Sensitivity of the SWAT Model to Soil Data Parameterization; Case Study in Saz - Çayırova Stream, Turkey

Creative Commons License

Oruç H. N., Çelen M., Gülgen F., Öncel M. S., Vural S., Kılıç B.

5th EurAsia Waste Management Symposium, İstanbul, Turkey, 26 - 28 October 2020, pp.503-513

  • Publication Type: Conference Paper / Full Text
  • City: İstanbul
  • Country: Turkey
  • Page Numbers: pp.503-513
  • Yıldız Technical University Affiliated: Yes


The accuracy of hydrological models such as SWAT model depends on a well-defined physical property of the watershed such as land cover, soil properties, topography and meteorological factors. The objective of this study is (i) to model the hydrological processes of Saz-Çayırova Stream where located in northwest parts of Turkey, (ii) to analyze the soil physicochemical characteristic of the catchment area with laboratory and field studies, (iii) to evaluate the effect of specified soil database and the FAO soil database over streamflow using SWAT model without calibration. Twenty-nine soil sampling points were identified to represent the whole of the catchment. Among the required soil parameters clay, silt, sand, rock / gravel, bulk density, organic carbon, maximum rooting depth and depth from soil surface to bottom of layer determined by field observations and laboratory analyzes. The most noticeable difference among these parameters is the clay and sand content. While the average clay and sand contents were measured 5.1% and 43.3%, respectively, in the SSM database, these values were defined as 31% and 27%, respectively, in the FAO database. Runoff values produced using FAO and SSM databases were compared with observation data without applying any manual or automatic calibration technique, because calibration processes mask differences which can occur as a result of soil data sets. Determination coefficient (R2), Nash-Sutcliffe coefficient of efficiency (NSE) and Root Mean Square Error (RMSE) values were used to evaluate the model results. NSE value was calculated 0.19 and 0.48 for FAO and SSM, while RMSE value was 0.34 and 0.25, respectively. In this respect, while the SSM database was making acceptable estimates even before calibration, the found NSE values between the runoff values produced using the FAO database and the observation data remained below the acceptable threshold value (0.36). Forming unique soil structure of the watershed, dominated by residential and industrial areas such as Saz-Çayırova stream, with laboratory and field studies allows the calibration processes to be made more consciously.