NAE Logo
In This Issue
Multiscale Sensing and Modeling Frameworks: Integrating Field to Continental Scales
Agriculture and Smarter Food Systems
The Impact of Mechanization on Agriculture
How Smart-IT Systems are Revolutionizing Agriculture
Information Technology and Agriculture: Global Challenges and Opportunities
Download PDF
Fall Issue of The Bridge on Agriculture and Information Technology
September 15, 2011 Volume 41 Issue 3

Agriculture and Smarter Food Systems

Thursday, September 15, 2011

Author: Matthew Denesuk and Susan Wilkinson

Requirements for greater transparency about where food comes from and how it is treated along the way are becoming more stringent.

The growing world population is putting increasing strains on natural resources, including agricultural resources, and recent economic and environmental trends are making the problem even more acute. The unprecedented growth of the global middle class is accelerating demand for foods with higher agricultural footprints, such as meats. At the same time, limited arable land areas, increasing water shortages, rising energy costs, and urgent climate/environmental concerns are creating a desperate need for substantial increases in factor productivities.1

In addition, in response to issues related to public health and food supply chains, requirements for greater transparency about where food comes from and how it is treated along the way are becoming more stringent. To meet those requirements, we must find ways to quickly trace the sources of contamination or other health-related issues.

The Transformational Potential of “Physical Meets Digital”

In many domains of human activity today, the infusion of information technology (IT)-enabled “intelligence” capable of driving major improvements in systems-level productivity and performance is accelerating. These improvements will have the most profound impact on “highly physical” domains, that is, domains that involve large-scale infrastructure (e.g., machinery, large vehicles, static structures, and productive land masses) and that are often exposed to natural elements. Examples include traffic systems, utilities management (water or electricity), transportation and shipping, mining, oil and gas, and of course agriculture and the broader food system. In these cases, the use and performance of such infrastructure tend to substantially determine overall system performance.

To varying degrees, many aspects of these kinds of systems have traditionally been resistant to the infusion of intelligence, due largely to the difficulty and cost associated with (1) obtaining detailed, timely data about the system components and surrounding environments and (2) remotely controlling actions based on analyses of those data.

However, manufacturers of heavy equipment are increasingly outfitting their products with deeper levels of instrumentation—various types of transduction capabilities that produce data reflecting the status of the equipment, potentially enabling management or operational changes to be performed remotely. In addition, technologies for generating data that reflect environmental conditions (e.g., in situ sensors, remote sensing satellites, airborne vehicles) are becoming increasingly capable and cost effective. Combined with cost-effective communications and interconnection technologies, comprehensive data reflecting the state of the infrastructure and environment can now be collected with tolerable latencies, enabling effective actions to be executed.

In addition, technology for effectively instrumenting “products,” such as food items, pharmaceuticals, and so on, is also becoming more capable and less expensive. Think, for example, of the increased availability of radio-frequency identification and 2D barcode technology, wireless environment/gas sensors in shipping containers, and global positioning system and other location-tracking technologies.

Smarter Agriculture

Broadly speaking, IT is increasingly impacting agriculture and the overall food system in myriad ways throughout the value chain. Impacts range from fundamental inputs, such as genomics and computer modeling that can help drive the next generation of seed and planting technology, to food distribution, such as smarter logistics that can help deliver food more quickly using less fuel and fewer machine resources and with less spoilage en route to the point of consumption.

The clear social need and inevitably growing market demand are driving investment by governments, IT firms, and private equity and venture capital firms. Many venture capital investments have focused on aspects of “physical meets digital,” such as traceability, sensing for reducing spoilage, and sensing for agricultural optimization and/or water productivity.

“Agriculture 2.0,” which implies decisive shifts in the ways and means of agricultural production, is becoming a term of art in this context. Because venture capital tends to flourish in environments that are experiencing market or business model disruptions, such investments should be watched closely.

The present article focuses predominantly on two related areas (see Figure 1) in which “smarter” systems enabled by physical-digital integration can have a positive and global impact: (1) track-and-trace technologies to support food safety and ultimately optimize food supply chains; and (2) increasing farm multifactor productivity by improving water logistics and application, optimizing machine/fleet maintenance, and improving farm operations/processes. Multifactor productivity includes agricultural business optimization as a result of the integration of these approaches with additional downstream business information.

