Modeling of water and nitrogen components in irrigated agricultural areas using SWAT model: The case of lower Seyhan Plain

As developments in the field of agriculture continue as a requirement of modern agriculture, environmental risks arising from increasing agricultural inputs are also increasing. Model studies, which started in the 1970s in order to support agricultural production and reduce possible risks, have increased and accelerated with artificial intelligence. Models are comprehensive programs that can simulate complex and large basins spatially and temporally, calculate the reactions occurring within the basin and the interactions of these responses with each other, and present quantitative results to end users. Models will continue to be used more frequently in the future, especially for decision-making mechanisms to combat climate change and prevent point pollution. Monitoring nitrogen fertilizers, which are the source of non-point pollution, in basins with a highly heterogeneous structure in soil, land use and agricultural practices, such as this study area, is very important from an environmental and economic perspective. Models are a very useful and economical tool for monitoring and predicting changes in heterogeneous areas. The study area is an artificial basin consisting of irrigated lands with well-known hydrological boundaries, where the amount of incoming and outgoing water is measured, with semi-arid climate conditions and where citrus, wheat and first crop corn are grown. There has been a meteorological station in the study area within the scope of large-scale projects since 2006, and daily drainage flows at the basin outlet and NO3 concentrations in the drainage water were measured daily with the help of an automatic sampler. Summer and winter cropping pattern maps were developed by checking locally and at regular intervals. Soil samples were taken from different soil profiles and surveys were conducted to collect information about irrigation and fertilization activities applied by farmers in the field. In the study, the SWAT model was used in the ArcSWAT interface, and the warm up phase was completed in 2008, the calibration phase was completed in 2009-2012, and the validation phase was completed in 2013-2014. While the statistical values R2, NSE, PBIAS values for daily drainage flow efficiency after the calibration phase were found to be 0.75, 0.74 and -3.2, respectively, these values were recorded as 0.17, -0.31 and -8.3 for validation. NO3 concentration values in daily measured drainage waters were first converted to NO3-N and then to N load in kg N per day by flow rate. While the statistical values R2, NSE, PBIAS values for daily N load were 0.38, -0.59 and 85, respectively, these values were found to be 0.01, -1.60 and 77 for validation. When the statistical performance values were examined, it was seen that the model's drainage current prediction capacity was higher than its N load prediction capacity. It is foreseen that these performance statistics can be improved with longer-term and homogeneous years of calibration and validation data, and can be used successfully in areas with highly heterogeneous characteristics, such as the study area