نوع مقاله : پژوهشی
نویسندگان
1 دانشیار، گروه هیدرولوژی و توسعه منابع آب، پژوهشکده حفاظت خاک و آبخیزداری، سازمان تحقیقات، آموزش و ترویج کشاوزری، تهران، ایران.
2 استادیار، گروه هیدرولوژی و توسعه منابع آب، پژوهشکده حفاظت خاک و آبخیزداری، سازمان تحقیقات، آموزش و ترویج کشاوزری، تهران، ایران.
3 دانشیار، گروه مهندسی رودخانه و سواحل، پژوهشکده حفاظت خاک و آبخیزداری، سازمان تحقیقات، آموزش و ترویج کشاوزری، تهران، ایران.
چکیده
کلیدواژهها
موضوعات
عنوان مقاله [English]
نویسندگان [English]
Introduction and Goal
Accurate estimation of annual runoff in ungauged watersheds remains a fundamental challenge in water resources management. Although employing empirical models is a common approach for runoff estimation in such regions, their efficiency necessitates the optimization and localization of their regional coefficients based on the specific geographical characteristics of each basin—a research gap that is still evident across large watersheds in Iran. Accordingly, the present study aimed to model and derive the regional relationships of the Justin and Icar coefficients using the available parameters of the Atrak watershed. The ultimate objective is to develop an optimal model for runoff estimation that can be generalized to adjacent basins with similar climatic conditions.
Materials and Methods
The Atrak watershed, covering an area of 18,773 square kilometers, is located in northeastern Iran, extending across three provinces: eastern Golestan, northern North Khorasan, and a portion of northern Razavi Khorasan. In this study, the boundaries of the Atrak watershed and its sub-basins were delineated using base maps, including a Digital Elevation Model (DEM) and the drainage network. Initially, hydrometric stations within the region possessing a minimum of 25 years of observational data were screened as primary candidates. Subsequently, to standardize the length of the statistical period, missing data were reconstructed using correlation methods. Following the evaluation of data adequacy and randomness via the Run test, 15 hydrometric stations with a common base statistical period of 33 years were ultimately selected for modeling and further analysis. Using the mean annual discharge values of the selected stations, along with the extracted physiographic and meteorological data, the regional coefficients for the Justin and Icar empirical models were estimated at each hydrometric station. In the subsequent phase, univariate and multivariate regression models were developed using morphoclimatic indices and the stations’ regional coefficients. The proposed regional relationships were evaluated using performance metrics, including the Correlation Coefficient (R), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Finally, the best-performing model was applied to estimate the regional coefficients across the entire Atrak watershed.
Results and Discussion
Initially, the runs test indicated that the data were random and statistically valid. In the Justin model, the univariate correlations with area, slope, land use, and geology ranged from 41% to 87%, whereas the multivariate model achieved a correlation of 0.94 with lower errors (RMSE = 0.09, MAE = 0.06). For the Icar model, the univariate correlations were 45%, 63%, 47%, and 78%, respectively, with larger errors, while the multivariate model reached a 93.7% correlation and lower errors (RMSE = 0.28, MAE = 0.21). Overall, multivariate regressions performed better in estimating the regional coefficients for both models, and both models had correlations above 90%. However, the error of the Icar model was roughly three times that of the Justin model; therefore, the Justin model is better suited to the Atrak watershed. The range of the regional coefficient for Justin and Icar in the hydrologic units of Atrak is 0.027 to 0.722 and 0 to 2.42, respectively. Regarding factor effects, slope has a positive relationship with the regional coefficient, while area, land use, and geology have negative relationships. Geology showed the highest correlation; permeable formations reduce the regional coefficient and runoff by absorbing surface water. Land use also affects the runoff coefficient through permeability, evaporation, and vegetation cover; in forests and rangelands, greater interception and infiltration result in lower runoff coefficients. Overall, slope and area have lower correlations, and in small, high-elevation catchments with lower precipitation, both flow depth and runoff coefficients are lower.
Conclusion and Suggestions
The findings of this research indicate that localizing the coefficients of empirical models, such as Justin and Icar, using physiographic and climatic variables is an essential requirement for the management of ungauged watersheds. Given the climatic variations within the study area, the application of constant and generalized coefficients can lead to substantial errors in runoff estimation. Consequently, the regression relationships developed in this study provide a practical tool for water resource experts and managers in the Atrak watershed, enabling more accurate estimations of the annual runoff potential without the need for hydrometric stations. Accordingly, it is recommended that future studies evaluate the performance of these relationships in adjacent watersheds with similar climatic conditions. Furthermore, considering recent changes in land use and climate, it is suggested to investigate the impact of vegetation parameters and soil permeability coefficients using non-linear or artificial intelligence models within this watershed to further enhance the accuracy of the estimations.
کلیدواژهها [English]
)