Figure 1

Food Tracking and Tracing

The increasing focus on food safety and security is being driven by a number of factors: (1) the number and/or visibility of food safety-related incidents (Figure 2); (2) the increasing globalization and complexity of the food supply system, which contributes to both higher risks associated with the number and diversity of handling entities and diminishing visibility/transparency. The latter effect produces uneasiness because of uncertainties about where food comes from, which greatly complicates efforts to trace problems back to their sources; and (3) fears of terrorism or other deliberate attempts to damage or contaminate food supplies.

Figure 2

The purpose of food tracking and tracing is to be able to follow food and food components as they flow through and are transformed along various pathways in the value chain, and, when needed, to be able to follow the flow backward to identify the sources and/or conditions to which the food was subjected along the way. The latter capability is a critical component of food safety systems.

Tracking and tracing capabilities in food systems can be complicated and require coordination among many firms and functions, which may have substantially different business motivations. Broadly speaking, motivation can be compliance driven, that is, dictated by some form of government regulation or customer requirements, or based on business value, that is, focused on branding, overall risk mitigation, or improving value-chain operational performance (e.g., by cross leveraging tracking and tracing capabilities).

One clear benefit of tracking and tracing that applies throughout the value chain is the ability to minimize disruptions caused by a food-safety event. For example, if the cause of the event can be isolated to a particular food from a particular part of a particular farm, the problem can be effectively quarantined, and food sales should return to normal levels as customer confidence recovers. The alternative, which frequently occurs, is that consumers stop buying the type of food (e.g., cucumbers) initially suspected as the cause. Even though the initial suspicion often turns out to be incorrect, sales of many types of food can be adversely impacted as health officials and food-chain participants work to find the culprit.

Even after the culprit is found, it frequently takes a substantial amount of time to identify the location of, and remove the contaminated product from, the market. The longer this takes, the more consumer confidence is eroded, and the longer it takes for sales to return to normal.

Standards and Regulations

Under the Food Safety Modernization Act, signed by President Obama in January 2011, the Food and Drug Administration is required to execute traceability pilot tests and use what they learn to develop new regulations to improve the ability of the food industry to identify the source of food-safety problems and quickly remove contaminated products from the market. For tracking and tracing to succeed, however, there must be a willingness to share information across the supply chain, as well as agreements among trading partners on standardized expressions of key data elements.

GS1 standards could be implemented throughout the food supply chain to enable traceability. GS1 is a not-for-profit organization dedicated to the design and implementation of global standards for identifying goods and services to improve the efficiency and visibility of supply chains. There are GS1 member organizations in 108 countries, and their well-known global trade item numbers (GTINs), including UPC (Universal Product Code), SSCC (Serial Shipping Container Code), and EAN (European/International Article Number), have been used by retailers and suppliers of packaged goods for decades. The adoption of GS1 standards varies by country and sector but has increased significantly every year, and efforts are under way to increase adoption by companies in the upstream supply chain.

GS1 standards for product identification (product type and lot numbers) are the basis of a major initiative undertaken by the produce industry to enable traceability back to the farm. The goal of the initiative, called the “Produce Traceability Initiative” (PTI), is to achieve adoption of electronic traceability throughout the supply chain for every case of produce by 2012. Although participation in PTI is currently voluntary, U.S. food retailers and their major produce suppliers are actively moving toward compliance.

Execution Issues

Tracking and tracing capability has three major pillars: (1) establishing a premise ID; (2) establishing a product ID; and (3) establishing a means of tracking movements and transformations (Figure 3).

Figure 3

Establishing a premise ID. A unique identifier is required for the source location and for each location an agricultural good passes through on its journey from the farm to the retail outlet. This identifier must be at a level of temporal and spatial granularity appropriate for the purpose and must include the identity of the firm or organization that owns or operates each premise.

On the compliance level, the identifier may, for example, relate to produce from a particular section of a field in a window of several weeks. On the value level, it may include information related to more sustainable or socially responsible practices on the farm, special handling, and so on.

Establishing a product ID. The product ID relates to the type and batch of agricultural good produced. On the compliance level, it might relate to a particular type and subtype of vegetable linked to a particular premise, such as a field-packed box of organic iceberg lettuce from a particular section of a particular farm in central California. On the value level, it might include information on the seeds used, irrigation characteristics, and more precise time stamps.

