From CEA to PIA: Drawing Lines in the Soil

Introduction

The field of Controlled Environment Agriculture (CEA) is both expansive and transformative. It includes a range of applications from protected field production to ambitious initiatives to grow food in outer space. Given this extensive scope, it's critical to introduce a more nuanced classification within the field. This article proposes the term Precision Indoor Agriculture (PIA) as a subfield of CEA that stands apart through its transformational adoption of technologies that fundamentally change operational paradigms. We will outline the key points of delineation that define PIA and present a conceptual framework to understand its untapped potential.

What is Controlled Environment Agriculture?

Definition and Range

Controlled Environment Agriculture covers a broad array of methods where the objective is to regulate the environmental factors affecting plant growth. This encompasses everything from basic polytunnels to sophisticated aeroponic systems on the International Space Station.

Key Objectives

• Environmental Regulation: A degree of control over light, temperature, humidity, and other climatic factors.

• Resource Efficiency: Improved utilization of water, nutrients, and space.

• Risk Mitigation: Reduced susceptibility to pests and diseases.

• Year-Round Production: Less dependence on seasonal favorability.

Introducing Precision Indoor Agriculture

What Sets It Apart?

Homogenous Environments

While most CEA applications aim to mediate environmental conditions, PIA applications are those which precisely supply all of the factors of production. PIA projects pay close attention to the details and, for example, not merely satisfied with keeping the room at a given temperature, but cognizant to attend to the microclimates within canopies. They understand that it is only by achieving uniformity across the entire range of inputs, from plantlets to fertigation, that a crop can be optimized as a single unit. Otherwise, variances inevitably cause the grower to produce below the ”efficiency frontier”, a term used to describe the theoretical limit to an agricultural system’s productivity.

Growers universally leverage a range of tools and platforms that facilitate data-driven agricultural decisions. Yet, it's a truism among operators that every crop cycle, even in the PIA context, presents a unique set of variables. Thus, crop management a highly intricate task requiring multi-factorial optimization. As you moves toward the PIA end of the CEA spectrum and the factors under your control increase, management becomes even more complex given the interplay of mechanical, chemical, and biological factors that must be harmoniously managed. However, these types of complex, multi-dimensional problems are precisely where machine learning algorithms demonstrate their strength.

Dynamic Biofeedback Controls

Operators of PIA farms are moving away from traditional heuristic-based management approaches, which, although largely effective, lack the precision and adaptability offered by a modern approach. Instead, there is a growing inclination towards systems capable of active feedback mechanisms, modulating inputs based on information taken directly from the crop. Such systems leverage near-term metrics, like derivative models of photosynthesis and respiration rates, as opposed to long-term indicators like overall harvest yield.

By operating in a rapid, iterative feedback loop, these systems make adjustments in real-time. This capability enables hour-by-hour and day-to-day optimization rather than waiting for an entire crop cycle to evaluate performance. The result is a compressed timeframe for deriving actionable insights and their subsequent implementation, minimizing the need for human oversight. This approach is emblematic of the paradigm shift towards a more agile, responsive, and automated form of agricultural management.

The data landscape for a Precision Indoor Agriculture (PIA) operator is fundamentally distinct in its qualitative aspects. Traditional production systems, which tolerate variability and confounding factors, generate data with a poor signal-to-noise ratio, thereby diminishing its analytical efficacy. To illustrate, if the gaseous composition of a cultivation environment is subject to intermittent or inconsistent ventilation, the cheaply obtainable markers for aggregate photosynthesis (oxygen levels) and aggregate respiration (carbon dioxide levels) become unreliable.

However, these limitations can be addressed if data streams are meticulously considered during the initial design phase. Through calibrated dosing and controlled ventilation systems, operators can maintain the integrity of valuable indicators related to crop health and productivity. The cornerstone for achieving a high-quality dataset, suitable for advanced machine learning algorithms, rests on three pillars: strategic sensor deployment, an environment validated for homogeneity, and centralized data integration involving all sensors and control systems.

This approach holds the promise of fully automating cultivation management. To draw a parallel with the development stages of autonomous vehicles, current advanced PIA deployments appear to be at a maturity level analogous to Level 2 or Level 3. In this current stage, most cultivation management functions are controlled by setpoints and schedules, and can make automatic adjustments, even tied to biofeedback. However, no commercial system currently exists that is entirely self-sufficient in its operations, or can conduct optimization autonomously.

