Evaluation of Empirical Methods to Estimate Streamflow in Ungauged Basins(Case Study: the Sefidroud Watershed)

Estimation of discharge in ungauged basins is of prominent importance in hydrologic and water resource management studies; however, it is not possible to determine the runoff coefficient in different watersheds without streamflow data. In many study areas of the country (unit of hydrological basis and balance of water resources in the studies of the Ministry of Energy), there is no hydrometric station to measure the surface flow out of the area. Several methods have been introduced to estimate the discharge of ungauged basins, which can be classified into three main categories. The first category contains the methods that make a relation between precipitation and the produced runoff (such as the Inglis and De’Souza and the Indian Department of Irrigation (IDOI) methods). The second category includes the methods that estimate annual runoff deficit and predict the yearly runoff accordingly (such as Turc, Langbein, Coutagine, and Khosla methods). The third category covers the methods that take into account the physiographic characteristics of basins to estimate runoff (such as the Indian Council of Agricultural Research (ICAR), Justin, and Lacey methods). The SCS Curve Number (CN) method is also among the most common methods of estimating runoff produced by rainfall and considers various conditions in its formulations; nonetheless, determining the CN and its initial absorption coefficient is still challenging. The aim of the present study is to evaluate the efficiency of different empirical methods in the estimation of runoff in watersheds with different hydrologic and physiographic characteristics and climatic conditions in addition to giving some insights on the selection of the proper runoff estimation methods in ungauged basins.In this study, the application of empirical methods in the calculation of the outgoing discharge from various areas in the Sefidroud watershed was investigated. The Sefidroud watershed has a total number of 11 areas, 10 of which have hydrometric stations in their outlets. For these ten sub-basins, the observed annual runoff was compared with the results yielded by the aforementioned empirical methods, and the efficiency of each method was assessed accordingly for each sub-basin. The Root Mean Squared Error (RMSE), Standard Deviation (SD), Correlation Coefficient, and the Centered Root Mean Squared Deviation (CRMSD) were used to analyze the data. The runoff estimation methods investigated in this study included Khosla, Lacey, Inglis De’Souza, Coutagine, Turc, ICAR, IDOI, Justin, and the SCS-CN methods. Moreover, the authors of the present study tried to find the optimized value of the initial absorption coefficient in the SCS-CN method in order to obtain a reasonably accurate estimation of runoff for each sub-basin. 

The results of the present study indicated that the Khosla and the SCS-CN methods with an initial absorption coefficient of 0.05 and 0.2 showed the poorest performance in all sub-basins. Moreover, the Inglis De’Souza method was not applicable in Iran’s sub-basins due to its different approach in dealing with plains and highlands. Because the study areas in the catchments of Iran are all a combination of plains and elevations and sometimes include a combination of several plains and several elevations with different characteristics. The optimized values of the initial absorption coefficients varied between 0.0006 and 0.25, which implies that a specific value of initial absorption cannot be used in all of the sub-basins to achieve the best accuracy in the estimation of runoff. Comparison between the results yielded by other methods (i.e. Turc, Coutagine, IDOI, ICAR, and Justin) with the observed streamflows indicated that the choice of the best method depends on the error index used for comparison. In other words, the Justin method had the best performance in terms of correlation with the observed runoff in the Sefidroud watershed. But, in terms of the RMSE error index, the IDOI method generally performs better. Finally, the Coutagine method had a good performance in terms of both correlation and RMSE in the main study areas. 

According to the results of the present study, the Justin method is recommended for areas that have a high altitude and temperature gradient and at the same time have a high flow coefficient. The IDOI method performs best for sub-basins that have a high runoff to rainfall ratio. As this ratio decreases below 0.2, the IDOI method is likely to produce poorer results. The Coutagine method showed a moderate performance in most of the studied areas, which suggests that it can be employed to produce conservative results in many areas under study.