Establishing a means of tracking movements and transformations. As food moves through the value chain, it is physically transported in various types of containers and through various local environments (temperature, humidity, various gas concentrations); it sits for varying periods of time at transition points; and it is transformed in various ways (e.g., blending, canning, freezing, preserving). Thus, maintaining the integrity of the data is a challenge. On the compliance level, it may include tracking points of origin and entry, type and ID of transport, basic information about transformation, and basic information about dates and times. On the value level, it may include environmental characterizations (especially temperature history) and geospatial history.

Implementation Issues

Government regulations and the requirements of business customers will encourage a minimum level (at least) of tracking and tracing capabilities. However, although the deployment of advanced tracking and tracing technology can significantly improve supply chain and other operational efficiencies, adoption has been slow, largely because of the number and complexity of food and food-component pathways and the concomitant need for multiple levels of process alignments and information sharing (including adoption of a consistent semantic model and a system for conveying relevant information and protecting otherwise sensitive information).

In addition, the governance mandate for agriculture and the food supply industries is often split among jurisdictions (e.g., states and provinces) and national bodies. As a result, solutions must support concurrent jurisdictional and national policies.

Therefore, it is essential that all stakeholders be identified, that relative benefits be defined, and that buy-in and cooperation be established. Although all parties will benefit to some degree from risk mitigation and reduced liability, other benefits will vary substantially by stakeholder. Key benefits beyond compliance with regulations may include increased access to global markets, better quality control, improved demand visibility and forecasting, and improved brand image and increased sales.

For incremental adoption, it is important that one define “slices” of a hypothetical comprehensive tracking and tracing system that cover enough space to provide significant benefits to all stakeholders, but are also contained enough to be manageable (and financeable). It is often helpful to define such slices using a so-called “cube model” that has three dimensions: (1) food type; (2) value-chain position; and (3) geographical/regulatory region (e.g., state, province). Manageable slices can often be defined in which one dimension is fixed and the other two vary across the relevant values (e.g., packaged leafy vegetables from all relevant regions from inputs through retail).

Another issue relates to the classic economic “agency problem,” that is, that costs and benefits can accrue substantially differently in different areas of the value chain. In practice, costs may be borne disproportionately by food producers/farmers and early processors relative to later processors, distributors, and retailers.

There are signs, however, that the business ecosystem is evolving as third parties increasingly provide technology and process components, such as low-cost labeling technology, tracking-infrastructure-as-a-service, and food environmental monitoring and response capabilities. By leveraging scale and specialization, these changes can lower adoption and management costs.

Country and Regional Issues

Structurally, food systems are globally distributed and integrated in a variety of ways. Different structures reflect economic and policy characteristics associated with the end destination, as well as with characteristics associated with production and distribution points along the way.

For example, food in the United States can come from farms nearly anywhere in the world through a variety of distribution pathways with distinct profiles for risk of contamination, spoilage, and additions to cost. These pathways cross through many different countries and regions, each with potentially different availabilities of relevant technology and deployment capabilities. And the relative economic motivations (e.g., based on labor costs, risk tolerance) for employing necessary processes can vary widely from country to country and even region to region. Nevertheless, companies and policy makers can insist on certain levels of supply chain transparency and can select vendors who can comply.

The food supply for a typical, large, rapidly growing city in a developing country, however, may receive the majority of its food from relatively undeveloped rural farms in remote parts of the country, where the technology for efficient labeling and scanning is either not available or too expensive. In such cases, there may be little or no tracking and tracing, and retailers and consumers may have essentially no insight into where their food comes from or what the safety risks are.

However, as mobile phones and related sensing and communications technologies reach higher levels of penetration and continue to improve in capabilities, they may help make cost-effective, easy-to-use systems available to rural farmers. Mobile phones and so-called “social web” technology also have the potential to convey important, targeted information to rural farmers, directly and through social contexts, that will help them increase output levels and productivity (e.g., Agarwal et al., 2010; Patel et al., 2010; Veeraraghavan et al., 2007).

Enhancing Multifactor Productivity

“Physical meets digital” innovation can help decisively in many respects to increase factor productivity directly on the farm and interactions with early food processors. A few representative examples are described below.

Optimizing Water Logistics and Use

The amount of water used in agriculture is staggering (Table 1). For example, it is estimated that producing 1 kilogram (kg) of wheat requires 500 to 4,000 liters of water, and 1 kg of meat requires 5,000 to 20,000 liters of water (e.g., Lundqvist et al., 2008). Thus it is not surprising that agriculture dominates the human use of water—estimates are in the range of 70 percent.

