1. Introduction
Population growth, urban development and industrialization of the communities have adversely affected the hydrology of the catchment area, which exacerbates the floods, increases pollution in the riffle, decreases the base flow and reduces the nutrition of groundwater. Increasing the impenetrable levels of these areas affected by this trend, reduced their permeable level, which plays a major role in absorbing rainfall, and subsequently adds to the total volume of runoff [1] [2] [3] . The result of these changes will shorten the focus time and increase the intensity of flood drainage in the area. [2] , using the LISFLOOD model, investigated the role of land use change on peak flood discharge in the two Oder and Meuse areas. [4] [5] [6] , by combining the HEC-HMS and HEC-RAS models in GIS, analyzed different flood control options and selected the best option. Also, [7] presented a regional flood model with a combination of these models in the San Antonio area and noted the model’s efficiency in predicting flood events. Historical reports and documents show that the watersheds of southern Iran, especially Shiraz, have a high potential for flood [8] [9] [10] . Due to the lack and adequacy of the statistics of flood, especially in urban basins, this study attempted to compare the current method of estimating flood in these basins with actual statistics. To do this, we attempt to simulate the flood hydrograph in the natural and urban basin of the dry river of Shiraz using the HEC-HMS software.
2. Location of Study
Part of the catchment area of the Shiraz Khoshk River with an area of 900.3 square kilometers is located between the geographical lengths of eastern 52˚12'30'' to 52˚41'30'' and north geographical latitudes of 29˚34'37'' to 29˚58'07''. This area is bounded to the northern plains of Zareghan and Beyza and from south to Shiraz city, from the east to the Maharloulake and from the west to the Shool col. The only river in this basin is the seasonal river, which has no water flow in the dry months. The nourishing sub-basins of the Khoshkriver are: Nahr Azam, Ali Abad, Chenar Sokhteh and Chenar Rahdar, which totally are four sub-basins. Their specifications are presented in Table 1.
3. Statistics and Required Data
The meteorological stations network in the study area includes six rational stations, one synoptic station and five hydrometric stations. The length of the station's statistical period varies from seven to fifty five years, which was eliminated for better use of this short-term station score. Finally, a 20-year common period was used from the year 1985 to 2005, and incomplete stations statistics with correlation method was reconstructed. Here, we study the Khoshk River catchment area, which is the data required for the model is the amount of CN (Curve Number) that is used by the soil map of the province of Fars and the basin boundary that has been digitized in the GIS software. Subsequently, the map of the hydrological groups in terms of depth and the soil texture of the basin are prepared.
4. HEC-HMS Model
To run this model, some parameters are needed, such as: initial losses, the conversion of excess precipitation to surface flow, determination of base flow, calibration and validity of the model, which will be explained below the calculation method of each one.
5. Basin Losses
Some of these factors include interception, evapotranspiration and reservoir
water in the holes. There are 7 different methods for calculating these factors. In this research, the SCS method is used.
(1)
(2)
(3)
Pe: Height of Effective Precipitation (Run off) (mm).
P: Height of Precipitation (mm).
S: Maximum Potential Storage of Basin (mm).
CN: Curve Number.
Ia: Initial Loss (mm).
6. Transfer Model
The SCS hydrograph method is used to convert excess precipitation to surface flow.
(4)
(5)
(6)
(7)
7. Calibration
In this part, the parameters of the model are calibrated with the group of data. Then, model validation is performed through the run of the model with
![]()
Table 2. CN Values and primary losses.
optimized parameters for the second group of data and finally, the simulated hydrograph is compared with the observed hydrograph. Also, in the validation phase, using the Nash-Sutcliff and variance coefficient were evaluated. It should be said when the coefficient of variance is lower the performance of the model will be higher. In the criterion of the Nash-Sutcliff, one indicates the complete matching of the simulation and observation hydrograph.
![]()
Table 3. Hydrologic element results.
Qoi, Qsi observed and simulated values Qo mean of observed discharge.
8. Research Methodology
In order to simulate the flood hydrograph, HEG-HMS 4.2 model we used. Firstly, using ARCGIS, topography maps, vegetation, land use and soil hydrological groups were provided and integrating land use maps and soil hydrologic groups the CN map was prepared and its mean values were determined for sub-basins. In order to determine the values of Intensity-Duration-Frequency using the suggested method Qahraman-Abkhezr, the amount and intensity of rainfall were calculated. Figure 1 shows the prepared CN map. The values of the CN and the calculated primary losses for the studied sub-basins are shown in Table 2.
Then, by inserting the calculation (primary losses, CN, lag time, area of basin, rainfall values, etc.), in the HEC-HMS, the model was implemented (Figure 2). The Soil Conservation Service (SCS) method was used for the flow and water losses and the results are presented in the Table 3.
Using the Nash coefficient and the sum of the squared error, the model was calibrated. The flood of November 1986 and December 2001 confirmed the model’s accuracy. The estimation of the Nash criteria in the sub-basins of Chenar Sokhteh Nahr Azam, Ali Abad, Chenar Sokhteh Khoshk, and Chenar Rahdaris 0.74, 0.83, 0.71 and 0.85, respectively. The calibration results of the model as an example in the event of the year 2001 is presented in Figure 2.
9. Discussion and Conclusion
Investigating the results of the HEC-HMS model in the catchment basin of the Shiraz Khoshk River indicates that although the volume of the flood and the shape of the simulated hydrograph are not consistent with the observed hydrograph, nevertheless, it has estimated the peak flood discharge well with an error rate of two to five percent.