Estimation of flood due to rainfall has been always a main subject in hydrology science. According to the importance of unit hydrograph for river flood forcasting and design of hydraulic structures and being not sufficient data at hydrometry stations of Khuzestan watershed, necessity of using artificial unit hydrograph for estimating of flood in rivers with no data is obvious. Moreover, because of the importance of rainfall in producing runoff, evaluation of the effect of loss models in production of water surface is important. In this research, two different models were evaluated:
Initial- continues loss and SCS loss. So two direct runoff models include SCS and Snyder were applied for simulation of flood hydrograph due to excess rainfall. Therefore with 15 series rainfallrunoff data of Abolabbas watershed, HEC-HMS model was applied to calibrate and validate selective models. Three statistical indexes (mean relative error, root mean square error and coefficient of correlation) are applied to estimate the errors. The results of mean relative error of flow peak and time of flood peak in evaluation models show that the best estimation of flow peak and time of flood peak obtain in SCS loss model with Snyder unit hydrograph and in initial- continues loss model.Comparison of the average values of root mean square error show that minimum value of RMSE is related to SCS loss model with Snyder unit hydrograph and in initial- continues loss with SCS unit hydrograph. This show that these models the best results among other models for their ability in separation of excess rainfall. Finally, the occurrences that are simulated by SCS unit hydrograph and SCS loss models have the best correlation with actual conditions. Therefore these
methods are able to accurately simulate the flood extent within a poorly gauged basin.