Table 1

If, as expected, the global middle class, which eats more meat and delicacies, continues to grow rapidly well into the current century, water requirements will increase much faster than population growth. Around the world today, difficulty obtaining sufficient amounts of clean water for irrigation is already a critical issue.

Increasingly, various types of sensor and communications technology are being used to provide data and perform analyses necessary to increase water productivity and reduce irrigation requirements (e.g., Aqueel-ur-Rehman et al., 2011). In dry areas (e.g., North Africa, western Asia), where water is sometimes a more limiting resource than land, farmers may focus more on water productivity than on yield per unit of land (Oweis and Hacham, 2006).

In such areas, deficit irrigation is becoming more prevalent (Geerts and Raes, 2009). This technique involves using measurements and modeling to design and execute strategies for supplying the minimum amount of exogenous water that will support stable and acceptable yields. Successful deficit irrigation requires field instrumentation that provides data reflecting soil conditions, water inputs, evapotranspiration, and spatial crop distribution. Models reflecting such factors as crop stage/type and crop water stress response can then be used to optimize output based on the water restrictions.

It is estimated that spoilage and other waste in the food chain cause about a 30 percent food loss. Therefore, smarter supply chains—both in terms of logistics and the use of environmental monitoring and response systems—can substantially increase the effective “water productivity” of agriculture.

Equipment Health and Fleet-Level Uptime and Maintenance Optimization

Uptime and utilization of heavy equipment in “highly physical” industries, whether on the farm, at a mine site, or on an oil rig, are often key factors in business performance. Fortunately, heavy equipment is increasingly being instrumented with sensors that collect data that can be used to model the health status of the equipment. Combined with other types of inspection data and environmental data, these models may be improved to the point that they ensure the optimization of maintenance of the equipment and maximization of uptime and utilization. Thus leading firms are moving away from “fix on failure” and scheduled-maintenance policies toward condition-based maintenance and ultimately predictive maintenance approaches.

However, these approaches are still in their infancy: accuracy and specificity issues often lead to false positives, as well as a lack of targeted information as to what specifically needs to be fixed/done; and prediction lead times still tend to be too short to take action to optimize maintenance and/or production schedules based on anticipated problems with equipment. Nevertheless, as the volume, diversity, and quality of data improve, and with increasingly creative modeling and analytics, condition-based maintenance and predictive maintenance capabilities are expected to have a major impact on agricultural productivity and cost efficiency in the future.

Biofuels Production for Optimizing the Harvesting and Transport of Highly Perishable Crops

The increasing use of biofuels as an alternative or complement to fossil fuels has had a significant effect on agriculture both in terms of changing demand and production characteristics and in terms of political/social engagement. When traditional food crops such as corn and sugarcane are used as feedstock, upstream biofuels production is similar to food production.2 Downstream processes, however, resemble more traditional industrial production processes. The need for detailed tracking and tracing of biofuels is generally lower than for other products, but optimizing the process from planting to production can be a priority.

Ethanol production from sugarcane is an illustrative example. Many factors affect the level of sugar in sugarcane (e.g., McLaren, 2009), which is a key factor in the ultimate fuel yield. In addition, because of certain natural biophysical processes, sugar levels tend to decrease rapidly after the sugarcane is harvested. Thus, minimizing the harvest-to-mill time interval is a high priority.

Solutions to this problem include optimizing crop planning and logistics. Using GIS technology and various modeling and analytics approaches to balance water availability, soil characteristics, distance-to-mill, road and weather conditions, traffic, and mill schedules, producers can optimize their planting, harvesting, and transport activities to maximize ethanol yields.

End-to-End Operational and Financial Management for Farms

True end-to-end business optimization is often impeded by siloed IT architectures and a lack of timely, reliable information. Typically, different data models are used for different functions. In addition, data are often of varying quality and freshness making it difficult to optimize processes that require trading-off factors associated with different functions/silos in the business.

Nevertheless, cross-functional optimizations, such as aligning harvesting specifics with current market dynamics, can be highly beneficial for agricultural firms. Thus, farms are increasingly instituting capabilities for collecting timely, consistent data about their assets and business processes and using analytics and modeling to enable end-to-end optimization and proactive decision making.

By tracking such factors as per unit costs and revenues, water usage, energy consumption, the health status of equipment, soil characteristics, and crop types/states, and then relating them to yields, metrics of market demand, and current and predicted distribution/logistics characteristics, such optimizations are possible, and managers can have access to decision-support information for balancing business objectives such as cash flow, profit, customer satisfaction, risk levels, and predicted future output capacity.

