Agricultural systems are a critical component of this infrastructure, providing crucial food products as part of the human food system, but also providing vital agricultural by-products as inputs for other industrial systems. They are deeply coupled with the natural environment and undergird human civilization. As this web of systems — industrial, agricultural, environmental, human — grows more intricate, it becomes increasingly difficult to understand and map out the consequences of our actions. Agricultural systems are, thereby, commonly assessed to improve economic performance, reduce environmental impacts, and improve social sustainability. System assessment begins with the representation of a system through some modeling process, and a subsequent evaluation of the model with respect to certain attributes. This process involves the collection, structuring, manipulation, and storage of a variety of data. System assessment often involves multiple modeling processes as each model is created to represent a particular aspect of a system and its performance. While instrumental in im-proving our understanding of these complex systems, these efforts have led to a fragmented field with tensions and pulls in different directions, and have resulted in a duplication of effort, and moreover, data and processes that are disconnected. Over the past decade, small- to medium-scale sustainable farms have been growing in popularity. There is an increasing demand for food that is grown sustainably, raised humanely, and produced with fair and just labor practices. This has spawned many efforts to curtail environmental impacts: including, numerous regulations, eco-labels, indoor cannabis grow system and certifications. Such efforts often require farmers to engage in additional record keeping that can be both tedious and time consuming.
Farmers, in essence, pay for the privilege of growing food sustainably. Despite growth in this sector, there is a lack of appropriate technological tools, arising in part from a mismatch between existing tools — which are typically designed for large-scale industrialized agricultural practices — and the needs of farmers who work at smaller scales. With this comes a growing need for systemic mechanisms to understand, analyze, manage, and further improve, sustainable agricultural systems. There is no dearth of analyses of the environmental impacts of our food systems, but there is a lack of connectivity across the plethora of models created and a lag with the changing real world that is represented.Agricultural systems are composed of heterogeneous subsystems with varied environmental impacts. For example, Figure 1.1 shows Alegria Fresh Farms, an urban farm in Orange County, California. It is a small-scale farm that produces fruits, vegetables and a variety of agricultural byproducts such as organic soil. It is not a single product system. A number of subsystems are present within the farm for irrigation, solar energy production, hydroponic cultivation, hydroponic vertical cultivation, certified organic cultivation, vermicomposting, and a nursery, among others. To understand which subsystems are more responsible for certain environmental impacts, one would need to be able to attribute impacts to particular subsystems and processes.This would show the farmer which cultivation method, for example, is more environmentally friendly. The hydroponic vertical cultivation subsystem shown in Figure 1.2 exemplifies the complexity of modern agriculture, even at this small scale. Many of the materials described are sourced externally from other industrial systems. Many external data are therefore required to accurately assess the environmental impact of the farm.
Availability and access to the environmental impact data of these external components is not guaranteed. Agricultural systems are also constantly evolving: equipment is upgraded, cultivation methods are refined to optimize certain metrics , subsystems grow and shrink, and food types grown change according to supply and demand, in addition to season. Figure 1.3 shows two satellite images of Alegria Fresh Farms: the image on the left is the farm shortly after it was set up. Alegria Fresh Farms was one of a set of small experimental and demonstration farms that were introduced to the Orange County Great Park in around 2009. The image on the right was taken after several years of farm activity. The farm has since been relocated. As changes are made in an agricultural system, any models that were initially created become outdated. When models are constructed for agricultural systems, they are a static representation of a dynamic system. Once data have been collected and things change, the model is no longer an accurate representation of the system. Over the last six years, the Alegria Fresh Farms has grown, systems have changed, and as a result, the relationship of the farm with the environment has been affected. However, many models, from satellite images to layout diagrams, cannot necessarily capture the change in the system and its en-vironmental impacts. While there are many techniques that allow for detailed analyses of an agricultural system’s environmental performance, they are primarily expert-driven technique, which means that once a model is created, it is difficult for the farmer to maintain or update it for continued use.Software systems for agricultural modeling are primarily designed for activities such as farm simulation and yield maximization.