Enhanced Biosecurity

Another point of differentiation when describing PIA can be made in the approaches to managing pest and disease. PIA operators seek to exclude all sources of variability, including pest and disease. Integrated Pest Management is a holistic approach to managing pest and disease on a farm, a paradigm that often recognizes that many CEA applications have endemic pest and disease. Thus, management strategies should justify and limit impactful interventions such as chemical pesticide applications. For PIA operators, greater emphasis is placed on prevention and eradication than management. The business case for producing in a pest-free environment includes not only the reduce labour and consumables cost of a conventional IPM program, but also avoidance of the drag on productivity and loss of product quality.

Biosecurity is easier to maintain, than to remediate, leading to the greater adoption of phytosanitary inputs. Operating pest and pesticide-free requires both emphasis on biosecurity through facilities and operational design, as well as management strategies for the elimination of anything that bypasses those aforementioned layers of protection. Success in this domain can be both practical and economical in the long run, and opens up a world of opportunity in automation and productivity.

IPM in the context of PIA recognizes that personnel are the most difficult vector to control for, and the elimination of human crop labour remains a promising frontier. While much attention has been given to materials handling automation, which is welcome news for those who’ve operated at heights, a pair of more recent technologies have radically transformed the timeline for the development of genomes that are adapted for PIA applications and facilitate the elimination of crop labour, and enhance the business case in other ways.

Distinct Genomes

The agronomic traits which have been commercially selected for, in the tedious and resource intensive conventional breeding work practiced to date, are not commensurate with the criteria for PIA. It would take decades to conventionally incorporate those traits that are best adapted to the rapid-cycling, high density, PIA approach. However, CRISP-enabled targeted gene editing, and Next Generation Sequencing technology are allowing for the economical validation of genomic interventions that target trait transformation using known pathways. PIA adapted cultivars will be distinct from those for other production contexts, often short stature, rapidly maturing, and high-yielding, with less emphasis on weather and pest resistance. Recent regulatory clarity in Canada and the US has paved the way for commercial adoption of such breeding approaches under existing Plant Breeders' Rights frameworks opening the products of this technology for commercialiation.

Why is PIA worth defining now?

There is a constellation of technological, economic and social trends that are converging to make advanced indoor agricultural operations economic reality. While the scaling of horticultural LED manufacturing, and the proliferation of digital sensing and controls have lowered CapEx and time to RIO, energy and labour costs have remained prime impediments to financing viability. However, a great deal of operational automation is merely a technical challenge, and the current reshaping of electrical generation infrastructure in Canada and the US is poised to hand a direct subsidy to PIA.

The major automotive electrification push is tied to efforts in modernizing grid infrastructure. By placing energy storage in everyones garage, grid operators are able to incent off-peak consumption and deploy a greater proportion of output as low-carbon base load, rather than high-carbon peaker capacity. Overnight rates are becoming extremely low and subsidized by on-peak charges, giving rise to competitive advantaged for those asynchronous with the sun. In fact, operating overnight gives better heat shedding performance for major sources of consumptions like chiller plants giving greater motivation to adopt PIA and more energy intensive crops.

Finally, consumer consciousness surrounding food quality is increasing demand for locally-sourced, pesticide-free, highly nutritious foods. This awareness is creating economic niches that may erode established food systems structures as PIA operators shorten supply chains and deliver fresher, better food. For those who want out-of-season produce (don’t we all), and are concerned that our blueberries are being flown from Peru, many consumers would welcome sustainably, locally-produced foods that are picked ripe and sold fresh.

Conclusion

The synergies between the technologies and management practices described above, prime the stage for a sea-change in industry practices. PIA represents a radical transformation, necessitating different expertise that conventional CEA. Both high skilled and low skilled agricutural work will be disrupted, and so will supply chains and downstream business models. To be sure, there are still uncertainties and a great deal of work by some very talented teams before PIA meets mass deployment, but the time to redundancy is currently very short, as evidenced by Plenty shuttering of facilities opened as recent as 2019. While some observers suggest this, and other high profile closures represents the non-economic viability of PIA, an alternative view would be that we just experienced a steep phase of technological innovation, resulting in the justification to rebuild, rather than operate inefficient designs. Innovation is occuring at a rapid pace, and there has never been a better time to think and grow differently.