Developing an integrated technology-environment-economics model to simulate food-energy-water systems in Corn Belt watersheds

https://doi.org/10.1016/j.envsoft.2021.105083Get rights and content

Highlights

  • Selected data techniques are applied to develop suitable surrogates for different process models.

  • The integrated Technology-Environment-Economics model based on surrogates is computationally tractable.

  • Insights on surrogate-based model coupling and tradeoffs of integration design choice are discussed.

  • The integrated model is demonstrated for FEW systems in the Corn Belt.

  • The sensitivity of 16 key parameters from component models on the integrated model is evaluated.

Abstract

To facilitate understanding and decision making in the food-energy-water (FEW) nexus context, we develop an integrated technology-environment-economics model (ITEEM) at a watershed scale. ITEEM is built as an integration of various models, including models for grain processing, drinking water treatment, and wastewater treatment (technology); a watershed model for hydrology, water quality, crop production, and nutrient cycling (environment); an economics model assessing total benefit, including non-market valuation of environmental benefits. Different data techniques are applied to develop suitable surrogates for computer-based models, including a response matrix method, artificial neural networks, and lookup tables. Empirical equations are applied to develop models of economics and drinking water treatment. The input-output relationships between the models are formulated in a unified computational framework. ITEEM, a spatially semi-distributed dynamic simulation model, can be used to quantify the environmental and socioeconomic impacts of various management practices, technologies, and policy interventions on FEW systems in the Corn Belt.

Introduction

Food-energy-water (FEW) systems in the US Corn Belt are highly interconnected and sensitive to stresses and threats. Grain production and subsequent utilization for animal feed, human food, and ethanol production have pervasive effects on water quantity and quality in downstream environments both locally (e.g., lakes and rivers with elevated nitrogen and phosphorus) and nationally (e.g., Hypoxic zone in the Gulf of Mexico) (US EPA, 2017). Water stress associated with increased climatic variability is anticipated to increase (Muttiah and Wurbs, 2002), especially in many mid-sized cities in the Corn Belt that interact with neighboring agricultural lands, major industrial needs (Li et al., 2018), and their shared watersheds. Energy demand and overall costs for wastewater and drinking water treatment have increased, and this trend is expected to be exacerbated by continued expansion of food and bioethanol production (Simpson et al., 2008; Twomey et al., 2010). To deal with these threats to and risks within FEW systems, long-term efforts have been made to resolve the conflicts between agriculture, food industry, water supply, and environmental protection. For example, wastewater treatment and corn ethanol refinery facilities have begun extracting nutrients from “waste” and process byproducts, which results in both the reuse of extracted materials as inorganic mineral fertilizers (e.g. struvite and calcium phytate) and the reduction of point-source discharge of nutrients to the environment. For example, recovering phosphorus (P) can conserve a finite resource (e.g. phosphate rock) (Cordell et al., 2009; Juneja et al., 2019; Margenot et al., 2019); cost-effective water treatment technologies are adopted to conserve energy use (Bhatnagar and Sillanpää, 2011); agricultural best management practices (BMPs) reduce nutrient and soil loss from farmland in upstream watersheds (Lemke et al., 2011; Rao et al., 2009). Researchers have called for holistic integrated modeling development and assessment for FEW systems at various scales to avoid fragmented status quo decision making (Leck et al., 2015; Little et al., 2019). This paper presents an integrated technology-environment-economics model (ITEEM) which unites a set of surrogates and empirical models derived from the various primary models simulating key processes at a watershed scale. The developed ITEEM is capable of analyzing complex systems and specific solutions to interconnected problems in FEW systems in Corn Belt watersheds.

There are several major challenges when integrating models from different disciplines. First, most physical models are developed using discipline-specific computer programs or software packages (e.g., SWAT for hydrologic processes, GPS-X for wastewater treatment), which causes a barrier for automatic information transfer. Recently, some interfaces have been developed for simple automated data exchange between two models (Anderson et al., 2018; Xiang et al., 2020). For a large interdisciplinary integrated model involving agricultural, hydrologic, and engineering components developed in various computer programs (including commercial software), as the case of our study, the level of complexity can be overwhelming to modelers, and it usually turns out to be infeasible to directly integrate different models due to incompatibilities among discipline-specific computer programs (Little et al., 2019). Second, some engineering design models (e.g., GPS-X for wastewater treatment, SuperPro Designer for Grain processing) are proprietary which may impose costs and legal constraints on direct coupling. Third, inputs and outputs from separate models are likely to have different temporal and spatial scales with distinct data formats, which need to be harmonized at the points of interaction between models (Cai, 2008). Appropriately building the interactions between various models is a key step to enable information transfer endogenously within a consistent model. Fourth, complex physical models can be highly computationally expensive; an affordable computational burden is especially important if the research of interest will address stochastic problems (Little et al., 2019). Thus directly integrating many computationally heavy models is often computationally infeasible.

