Evaluating the impacts of soil data on hydrological and nonpoint source pollution prediction

https://doi.org/10.1016/j.scitotenv.2016.04.107Get rights and content

Highlights

  • ā€¢

    The impacts of available soil data on NPS modeling are quantified.

  • ā€¢

    This paper provides information for the appropriateness of each soil database.

  • ā€¢

    Error from soil data to watershed management strategy was assessed.

  • ā€¢

    The choice of soil data shows great impacts on watershed models.

  • ā€¢

    This paper also indicates that NPS-TP outputs are more sensitive to soil data.

Abstract

Soil data are one key input for most hydrological and nonpoint source (H/NPS) models, and quantifying the error transmission from soil data to H/NPS predictions is of great importance. In this study, two typical soil datasets were compared using the Soil and Water Assessment Tool (SWAT) in a typical mountainous watershed, the Three Gorges Reservoir Region, China. Besides, the effects of soil data resolution were evaluated, and the error transmission from soil data to watershed management strategy was assessed. The results indicate that model outputs are not sensitive to changes of soil data resolution but the choice of soil data greatly impacts the application of watershed models, in terms of the goodness-of-fit indicator, predicted data and related uncertainty. This soil data-induced error would be inevitably magnified from the flow simulation to the NPS prediction stage. This study could indicate that the choice of soil data will lead to significant differences in management schemes for specific pollution periods. These results provide information on the impacts of soil data on the functionality of watershed models and valuable information for the appropriateness of each soil database.

Introduction

Soil, which represents the historical processes of the Earth's surface, plays a key role in determining the allocation of water between rainfall, evaporation, infiltration and direct runoff (Essaid et al., 2015). Its physical, chemical and biological properties influence the fate of pollutants within watersheds, especially from nonpoint sources (NPSs) (Shen et al., 2013a, Singer and Warkentin, 1996). Thus, soil data which is expressed as attribute layers in a GIS format has become an important input for most hydrological and NPSs (H/NPSs) predictions (Geza et al., 2009, Ramos and Martinez-Casasnovas, 2015). However, due to the application of various sampling/mapping methods, soil data-induced prediction error or uncertainty remains a key challenge for the usage of watershed models (Geza and McCray, 2008).

Typically, there exist various soil datasets that are developed by different agents; therefore, one question that raises is how different mapping approaches affect model outputs (Liu and Gupta, 2007). In the United States, the discrepancies between two commonly used soil data sources, the Soil Survey Geographic database (SSURGO) and the State Soil Geographic database (STATSGO), have been researched widely. The impacts of mapping approach have been demonstrated in simulating total flow (Kumar and Merwade, 2009), peak flow (Wang and Melesse, 2006), and nutrients (Geza and McCray, 2008). These studies have demonstrated that the impacts depend on weather conditions, features of study watersheds, and soil type distributions. For example, Geza and McCray (2008) noted that a SSURGO-based model would produce larger simulated flow, sediment and attached nutrient in most cases. However, Kumar and Merwade (2009) showed that a STATSGO-based simulated flow is relatively higher compared with that using the SSURGO data. Moreover, Wang and Melesse (2006) indicated that although a SSURGO-based model results in better performance for predicting total flow, the two soil datasets have comparable impacts on the peak flow simulation. In this sense, additional studies are needed for different conditions to evaluate the relative appropriateness of each soil database.

Another question is regarding the appropriate resolution/scale of soil maps for describing H/NPS processes at the watershed scale. This issue has been raised because high-resolution maps are often costly, time-consuming, and hard to obtain (Peschel et al., 2006, Schwen et al., 2014). The preparation of high-resolution soil data is especially difficult due to the numerous samplings and laboratory analyses required (Kumar and Merwade, 2009, Moriasi and Starks, 2010). Due to the increased availability of high-resolution data, more studies have focused on quantifying the impacts related to the resolution of digital elevation models (DEMs) and land use maps (Lin et al., 2010, Zhang et al., 2014). Few studies have reported the impacts of soil data resolution, focusing on the transmissions of errors into simulated evapotranspiration and soil water storage analyses (Muttiah and Wurbs, 2002). Even fewer researches have noted the impacts on sediment and nutrient exports (Chaplot, 2005). These studies have demonstrated that soil data resolution determines the basic units for describing various soil properties, e.g., the percentages of sand, silt, and clay, hydraulic conductivity, and bulk density, thus data resolution does impact hydrological prediction (Gatzke et al., 2011). For a specific catchment, a larger percentage of clay soils results in more direct runoff, whereas the presence of more sand or silt loam soils decreases runoff amount (Muttiah and Wurbs, 2002). Moreover, the spatially described soil chemical characteristics would have impacts on the fate of pollutants. For example, the initial accumulation of nutrients within soil layers determines the NPS-nutrient loadings from catchment to streams (Lin et al., 2015). To the best of our knowledge, few studies have investigated the impacts of soil data resolution in defining those hydrological process, as well as functions connected with NPSs.

