Mohmed Abdallah Mohmed Abdalhi2023-03-152023-03-152010-05http://hdl.handle.net/123456789/18994In fulfillment of the requirement For the degree of Master of Engineering - Hohai UniversityAbstract Abstract Hydrological modeling of water resources systems has contributed to a great deal of analysis-type knowledge. The understanding of spatial and temporal events has led to a higher degree of confidence in the design of most water resources projects. This research study seeks on how to apply the Artificial Neural Network rainfall runoff model and Xinanjiang rainfall-runoff model in humid and semi humid catchments of Guanyin Qiao, determines the optimum parameters of the two models used, forecast the real time of peak flood event and evaluate the models acceptability in the catchment. The study area is Guanyin Qiao catchment in Longyan area located at the Western region of Fu Jian province in China, with the total catchment area of 324 Kmz. The daily data from 1992 — 200lwere used and its sets were divided into two groups of data sets, which includes, calibration set and validation set for the two models. The manual calibration processes of the two models were successful and obtained the optimum parameters used in validation processes. The average relative errors for the ANN model and Xinanjiang model were 18.45% and 6,44% respectively, this means that, the average relative errors for the two models have not exceeded the acceptable relative error values, because each flood simulation or forecasting result is acceptable only if the percentage of the peak relative error between simulated and observed flood is less than or equal to 20%. The only one drawback of this study is that, the ANN model is able to make good forecasts in the short lead times of performance, but with increase in the lead times the results deteriorates rapidly. The manual calibration of Xinanjiang model is good for simple model but in long period of year’s calibration and complex model, it’s tedious and affects the accuracy of the parameters. Based on the overall results and the comparative study, the research proved that, the ANN rainfall-runoff model and Xinanjiang rainfall- runoff model applied in this study is more efficient and qualified to be applied in the Guanyin Qiao catchment and can be categorize in grade B base on the Chinese Standards classification (2000). Keywords: ANN model, Xinanjiang model, Flood forecasting, Guanyinqiao Catchment.flood forecastingwater resourcesThesis