Conclusions/Action

Information technology will be increasingly important for improving agricultural business productivity and cost efficiency and for providing safe food (and fuel) for a growing and increasingly wealthy and demanding world population. “Physical meets digital” technologies and processes, which are well suited to deployment throughout the food value chain, will have important near-term and longer term impacts. The following priorities are key to making this happen.

Data capture and management policies. Necessary data are becoming increasingly available, but policies and processes to ensure that they are captured and managed electronically in a way that ensures their quality, availability, and integration (especially in federated models) have not been put in place.

Sharing and scale. Useful analytics and modeling often require large amounts of detailed data about many different kinds of assets and processes in the context of actual use. It is in the interest of everyone involved that data be shared as much as possible to accelerate the development of models and analytics that are accurate enough to support critical business decisions. Because important data are generated by many firms and other stakeholders, policies and procedures must engender mutual trust, and security must be provided as needed.

Ecosystem thinking. The kinds of improvements discussed in this paper will require that stakeholders collaborate more closely in mutually beneficial ways based on a sense of shared destiny. Relatively speaking, agricultural and food industry players have tended to be less inclined to adopt this ecosystem mindset.

Tailoring. Approaches and technologies must be adapted for the unique needs and capabilities associated with widely differing sociopolitical and economic development conditions around the world.

Business structure and models. New vendor business models will be needed to enable the widespread adoption by agricultural businesses of necessary technologies and processes. These models should include third-party vendors who can offer key capabilities at lower cost based on scale and specialization. Financing and/or cloud-oriented “as-a-service” models should also be used to reduce adoption and management costs.

References

Agarwal, S.K., K. Dhanesha, A. Jain, A. Kumar, S. Menon, N. Rajput, K. Srivastava, and S. Srivastava. 2010. Organizational, Social and Operational Implications in Delivering ICT Solutions: A Telecom Web Case-study. Paper presented at ICTD2010, London, U.K., December 13–16, 2010.

Aqeel-ur-Rehman, A.Z. Abbasi., N. Islam, and Z.A. Shaikh. 2011. A review of wireless sensors and networks’ applications in agriculture. Computer Standards and Interfaces (April). doi:10.1016/j.csi.2011.03.004.

Geerts, S., and D. Raes. 2009. Deficit irrigation as an on-farm strategy to maximize crop water productivity in dry areas. Agricultural Water Management 96(9): 1275–1284.

Lundqvist, J., C. de Fraiture, and D. Molden. 2008. Saving Water: From Field to Fork—Curbing Losses and Wastage in the Food Chain. SIWI Policy Brief. Stockholm: SIWI. Available online at http://www.siwi.org/documents/Resources/Policy_Briefs/ PB_From_Filed_to_Fork_2008.pdf.

McLaren, J. 2009. Sugarcane as a Feedstock for Biofuels: An Analytical White Paper. NCGA Whitepaper. Available online at www.ncga.com/files/pdf/SugarcaneWhitePaper092810.pdf.

Oweis, T., and A. Hachum. 2006. Water harvesting and supplemental irrigation for improved water productivity of dry farming systems in West Asia and North Africa. Agricultural Water Management 80(1-3): 57–73.

Patel, N., D. Chittamuru, A. Jain, P. Dave, and T.S. Parikh. 2010. Avaaj Otalo—A Field Study of an Interactive Voice Forum for Small Farmers in Rural India. Pp. 733–742 in Proceedings of ACM Conference on Human Factors in Computing Systems (CHI 2010). Available online at http://www.stanford.edu/~neilp/pubs/chi2010_patel.pdf.

Veeraraghavan, R., N. Yasodhar, and K. Toyama. 2007. Warana Unwired: Replacing PCs with mobile phones in a rural sugarcane cooperative. Presented at the International Conference on Information and Communication Technologies and Development, 2007. ICTD 2007. Available online at http://people.ischool.berkeley.edu/~rajesh/index_files/ WaranaUnwired.

FOOTNOTES 

1         For example, consider water as one factor in agricultural production. An increase in water productivity implies that the same level of output could be obtained with less water. Other factors could include labor, seeds, fertilizer, types of machinery, etc.

2         Non-food stocks have many advantages, but this topic is beyond the scope of this article.

 

About the Author:Matthew Denesuk is manager of Natural Resources Modeling and Social Analytics, IBM Research. Susan J. Wilkinson is the global subject expert for Food Safety and Traceability, IBM Global Business Services.