7 Factors That Could Make or Break Digital Farming

Digital farming is the new technology genie – a genie that is now out of the bottle and cannot be put back. It is the overarching concept that embraces the precision afforded by global positioning, integrates the resolution and fidelity of new sensors and controls, and unleashes the economics, computer horsepower, and storage capabilities of the digital revolution — the now-emergent fourth Industrial Revolution, an Agricultural Revolution. Digital farming is the digital modeling of the entire cultivation – i.e., “a digital twin,” a digital replica of physical assets, processes, and systems. And its integration with a farm management system in a digital ecosystem enables all stakeholders to investigate alternatives and generate decision-quality information when it is most critical, before the need arises.


This digital genie is out of the bottle, but before the genie can really work its magic, there are some core elements of the digital and agricultural ecosystems that must be developed and integrated to a whole system. Digital farming must automate data ingestion and analysis, and it must scale to provide the value creation that is promised. To do this, there are still some gaps that must be filled. Here are seven core concepts of digital farming and why they are important in the agricultural and digital world – the agricultural digital ecosystem.

1. An Integrating digital platform. Needed first and foremost in this digital ecosystem is an integrating digital platform. We are beginning to see what this means with the emergence and dominance of platforms like Amazon for online purchasing. Before there can be anything similar to online purchasing in agriculture, there must be a digital platform that enables and protects stakeholder access and information; automates the development and analysis of massive bodies of data; and develops, reveals, and manages the potential costs – and revenues – of these decisions.

2. A sequence of actions. Second, as with every process, there needs to be an order of precedence, a sequence of actions. Simultaneous with the advent of new digital ecosystems has been an explosion of new technologies and insights across the cultivation cycle. These technologies and learnings need to be organized, structured, modeled, and scheduled in order for their decision insights to be practically applied in the field. Many of these technologies currently exist outside of the digital ecosystem, and they need to be digitally modeled and integrated to improve decision processes and create stakeholder value.


To help make this happen, each technology can be viewed relative to its contribution to the decision process. For example, imaging must be digitally curated, analyzed, and integrated with other already-known information. In the digital environment, image analysis must be automated because human analysis is inconsistent, is not reproducible, and is always “after the event.” Once imaging is combined with already known information, like a previous presence of weeds, it can provide an indication of when, where, and why plant stress might again emerge. In this case the application of imaging might follow a sequential process that includes steps such as Indication, Detection, Identification, Localization, Decision, Mitigation, and Validation. Because we know what to look for, we can select the right sensor and the right time. And once the types of stresses and their locations are known, mitigation can be accomplished with precision and economy. Known stresses will emerge again, and their emergence will be anticipated.

3. Automation of analytics. Third, a key use of the digital environment is the automation of analytics. Automation manages the analytic processes applied to every file and every data element of that file. It analyzes this data to support some anticipated event and its required decision. This data exists in very large files, and these files must be transmitted across available communications systems. These files must be in forms that are both accessible and actionable, able to be stored and analyzed in a timely, as-needed, and cost-effective way.

This is an important piece of the puzzle. Stakeholders should not be forced to learn how to manipulate this data; instead they should have the ability to progress from data scheduling, presentation, and description to developing predictions, diagnostics, and prescriptions for action. The digital platform applies the analytics and adds knowledge to the process; it must traverse the data from what happened, to why it happened, and how it can be mitigated.

4. Soil as the productive asset. Fourth, soil is the productive asset. There needs to be a digital soil model in this ecosystem – i.e., again a digital twin, which models the soil and its productive capability. Soil is the common denominator, from one cultivation to the next. This digital twin describes and analyzes all aspects of the soil that will have an impact on cultivation. The digital modeling of the soil starts with factors such as topography, soil horizons, and soil content. This modeling includes identifying buried appliances, soil treatments, and crop protections. These all need to be geo-located; ranked sequentially from least variable to most variable; and fully described, modeled, and made digitally accessible to the analysis efforts in this ecosystem.

5. Digital description of the seed or crop. Fifth, the seed or the crop to be grown needs to be digitally described and a body of growth metrics developed that characterize the cultivation of that crop and variety. In this case, this step by itself should be a sufficient and persuasive justification for the development of digital farming. When seed production – or similar historical – information is maintained for any crop in a digital ecosystem, a wealth of data and metrics is available for the next cultivation. This would include crop and variety information like first emergence, timing of vegetative states, root growth, nutrient and moisture utilization, and fruit size and maturation. These metrics can be statistically described, and these results can be used to digitally predict and schedule when certain events in the next cultivation should be happening.

6. Weather prediction. Sixth, weather prediction already is significantly better than it was even a decade ago, and today – even as weather is happening – remote monitoring and measurement have improved significantly. This means that it is now possible to correlate weather in terms of sunlight, rain, temperature, and other variables in the cultivation. Further, it is now possible to provide real-world models that integrate the soil, seed, and weather models to provide continuous monitoring, prediction, and scheduling. This coupling of models can provide additional benefits to determine crop utilization of moisture throughout the cultivation and nutrient leaching and depletion, and provide a view of other processes that are dependent upon water as the medium for transport.

7. Stresses are persistent. Last, stresses that impact the crop are persistent. In general, if there was a weed present in the last cultivation, there probably will be a weed in the next cultivation. The location of this weed stress is known and can be geo-located within inches. This geo-location and persistence applies not only to weeds but to fungi, insects, and biologicals. Early indications of these stresses are locked in the memory of the soil, and the digital twin will remember exactly where. These stresses can be identified and – like crop metrics – these stress metrics can be monitored for when and where to look to detect the next emergence. Once these stresses are indicated and detected, further resources can be applied to identification, localization, mitigation, and validation.

