Flood modeling and flow routing in ungauged watersheds (Case study: Tamar watershed, Golestan province)

Flood modelling and watershed routing are the main topics of interest for hydrologists in the process of transforming rainfall into runoff hydrograph and streamflow through a watershed. In the design of some water systems, it is necessary to know the changes in flow over time (hydrograph). Empirical formulas, drainage pattern, simulation method, statistical estimation of maximum instantaneous flow rate, regional analysis and flood index method can be used to estimate maximum instantaneous flood in ungauged watersheds. Each of the flood modeling methods can be used in specific conditions. In the meantime, only methods that lead to the preparation of hydrographs are able to provide accurate details of flood hydrograph characteristics. Flood management based on hydrograph analysis approaches in ungauged watersheds is very difficult, so the use of empirical methods and physiographic parameters in flood prediction models can be considered as an in-hand and easy to use method. The direct runoff of a watershed, in addition to the temporal and spatial distribution of rainfall, depends on soil properties, topography, geomorphology and watershed drainage network. Obviously, there is a close relationship between the geomorphological characteristics and the hydrological response of a watershed. The use of Gamma distribution has been significantly expanded due to the production of smooth shape and flood hydrograph as well as the ease of extracting its parameters from the geomorphological characteristics of the watershed. The main aim of this study is the application of Gamma distribution using catchment characteristics to modeling the direct runoff hydrograph of Tamar watershed.

The Tamar watershed which located at east of the Golestan Province (1515 km2), was divided into subwatersheds and required parameters for simulation procedure were calculated and rainfall-runoff process modeled for each hydrologic unit, as well as simulated flow routed through the main reaches of the watershed. Since the reservoir gamma distribution is considered linearly in the two-parameter distribution, the storage coefficient of this reservoir can be estimated using the watershed characteristics. In this regard, the data required to determine the parameters used in the gamma distribution in flood simulation were extracted. Then, using the gamma distribution, the flood current in each sub-watershed was simulated. In the next step, using the Muskingum method, the flood hydrograph was routed through sub-watersheds. Then, taking into account the location of sub-watersheds and routing intervals, the unit hydrograph of flood moments was estimated for the whole watershed and converted into a unit hydrograph equal to one third to one fifth of the watershed lag time based on SCS method. In the next step, by multiplying the dimensions of the resulting hydrograph in the effective precipitation pulses, the flood hydrograph was modeled for each storm. Based on the available data and their relevance, eight corresponding flood and rainfall observational events were selected. The simulation results were evaluated with the criteria of coefficient of determination, relative efficiency coefficient, agreement index, and relative agreement index.

Based on the presented results and the visual comparison of the simulated hydrographs, it can be stated that the model results in estimating the components of flood hydrographs had an acceptable accuracy. The results of simulated and observed hydrographs comparison using coefficient of determination (60%), and index of agreement (74%) show that the accuracy of existing method is on the average at predicting complete shape of flood hydrograph. However, due to the use of experimental methods and coefficients in estimating the inputs of the modeling process, the existence of errors in the model results is largely expected.

As a concluding remark, it should be noted that using physiographic and easy obtained parameters can be considered as an appropriate approach in flood modeling, so the relationship between geographic and properties of the flood hydrographs be more elaborated in future research. The proposed flood modelling approach and application of Gamma distribution can provide a significant contribution towards flood management program in areas that limited data were available. It is identified that direct run-off hydrographs and model results are more responsive to variations in the value of the uncertain parameters. It should be noted that the modeling of complete hydrograph as well as the extraction of peak discharge components, time to peak and flood volume at storm events is one of the advantages of the proposed method. In other words, in addition to the possibility of comparing sub-watersheds in terms of flood occurrence, the shape of the hydrograph and its change over time in different parts of the drainage network can be obtained. Thus further investigation is required in order to evaluate variations in the model input parameters in different climate conditions.