These can be both computationally expensive. Such work, while interesting, relevant, and timely, primarily addresses the modeling and simulation of agricultural systems to improve crop yields and system management, with only a marginal focus on environmental assessment of these systems. Agricultural Ontologies: The Food and Agriculture Organization of the United Nations has maintained the AGROVOC project since the 1980s. It is a controlled vocabulary, designed for, and used by, information management professionals . It consists of over 32,000 agricultural concepts, gleaned from publications, research artifacts, and external thesauri. Similarly, the United States Department of Agriculture’s National Agricultural Library has provided a more America-centric glossary and thesaurus service since 2002. These initiatives are part of a broader goal within the agricultural information management community to standardize agricultural terms, concepts, and data. There are also several efforts — in early stages of development — to extend tradition agricultural thesauri into fully developed ontologies for use at the farm level to inform crop production, “foods-for-health” knowledge systems in the nutrition space, and to augment precision agriculture with plant-driven decision making capa-bilities. The AGROVOC team is also developing an ontology service in anticipation of semantic web requirements. Crop Modeling: Work in crop modeling tends to focus on operational and yield optimization. For example, Honda et al. present a service platform that uses a network of field sensors to obtain real-time field data to plan large-scale field operations e.g. fertilizer application. Meanwhile, Ponti et al. used statistical models to assess the yield gap between organic and conventional agriculture to obtain a deeper understanding of the range of performance in agriculture. Agricultural Simulation: Agriculture researchers often use simulations to understand and make predictions about the performance of various agricultural systems, while computation researchers apply their expertise to improving the performance of simulations in agriculture. For example, the Agricultural Production Systems sIMulator project is a modular simulation tool that allows for the investigation of relationships between plants, animals, soil, climate, and management involved in agricultural systems. Papajorgji et al. present different ways in which model-driven architecture, in particular the use of the Unified Modeling Language, can be leveraged to improve crop simulation models. Miralles & Libeourel study how Geographic Information System models can be brought to bear on crop simulation to allow for better integration of weather data. Life Cycle Assessment: LCA is a modeling technique used to assess the environmental impacts of products and the processes by which they are constructed. Farmers, along with environmental analysts, can conduct LCAs to quantify the environmental impacts of resource flows in a system, vertical grow racks and subsequently make improvements in the farming processes to reduce undesired impacts. Many software systems and databases exist to support LCAs, most of which are domain agnostic . Models of agricultural products and production systems are commonly created in the academic community using LCA. Examples of systems that have been successfully analyzed using LCA include: a large American feedlot where beef is produced; a soybean meal production chain that spans from Argentina to Denmark; and an Australian corn chips production chain that begins at the corn farm, and ends at the consumer-ready packet. While LCA models are intended to provide efficient and convenient access to information about the environmental performance of production systems, such as agricultural systems, there is a disconnect between the current LCA modeling process, the needs of sustainable farmers, and the systems represented.
The Bruntland report presents an umbrella definition of sustainable development as, “the ability of humanity to ensure that it meets the needs of the present without compromising the ability of future generations to meet their own needs”. While this definition is commonly used in sustainability research, it is too broad to allow us to engage in grounded, sustainable, human computer interaction work, or other forms of sustainable design activities. We often come away from such discussions with difficulty envisioning specific applications to design and actionable interventions to pursue. There have been many explorations of sustainability in the context of HCI, ICT for Development, and the coordination literature. However, just as most “sustainability-oriented work takes place outside HCI” , much design work for sustainable agriculture also occurs at the periphery of applied computing research. In this section, some of the more promising explorations within this periphery are reviewed.HCI and Agriculture: Interest in the intersection of HCI, design, and agriculture is growing. For example: Raghavan et al. recently suggested use of computation to design better agro-ecological systems, and Frawley & Dyson created non-human animal personas to enhance welfare in animal agriculture. For the most part, design for agriculture tends to be for specific subsets of agriculture. For example, Chinthammit et al. ran a Software Interest Group meeting, looking at “HCI in Food Product Innovation”. Whether it is design ideations specifically for urban residential gardeners or the design of platforms to assist in the creation of backyard permaculture systems, Di Salvo et al. point out that “there is a significant gap between the professional fields of industrial and interaction design and design research in sustainable HCI” . In an attempt to bridge this gap, my collaborators and I ran a workshop on at the ACM CHI conference in 2017 on “Designing Sustainable Food Systems”. We aimed to bring together HCI researchers, designers, and practitioners to explore, design for, and reflect on, opportunities for the HCI community to engage in creating more a sustainable food system. Coordination and Collaboration in Agriculture: Food production is an inherently collaborative process, with many stakeholders involved and varying organizational configurations across system types. Examples include: a qualitative study looking at coordination challenges in organic farm families, and an early warning management technique to enable collaboration among rice farmers participating in small-scale precision agriculture. Such work provides valuable context for understanding how the interplay between stakeholders on farms affects tool design. Farm Management: At the time of this writing, the Information Technology startup community began focusing on the design of tools for agriculture. Earnest examples include: precision farming tools, precision agriculture tools requiring specialized hard-ware ; daily farm management tools aimed at agribusiness; inventory management tools; and tools that provide analytics.Agricultural Environmental Policy: Environmental regulations have serious implications on farm-level assessment, data collection, and record keeping that farmers in California must engage in [38]. Regulations, both state-imposed, and federal, typically require some form of record keeping and form filling, and in some cases, the presentation of these documents during site inspections. Environmental policies that California farmers are subject to include: water, pesticides, and emissions regulations. In addition, farmers may participate in voluntary programs to demonstrate commitment to environmental protection, such as the environmental stewardship program. Environmental Labeling and Certification: The primary goal of environmental labeling and certification agencies is to provide farm quality assurance. These labels and certificates provide the purchaser, whether broker, retailer, or consumer, with the validation that a farm meets a particular standard set out by the certifying agency. The efficacy and usefulness of labels has long been debated from the early days of dolphin-safe labels to the present-day quandary of GMO-labeling. Nevertheless, many farms actively pursue labels and certifications, necessitating an additional layer of record keeping and data collection. There are four main types of labeling and certification in sustainable agriculture: government regulated National Organic Program; non-profit regulated ; retailer specific ; and product-specific .