An agro-hydrological process-based export coefficient model for estimating monthly non-point source loads in a semiarid agricultural area

Abstract

Export coefficient models (ECM) with few parameters and simple structures have been widely used to estimate nonpoint source (NPS) pollution loads. However, they are unsuitable to simulate characteristics and distributions of NPS pollution over semiarid agricultural watersheds of intensive irrigation, infrequent precipitation, and shallow groundwater. In this study, field monitoring was conducted to identify key impact factors of total nitrogen (TN) loads from NPS pollution over a typical semiarid agricultural watershed in northern China. Sources and characteristics of TN loads were examined by integrating the monitoring, a Pearson correlation analysis, an agro-hydrologic model, and a genetic algorithm into an ECM framework. This study quantitatively revealed the relationship between agro-hydrologic features and NPS nitrogen pollution in semiarid agricultural areas; it also simplified quantification of nutrient distributions in various seasons under pollution-source diversity. An accurate ECM was constructed to simulate variation of TN loads with seasons. Compared with traditional ones, the developed model reduces the simulation error of TN loads from 16.2% to 8.9% and improves R2 from 0.072 to 0.692. Simulation results show that the dominant TN nutrient source, i.e., farmland, accounts for 49% of annual TN loads. If water use efficiency increases from 40% to 50%, the largest reduction of TN loads would occur on farmland (about 14.4%), followed by grassland (12.0%), construction land (11.5%), bare land (11.0%), and woodland (8.3%). Improving water use efficiency and optimizing irrigation schemes and land use types should be preferentially recommended for NPS pollution control over semiarid agricultural regions.