The objective of this study was to undertake a systematic analysis into the impacts of soil data on H/NPS predictions at the watershed scale. In China, the most commonly available soil dataset for watershed models is the state soil geographic map, which is developed by the Nanjing Institute of Soil Science, Chinese Academy of Sciences (CAS-NISS) (Ye et al., 2011). Chinese agricultural agents have also compiled digital soil maps through generalized detailed surveys at the county level, which can serve as local sources of soil properties. To date, few studies have been conducted to quantify the error-transmission from these two types of Chinese soil maps to H/NPS predictions. Thus, the following tasks were performed: 1) these two commonly used Chinese soil maps were compared and their impacts on the H/NPS predictions were quantified; 2) high-resolution soil data were reassembled into coarser ones, and their impacts were evaluated; and 3) error transmission from soil data to the identification of priority management areas (PMAs) was assessed. The case study was performed using the Soil and Water Assessment Tool (SWAT) in the Upper Daning River (UDR) watershed in China.

Section snippets

Watershed description

This study was conducted in the UDR watershed, which is located in Wuxi County (affiliated to the municipality of Chongqing), in the Three Gorges Reservoir Region (Fig. 1). The UDR watershed, with a mean annual temperature of 18.4Ā Ā°C, is characterized by a typical continental monsoon climate in the northern subtropical temperate zone. The annual precipitation ranges within 1030Ā mmā€“1950Ā mm with a mean value of 1182Ā mm, though 78% of the rainfall occurs in summer due to the influences of the monsoon.

Impacts on goodness-of-fit indicator

In this study, five CAB soil datasets, at scales of 1:50,000, 1:100,000, 1:250,000, 1:500,000 and 1:1,000,000, were used as model inputs. The values of calibrated parameters were not changed for this purpose to avoid parameter uncertainty (Yang et al., 2008). As shown in Table 1, the ENS values ranged within 0.71ā€“0.72 for the hydrological prediction, 0.54ā€“0.55 for the sediment simulation, and 0.83ā€“0.80 for the TP modeling. To provide a static state instead of subjective personal judgment, the

Conclusions

In this study, the impacts of soil data on flow, sediment and NPS-TP prediction were quantified. The results indicate that data resolution has limited influence on hydrological and sediment prediction, but the choice of data source greatly impacts on model outputs. Compared with the CAS-NISS data, CAB data results in an underestimation of simulated sediment and TP, and it also generates lesser uncertainty related to hydrological and sediment prediction. From a scientific point of view, the CAB

Acknowledgements

This project was supported by the National Natural Science Foundation of China (nos. 51579011 & 51409003), and the Fund for Innovative Research Group of the National Natural Science Foundation of China (no. 51421065).

References (58)

  • M. Geza et al.

    Quantifying predictive uncertainty for a mountain-watershed model

    J. Hydrol.

    (2009)
  • L.T. Ghebremichael et al.

    Integrated watershed- and farm-scale modeling framework for targeting critical source areas while maintaining farm economic viability

    J. Environ. Manag.

    (2013)
  • K. Lin et al.

    An evaluation of impacts of DEM resolution and parameter correlation on TOPMODEL modeling uncertainty

    J. Hydrol.

    (2010)
  • X.X. Lu et al.

    Seasonal changes of nutrient fluxes in the Upper Changjiang basin: an example of the Longchuanjiang River, China

    J. Hydrol.

    (2011)
  • X. Ma et al.

    Assessment and analysis of non-point source nitrogen and phosphorus loads in the Three Gorges Reservoir Area of Hubei Province, China

    Sci. Total Environ.

    (2011)
  • M. Munafo et al.

    River pollution from non-point sources: a new simplified method of assessment

    J. Environ. Manag.

    (2005)
  • R.S. Muttiah et al.

    Scale-dependent soil and climate variability effects on watershed water balance of the SWAT model

    J. Hydrol.

    (2002)
  • O. Salazar et al.

    Application of an integrated framework for estimating nitrate loads from a coastal watershed in south-east Sweden

    Agric. Water Manag.

    (2013)
  • C. Santhi et al.

    A modeling approach to evaluate the impacts of water quality management plans implemented in a watershed in Texas

    Environ. Model. Softw.

    (2006)
  • A. Schwen et al.

    Vertical variations of soil hydraulic properties within two soil profiles and its relevance for soil water simulations

    J. Hydrol.

    (2014)
  • Z. Shen et al.

    Impact of spatial rainfall variability on hydrology and nonpoint source pollution modeling

    J. Hydrol.

    (2012)
  • Z.Y. Shen et al.

    Assessment of nitrogen and phosphorus loads and causal factors from different land use and soil types in the Three Gorges Reservoir Area

    Sci. Total Environ.

    (2013)
  • Z.Y. Shen et al.

    A comprehensive study of the effect of GIS data on hydrology and non-point source pollution modeling

    Agric. Water Manag.

    (2013)
  • Z.Y. Shen et al.

    Identifying non-point source priority management areas in watersheds with multiple functional zones

    Water Res.

    (2015)
  • M.J. Singer et al.

    Soils in an environmental context: an American perspective

    Catena

    (1996)
  • M. Strauch et al.

    The impact of best management practices on simulated streamflow and sediment load in a Central Brazilian catchment

    J. Environ. Manag.

    (2013)
  • R.P. Udawatta et al.

    Soil enzyme activities and physical properties in a watershed managed under agroforestry and row-crop systems

    Agric. Ecosyst. Environ.

    (2009)
  • Y. Wu et al.

    Modeling of soil erosion and sediment transport in the East River Basin in southern China

    Sci. Total Environ.

    (2012)
  • S. Wu et al.

    Modeling the effects of elevation data resolution on the performance of topography-based watershed runoff simulation

    Environ. Model. Softw.

    (2007)
  • Cited by (0)

    View full text