are considered as disastrous hazards which occurred often in the monsoon season.This study develops event and continuous hydrologic modeling in the Zijinguan basin of Daqinghe river using the Hydrologic Engineering Center-Hydrologic Modeling system and ANN model using Artificial Intelligence for prediction of runoff discharges.Hydrologic models are simplified,conceptual representations of a part of the hydrologic cycle.The Hydrological Cycle is the water cycle that circulates water from the land to the atmosphere and back again to the land.Study the loss methods provided by HEC-HMS and its suitability for model fit and development of an Artificial Neural Network model for rainfall-runoff modeling are the main objective of this research.The watershed was delineated with HEC-Geo HMS in Arc GIS and its properties were extracted from a 30 m × 30 m Digital Elevation Model(DEM).The model encompasses both the Soil Conservation Service(SCS)Curve Number and the Soil Moisture Accounting(SMA)loss methods for simulating surface runoff in the events and continuous models,respectively and Soil Conservation Service(SCS-UH)Unit Hydrograph and Muskingum were used for transform and flow routing.Specifically,eight rainfall events were selected for calibrating and verifying the event model.Similarly,for the continuous model wet seasons of eight different years were used for calibration and verification.The calibrated parameters of events models were used in the continuous model.The soil moisture and evapotranspiration data were decoded from the satellite data to set in a continuous model.Here,a continuous model has to compromise the overall modeling quality because of insufficient high-resolution data for developing,calibrating,and validating the model.For,the ANN model eight flood events of eight different years were used for training and testing.Rainfall was the input and discharge was target output for flood modeling.Feedforward Backpropagation Algorithm is used for training the network with the Levenberg–Marquardt technique.In the current research ANN model of,Multilayer Perception with one hidden layer was used for simulation.In this paper,the performance of the event-based model and continuous-based model was compared and was analyzed.SCN based model performs well than SMA based model.The ANN-based model was developed to make more conformity for the prediction runoff volume.The model evaluation index for the ANN model gave good and acceptable results.The sensitivity analysis of SMA based model parameter was done and analyzed.The strategy of the combination of fine-scale event and coarse-scale continuous hydrological modeling with(HEC-HMS)along with the ANN model has shown in this work.