Researchers have developed various integrated models (Cai, 2008; Carmichael et al., 2004; Gaddis et al., 2010; Housh et al., 2014). Cai (2008) shared reflective comments on the advantages and challenges of holistic modeling (tight coupling of different components in one consistent model) versus compound modeling approaches (“loose” coupling of different components via external data exchanges). Holistic models embed different components into a single consistent optimization model, such as hydrologic-economic models (Cai, 2008; Cai et al., 2003; Harou et al., 2009), hydro-biogeochemical model (Wu et al., 2016), and “system of systems” models, e.g. a biofuel (biomass and refinery)-infrastructure-environment model (Housh et al., 2014). Holistic optimization models are usually composed of mathematical equations including the objective function(s) and constraint function(s). Other system modeling approaches applied in FEW systems include agent-based models (Ng et al., 2011), life cycle assessment (Li et al., 2020), system dynamics (Feng et al., 2016; Gaddis and Voinov, 2010), etc. However, these system modeling methods have less focus on integrating detailed physical process modeling, but more focus on other perspectives. For example, agent-based models focus on simulating the behavior and decision-making of multiple stakeholders, life cycle assessment focuses on quantifying environmental impacts from cradle to grave, and system dynamics focuses on modeling the feedbacks among stock variables and drivers. The degree of process details at which those system modeling approaches have may not lend themselves to coupling multiple complex process models in a system of systems.

Little et al. (2019) proposed a generic tiered system of systems (GTSoS) to upscale physical models from the process level to the system level via integration while keeping computational tractability and minimizing the loss of fidelity (Little et al., 2019). Models that are developed at the process level in various computer programs (or software packages) with domain-specific knowledge and data can be replaced by surrogates (also termed reduced-order models, meta-models, or emulators), if process models cannot be integrated directly due to complexity and incompatibilities among discipline-specific computer programs. Various data techniques can be applied for emulating a process model, such as polynomial response surfaces, artificial neural networks, and supporting vector machine using numerical samples of inputs and outputs of the primary model under a systematic sampling strategy (Leperi et al., 2019; Lu and Ricciuto, 2019). Those surrogates typically build statistical relationships between inputs and outputs of a system modeled by a primary model. Another type of surrogate is based on hybrid theory and data (also termed as lower-fidelity physically-based surrogates) (Razavi et al., 2012). By replacing complex process models with appropriate surrogates, one can integrate them into a consistent model, maintaining reasonable fidelity of the primary process models without causing a serious computational burden, as most surrogates do not have a rich internal structure (Carmichael et al., 2004).

Although the GTSoS framework provides a promising direction on model integration for analyzing a system of systems, the development of such a framework is challenging. Specific challenges include the selection of an appropriate mathematical form of a surrogate for a particular process model, and the integration of the surrogates across multiple spatial and temporal scales (Cai, 2008). In addition, examples of real-world problems are needed to demonstrate the effectiveness and applicability of GTSoS to the various complex system modeling cases. Here, we explain the methodology used to overcome these challenges in the construction and execution of ITEEM. Disciplinary-specific process models are replaced by surrogates, and these surrogates are integrated within a unified computational software framework to form a holistic model. ITEEM is demonstrated in a watershed in the Corn Belt to analyze inter-connected problems of crop production, grain processing, water and wastewater treatment, and nutrient management, with consideration of technologies, management practices, and policies for multiple sectors.

Section snippets

Research problem

FEW systems are usually highly interconnected crossing multiple sectors.in many regions. For the Corn Belt watersheds, FEW systems are sensitive to stresses and threats with respect to food production, and increasing biomass production, and energy supply and demand, which pose impacts on water quality, water supply, energy demand and cost, resources conservation, and economic growth and financial stability. These interconnected components of FEW systems are depicted in Fig. 1. Managing

Development of ITEEM

Via multi-disciplinary teamwork, three process models (SWAT, WWT, and GP) and empirical (DWT) or theoretical-empirical models (Economics) are first established at the process level (the lower part of Fig. 2). Then the components of ITEEM are developed in the form of surrogates or empirical relationships, which are coupled by integrating input and output relationships crossing temporal and spatial scales at the interaction points between the components at the system level (the upper part of Fig.

Computational implementation of ITEEM in object-oriented programming platform

A coherent computational framework is developed to link and execute models from individual knowledge domains in an orderly manner. Standards of integrated modeling have been promoted by researchers to produce a useable and low-friction simulation environment, such as the Community Surface Dynamics Modeling System (CSDMS) project by Peckham et al., in 2013. The design criteria include but not limited to support of multiple operating system, use of open-source tools rather than proprietary

Demonstration of ITEEM in the Upper Sangamon River Watershed, Illinois

We demonstrate ITEEM via a testbed watershed, the Upper Sangamon River Watershed (USRW). Different scenarios are tested to explore a portfolio of alternative engineering technologies, policies, and BMPs; the results of the scenarios are compared to a baseline scenario in terms of multiple FEW systems indicators.

