What is Digital Farming — Really?

What is Digital Farming — Really?

The agricultural community has spent several years synthesizing an operational description for digital farming. In several cases, there have also been solutions put forth all claiming to be digital farming, but now, more than ever, there is as much ambiguity about digital farming as there was when the community labeled and named it. Agriculture is one of the most complex systems that can be analyzed, and most of what has been proposed are solutions like weather, imagery, and NDVI, including many proprietary point solutions. These are pieces of the solution, and they are parts of the operational concept. This article will define this concept.

Advertisement

Digital farming is applying precision location methods and decision quality agronomic information to illuminate, predict, and affect the continuum of cultivation issues across the farm. Here is a look at each part of the definition:

  • Precision is about geo-location services associated with the Global Positioning System and its extensions. It is the overlay of these geo-location services on a digital map for precision sensing, identification, predictive decision making, and action.
  • Decision quality information is timely and within the decision loop of the event. It is delivered by advanced sensors, descriptive models, and predictive algorithms that provide the required insight to the agronomic issue.
  • Cultivation is a beginning to end, an end-to-end, real-time, continuous, decision process that requires timely decisions and actions. It extends upstream of planting and downstream of harvest. Today, cultivation is about the knowledgeable but subjective observational powers of the individual. Tomorrow it will be about the sensed, objective, predictive power, and precision of the digital ecosystem.

Digital farming must also be about an operational system. For this reason, the following requirements ensure the system will scale to millions of acres, deploy across multiple crops, provide an end-to-end solution, exist within an ecosystem, and support the diverse agronomic and economic needs of hundreds, even thousands, of stakeholders at the same time. In a rather crude summary, it must be more than a curiosity for the sophisticated geek and must consistently serve the small grower all the way up to the largest.

Digital farming, at the next lowest level, must organize, analyze, and orchestrate the timely delivery of information from the bodies of data that constitute a field. It must be about breaking the field down into differentiable, geo-located, and individually homogeneous units of productive assets. This requirement is about each unit as a productive asset. Precision location ensures that information collected about that unit is measured, collected, analyzed, and actioned for the same location and is differentiated from all of the other surrounding assets. It is about being individually homogeneous and identical in size, its footprint and depth, so that the system can repeatably analyze the same unit. This means that across the footprint and at depth each variable for a productive unit will have the same variable value.

Predictions and prescription can be generated and production can be individually monitored for each productive asset. Each productive unit of the asset can be separately and precisely identified, analyzed, and actioned to produce a predicted outcome. The sum of the outcomes for all of these homogeneous units constitutes a field, and the sum of all of the predictions, prescriptions, costs, and yields are the economics of the field. By treating the field as a summation of productive assets, digital farming can surgically apply advanced data and analytic algorithms, real-time, to the management of every asset and, in summary, the entire field.

Digital farming is also about what it isn’t. Digital farming is not about genetics, weather prediction, etc. These factors are extremely important to the predictions but they are externally generated and applied as inputs. Therefore, for example, the seed selection process would examine seed selection based upon seed profiles developed and produced by the ag companies. The soil conditions, hydration, and other known information are developed from the sensors, or from the historical records for the region and field. The region and field data are historical and measured, and this data is used to establish a predictive foundation. Digital farming is about using known field, crop, nutrient, protection, and hydration information to predict outcomes based upon sensed, processed, and aggregated information.

Digital farming is about describing the asset in the digital domain; about creating a digital twin that can be evaluated repeatedly against many variables. Each unit has definable and consistent measures of the key agronomic variables across the productive asset, and it is about separating the data that is changing from that which is not changing. This means there are variables about the productive asset that can be ordered and analyzed to predict the performance of the productive asset. For example, and just looking at a description of the field, the following layers of information are presented in the order of their variability. They are:

  1. Topography: contours and slopes, inclination to the sun, run off of precipitation, etc.
  2. Buried or hidden artifacts: drain tiles, compaction zones, rocks or buried geological formations.
  3. Electrical Conductivity Mapping: an indication of soil types and hydration carrying capacity.
  4. Soil Sampling: validation of soil types, horizons, and composition, an infill map of the 3-D characteristics of the asset.
  5. Historical Stress locations: geo-locations of previous weed escape, insect infestations, fungal outbreaks, nematodes, etc.
  6. Historical supplements and treatments: previous crop or field protection applications, nutrient applications, etc.

These are all permanent, or semi-permanent, additions to the description of the productive assets that change slowly over time. The more permanent descriptive information is presented first, and, as each new variable is examined, each iteratively becomes more variable over time. The previous example is about the soil; the productive asset, but each of the domains of the digital farm deals with information in massive scales and moves from the invariant to the variant.

The most important attribute in this discussion is that the massive amounts of data represent the properties of each domain that influence the cultivation and are described down to the exact geo-location of the asset. All of these data can then be analyzed separately and summarized for a field, farm, region, etc. This is the data that needs to be geo-located, precisely measured, and described digitally. This data is the foundation for all of the subsequent efforts. It sets the foundation for the geo-location of real-time inputs and the application of precision technology.

Finally, in order to apply this data, the grower needs to collect, communicate, store/archive, retrieve, orchestrate, and analyze this data. The grower needs to get inside and ahead of his decision loops; the decision loops that make up the cultivation cycle. This cannot happen without timely information. The required timing and location of the sensing needs to be predicted and the analysis of the identification needs to be real-time to effect a positive change. The delivery of information a day, week, or month later is not digital farming. This also means that using any data that is days old or from sensors that detect too late in the stress cycle to influence the stress are not digital farming. In this case, the currency of the detection of the event upon which action is required is not only timely but critical. If the decision information is presented too late, the issue is already invested in the assets and any remedial action is often too late. Digital farming is about providing timely information to the grower when they need it, in real-time.

In summary then, digital farming is about precision location, real-time sensing and processes, and the generation of decision quality agronomic information across the continuum of the cultivation cycle. It is also about scalability, end-to-end processes, and generalized operational delivery to the agricultural community.

Leave a Reply

Peter Hathaway says:

Look at niolabs precision AG.

michael collins says:

Thank you Peter. I have been for a while. They are a piece of the solution. I like the additive nature of their approach and the microservices application. Michael