下载掌阅APP，畅读海量书库
立即打开
Nitrogen (N) is crucial to aquatic and terrestrial organisms, and its use as a fertilizer in modern agriculture is one of the main reasons that crop production has kept pace with the human population growth (Robertson and Vitousek 2009). N is also a major pollutant facing humanity today, since excessive N discharge originated from intensive urban and agricultural sources can lead to eutrophication of lakes and reservoirs, groundwater nitrate enrichment, loss of biodiversity, and other serious environmental and human health issues (Filoso et al. 2004; Galloway et al. 2008; Chen et al. 2016). To reduce the adverse impacts of N pollution, many countries have incorporated water pollution mitigation strategies, such as the Total Maximum Daily Load program and various Best Management Practices, into watershed management.However, to develop suitable nutrient management programs requires knowledge of spatiotemporal variations of N concentrations in rivers and streams, as well as the load contributions of various pollution sources (Nestler et al. 2011; Liu et al. 2013; Shen et al. 2014; Hashemi et al. 2016).
River drainage networks are hierarchically organized systems where first-and second-order streams, usually referred to as headwater streams, constitute at least 70% of the total stream length (Lassaletta et al. 2010). In addition to their dominance in numbers and cumulative length, headwater streams also exert controls on stream runoff and downstream fluxes of nutrients. For example, Alexander et al. (2007) found that first-order headwaters contribute approximately 70% of the annual runoff and 65% of the N flux in second-order streams in the northeastern U.S. Dodds and Oakes (2008) estimated that water chemistry was closely correlated with riparian land cover adjacent to the first-order streams in eastern Kansas. These results are similar to the findings in European headwater agricultural watersheds (Lassaletta et al. 2010). However,owing to their small size and intermittent flows, headwater streams are often omitted or inaccurately represented on stream maps derived from topography,and are therefore ignored by water pollution mitigation strategies, in spite of their vulnerability to human activities and significant influences on downstream water quality (Armstrong et al. 2012; Rasmussen et al. 2013).
China is faced with the severe challenge of widespread eutrophication due to excessive discharge of nutrients, including N, especially for the southeastern region with high population density and rapid economic development. Many lakes and reservoirs in headwater regions have been designated as drinking water sources to provide sufficient amounts of high-quality water (Li et al.2013). Although previous studies of N pollution of these waters have focused on agricultural drainage (Jeon et al. 2007; Liu et al. 2009; Shen et al. 2014;Hashemi et al. 2016), the pollution sources are usually more intricate (Wang et al. 2005; Ongley et al. 2010; Asian Development Bank 2011). For example, many small farms exist in the low mountains, where domestic sewage and waste are discharged to the environment without appropriate treatment. Besides, the improvement in people's living standards has driven growing appetites for meat and dairy products, but the burgeoning animal feeding operations usually lack sufficient waste disposal facilities in rural areas. Semi-continuous discharges of N pollutants from these mini-point sources can have a significant impact on both the watershed and downstream drinking water sources.
In addition to complex pollution sources, headwater agricultural watersheds are usually dominated by multi-pond systems in Southeastern China (Yin et al. 1993; Verhoeven et al. 2006; Chen et al. 2017). As a semi-natural wetland,a multi-pond system is composed of many tiny ponds, connected by ditches and streams. Such ponds have existed for centuries, serving as drinking water sources for farmers and livestock, and for washing, fishing, and irrigating.Recent studies show that multi-pond systems effectively retain NPS pollutants through sedimentation, adsorption, and uptake by aquatic plants (Liu et al.2009). For instance, a five-year monitoring study in the Liuchahe Watershed near Chaohu Lake estimated that the connected ponds reduced phosphorus in agricultural drainage by > 90%, if the total pond area occupied 6%-10%of the entire watershed (Yin et al. 2006). Reduction capacities also show a gradient from foothill ponds to riverside ponds (Qiang et al. 2006). However,the dynamics of many small water bodies are challenging to quantify in terms of their relationship to complex pollution sources, and further, it is difficult to develop suitable nutrient management programs.
Many NPS pollution models with various capabilities and degrees of complexity have been developed in the past decades, including simple export coefficient models like PLOAD, regression models like SPARROW, and process-based models, including HSPF, SWAT, and INCA (Yang and Wang 2010; Butcher et al. 2014; Wellen et al. 2015). The simple models are commonly used to estimate pollution loads from different sources, but they are difficult to validate, and fail to account for pollutant migration and transformation (Ma et al. 2011; Liu et al.2013; Lu et al. 2013). In contrast, process-based models have increasingly been applied to evaluate NPS pollution and associated uncertainties. In this study,HSPF was selected due to open-source characteristics, flexible input/output configurations, and the ability to simulate the fate and transport processes from various pollution sources to the final receiving waters. Nevertheless, HSPF has mainly been used to analyze nutrient discharge from agricultural drainage (Filoso et al. 2004; Jeon et al. 2007; Ribarova et al. 2008; Liu and Tong 2011; Li et al.2015), despite its ability to incorporate different mini-point sources and surface waters (Bicknell et al. 2005; Yang and Wang 2010). Poor characterization of pollution sources and scattered small water bodies may lead to overestimation of the relative contributions from agriculture and disproportionate targeting of potential pollution sources during process-based simulations (Lassaletta et al.2010).
The Chenzhuang (CZ) Watershed in the south of the Mt. Mao region was selected for this study as a testbed. It lies in the headwater region of two important basins with an environmental concern: the Taihu Lake Basin and Qinhuai River Basin. Although there are many small reservoirs designated as drinking water sources in peripheral areas, about 46% of them failed to reach the regulatory thresholds for N concentrations in the Chinese Surface Water Quality Standard (GB 3838-2002), which garners increased attentions from decision-makers and stakeholders (Li et al. 2013; Wang et al. 2016; Chen et al.2017). In view of the above scientific challenges and the severe water quality situation in the reservoirs, the objectives of this work are to (1) employ HSPF to simulate N discharge and transport processes from all known anthropogenic mini-point and nonpoint pollution sources in the headwater agricultural watershed, (2) analyze the spatiotemporal patterns of N concentration in the multi-pond system, and quantify their load contributions from each pollution source, (3) compare the results with previous studies on N source attribution,and (4) propose suitable strategies to improve water quality for both the watershed and downstream regions. This study is a first attempt to incorporate multi-pond systems into the process-based modeling of NPS pollution. The proposed methods and analytical results can inform other hydro-environmental studies which focus on scattered and small water bodies in urban and natural systems. The methods and results are also useful to water pollution prevention and mitigation for entire river basins.