Conclusions and future research

Addressing large-scale environmental sustainability challenges requires integrated analysis of complex inter-relationships within FEW systems. This paper presents the development of an integrated technology-environment-economics model (ITEEM) for typical watersheds in the Corn Belt. We use various data techniques to convert complex models simulating physical & engineering processes and socioeconomic relationships into computationally tractable surrogates and link these surrogates via

Data and code availability

The data and codes are available from the corresponding author upon request.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the US National Science Foundation (INFEWS/T1 award number 1739788). We are grateful to industrial, governmental, and agricultural stakeholders in Decatur, IL for providing valuable data for developing components of the ITEEM. We appreciate valuable comments and constructive suggestions from the editors Daniel P. Ames and Tatiana Filatova, and three anonymous reviewers, which have considerably improved the quality of this work.

References (51)

  • S.D. Peckham et al.

    A component-based approach to integrated modeling in the geosciences: the design of CSDMS

    Comput. Geosci.

    (2013)
  • N.S. Rao et al.

    Modeling watershed-scale effectiveness of agricultural best management practices to reduce phosphorus loading

    J. Environ. Manag.

    (2009)
  • Y. Wu et al.

    SWAT-DayCent coupler: an integration tool for simultaneous hydro-biogeochemical modeling using SWAT and DayCent

    Environ. Model. Software

    (2016)
  • Z. Xiang et al.

    DSSAT-MODFLOW: a new modeling framework for exploring groundwater conservation strategies in irrigated areas

    Agric. Water Manag.

    (2020)
  • N. Alemazkoor et al.

    Fast probabilistic voltage control for distribution networks with distributed generation using polynomial surrogates

    IEEE Access

    (2020)
  • R.A. Bagnold

    Bed load transport by natural rivers

    Water Resour. Res.

    (1977)
  • E.M. Bennett et al.

    Human impact on erodable phosphorus and eutrophication: a global perspective

    Bioscience

    (2001)
  • A. Bhatnagar et al.

    A review of emerging adsorbents for nitrate removal from water

    Chem. Eng. J.

    (2011)
  • X. Cai et al.

    Integrated hydrologic-agronomic-economic model for river basin management

    J. Water Resour. Plann. Manag.

    (2003)
  • X. Cai et al.

    Strategic planning for drought mitigation under climate change

    J. Water Resour. Plann. Manag.

    (2015)
  • S.R. Carpenter

    Phosphorus control is critical to mitigating eutrophication

    Proc. Natl. Acad. Sci. U.S.A.

    (2008)
  • E.J.B. Gaddis et al.

    Spatially explicit modeling of land use specific phosphorus transport pathways to improve TMDL load estimates and implementation planning

    Water Resour. Manag.

    (2010)
  • S.M. Gorelick

    A review of distributed parameter groundwater management modeling methods

    Water Resour. Res.

    (1983)
  • J.J. Harou et al.

    Hydro-economic models: concepts, design, applications, and future prospects

    J. Hydrol.

    (2009)
  • R.W. Harris et al.

    Process Design and Cost Estimating Algorithms for the Computer Assisted Procedure for Design and Evaluation of Wastewater Treatment Systems (CAPDET). Environmental Engineering Division, Environmental Laboratory, U.S. Army Engineer Waterways Experiment Station

    (1982)
  • Cited by (19)

    • Hydrological connectivity affects nitrogen migration and retention in the land‒river continuum

      2023, Journal of Environmental Management
      Citation Excerpt :

      The main method for identifying CSAs is spatial modeling using geomorphological and hydrological information (e.g., soil and water assessment tool). However, the modeling process is long and complex (Li et al., 2021). Some studies have also shown that CSAs fall short in defining the transport capacity of pollutants by surface runoff (Giri et al., 2017; Thomas et al., 2016).

    • Integrating fourth industrial revolution (4IR) technologies into the water, energy & food nexus for sustainable security: A bibliometric analysis

      2022, Journal of Cleaner Production
      Citation Excerpt :

      The fourth industrial revolution is changing production, organizational processes, and economic directions in nations, necessitating a strategic approach to evaluate the extent to which the technologies of the revolution affect three vital resources (water, energy & food) to the human economy. Qi, et al. (2021); Li et al. (2021) and Sun et al. (2021) opined that these three resources are critical in ensuring the socio-economic development of any nation. The United Nations stated that the instability relating to these resources is considered a global risk (Li et al., 2016).

    • A food-energy-water nexus meta-model for food and energy security

      2022, Sustainable Production and Consumption
      Citation Excerpt :

      It might be more beneficial to link existing sectoral models rather than to expand a single one. Model coupling of different FEW sectors is often challenging because of the diversity of inputs, the different spatiotemporal scales of operation, and the different programming languages used (Tabatabaie and Murthy, 2020; Li et al., 2021; Little et al., 2018; Cai, 2008). The FEW-N has led to the development of new tools that integrate all three sectors within one framework.

    • A modified response matrix method to approximate SWAT for computationally intense applications

      2022, Environmental Modelling and Software
      Citation Excerpt :

      Excluding sediment, the approximations maintain an RM-based approach, facilitating advantages such as the feasibility of linear programming methods. The approximations are especially well-suited to integration within a system of systems modeling framework where modelers wish to consider a mixture of non-point and point source models (Li et al., 2021a). For example, the nutrient effluents simulated by wastewater treatment models can be directly added via the RM method to accurately simulate the in-stream P load.

    View all citing articles on Scopus
    View full text