JWSS - Journal of Water and Soil Science

Estimation of flood flow rate represents a method of preventing damages associated with this natural phenomenon. This estimation is one basis in the design of various hydraulic structures, dam spillways, watershed management and flood control. The maximum flow rate of floods is determined by methods such as Creager, Jarvis-Meyer, Cypress-Creek, and rational method. Rational-probability method is an alternative to estimate peak flood rates, and is expressed as:

 Q(y) = F. C(y). I(tc.y).A

 where Q is maximum flood flow rate (m³/sec) y is the return period (year) C(y) is runoff coefficient with a return period of y A is watershed area (km²) I is rainfall intensity (mm/hr) for a specified return period equal to time of concentration of the watershed and F is the conversion factor equal to 0.278 when the above units are used. The basic concept of this method is the same as that in the rational method except that the return period is also included in the equation. Usually, runoff coefficient, C(y), is determined empirically from tables cited in the literature (e.g., Chow et al., 1988). In the present research, data from 18 hydrometry and 6 rainfall-recording stations (located in Caspian - Sea watershed) were analysed using TR software. The Caspian - Sea watershed (which covers eastern and centeral parts of Iran's No. 1 main watershed) has the sub-basins of Atrak, Tadjan, Chalus, Sardabrood, Siahrood, Gorganrood, Safarood, Kesilian, Babolrood and Neka. Runoff coefficients with return periods of 2, 5, 10, 25, 50 and 100 years were determined for these sub-basins and iso-coefficient curves were plotted. The results showed that computed runoff coefficients were less than the values given in the literature because they are determined from observed flow rate and rainfall intensity in each catchment. It was also shown that runoff coefficient increased with increasing return periods. Application of the computed runoff coefficients in three sub-basins of the area resulted in more accurate estimations of maximum flood rate than when the values for these coefficients cited in the literature were applied.

The rainfall-runoff process and flooding are hydrological phenomena that are difficult to study due to the influence of different parameters. So far, different methods and models have been provided to analyze these phenomena. The purpose of this study is evaluation of adaptive neuro-fuzzy inference system (ANFIS) for storm runoff coefficient forecasting. To that end, Barariyeh watershed was chosen in Neishabour and the data of 33 events were collected from 1952 to 2006. Factor analysis (FA) was used for determination of independent variables in storm runoff coefficient forecasting. Four variables were selected as independent variables, including average rainfall, third, first and fourth quartiles of rainfall intensity and also five other variables included &phi index and first to fourth quartiles of rainfall intensity. Other variables combined based on their hydrological role were considered as ANFIS inputs. The results revealed that the ANFIS inputs including first to fourth quartiles of rainfall intensity, &phi  index, and total rainfall of five days before can predict storm runoff coefficient with R²=0.91, RMSE=0.02506, MAE=0.0666 and CE=0.87.