These seven overarching concepts are just some of the core elements, and these need to be integrated into a whole that provides all stakeholders with the required decision-quality information when it is needed. These overarching concepts need to be pushed upstream and downstream of on-farm production to provide for cultivation and information provenance as well as agricultural stewardship.

Finally, digital farming must be economical and scalable, and the platform must create and deliver value for all stakeholders. Economics, scalability, and value creation ensure that the platform and its presented capability will be used and profitably integrated. The genie is out of the bottle, and these concepts are a part of digital farming and are leading an agricultural revolution.

Leave a Reply

Avatar for Bob Wheeler Bob Wheeler says:

Excellent article Michael. I think you hit a lot of key points or goals of where we’d love to see precision farming go. We have hundreds of different data inputs today and with IoT and other technologies we are only going to increase the number of data inputs. One of the things I believe we struggle with is putting structure around all the data inputs so that they can even be correlated. Perhaps defining a structure is also part of the problem versus leaving data in a unstructured format that Big Data platforms help bring together for information extraction. We’ve spent years trying to get data formats standardized so we could work with the data without 10 pieces of software taking binary data putting into to some flat file that is more useful for the layman. Whereas today we need more data hubs that consume data in any format that allow us to analyze and extract information in order to make good agronomic decisions.

Avatar for Michael Collins Michael Collins says:

Bob, thank you for the kinds words. This is the first article I have written for general publication. I have several articles I have posted on LinkedIn and many I have written as reports or documents for clients and customers. This article was an attempt to begin to put structure to my continuing interest and study in the Digital Farming domain. I believe that I am a strategic and enterprise architect. I find digital farming, along with a number of other domains like it, to be a real challenge. That said, all systems have an architecture; intended or unintended, an architecture exists. One of the key properties of a system is that its form is the organizing construct for all system information. Part of what I presented in this article is that the form of this system will enable us to store, access, and process/ analyze this data. My job is to reveal the form of this system so that we can understand, assess, and manage it.

Avatar for Liz Wiley Liz Wiley says:

Hi Michael – this was a great article and one that we can speak directly to. Our company, Spherical Analytics is doing exactly what you are recommending. We ingest and aggregate large disparate sets of data, proof its for trust and authenticity, blockchain the data to secure it, then conduct predictive and/or descriptive analytics and visualizations for key stakeholder. Check us out at https://www.sphericalanalytics.io
We can aim this platform at any industry segment – as the Food Security Lead at Spherical Analytics, I focus predominately on agriculture, water, and soil. I would love to talk with you or anyone else in need of these services that we provide.

Maybe agrirouter can be the digital platform for individual data-exchange mentioned in these excellent article
see more on http://www.my-agrirouter.com

Avatar for Ken Wagenbach Ken Wagenbach says:

Very Good article and feed back. Having spend personal resources/time consuming i.e purchasing as a grower and then as one validating and verifying value within the realm of an OEM manufacture consuming/developing /supplying Precision Ag technologies, perhaps, we miss the mark in the fact that the technologies of digital farming/big data must create real value to the producer, (dollars in their pockets), before it will have economically sustainable value to the rest of stakeholders. These are exciting times with positive intrinsic value for agriculture and environment as whole and just maybe the reply from Bob, may very well have hit on what is restricting the timely release of the “nutrient “value of the manure of ” Big Data” and Digital Farming.

Avatar for Roger A DuMond Roger A DuMond says:

Mr. Collins,

Thank you for a very insightful article, and a topic that i seem to dwell on a lot. It seems that many of us silver haired participants of precision ag / digital ag share a lot of the same thoughts. As agronomists, we are convinced that the end results we strive for in digital ag are valid, but just like the floating beach ball, many of the end rewards seem to keep floating just outside our reach. Each day the analysis and data management tools get better, so maybe the goals are coming within reach a bit more.

Anyhow, I would add two more factors to your list: 1. Human Capital (with the training and tools to know how to properly manage the massive data sets) and 2. A well defined “payback” monitor that will demonstrate to the end user how their investments are being rewarded.

Thanks again for a great article.

Avatar for John walter John walter says:

Michael: good article, congratulations!

John T. Walter

All your observations are a lot of Marginal Costs (MC) that will be larger than the resulting Marginal Revenues (MR). So which MCs are the lowest hanging fruit for MRs? I push MC until it equals MR and then I stop. Today the MR is pretty small and traditionally agriculture spends a lot of time with low MRs.

I have always loved technology until the BIG data agendas started taking aggregation liberties with my small data. At least Facebook users don’t “pay” to be aggregated.

Isn’t it about time the industry values and respects the small data generator with some kind of “peer to peer” personal data silo system that is protecting by some “bitcoin only better” type of digital encryption to take advantage of cloud technology to one’s personal large data silo and then we can find the true value of digit data because it would have to be purchased.

I have had great success with capable GIS finding correlations of my own racked and stacked data layers.

The MC=MR will determine which things will survive the make or break analysis. It will take time but it will accelerate with the next jump in MR…..which is not on the horizon just yet.


If your blockchain technology can give me my own controllable key to my data silo in the cloud then the ground fog just might lift and new possibilities can evolve for all data creators to the benefit of civilization for the equitable and right reasons.


Avatar for fslaviero fslaviero says:

Thank you Michael, especially for the focus on the ‘sequence of actions’, i.e. the inevitable focus on business processes. This is frequently forgotten as new technology is taking most of the consideration. We’re running digital integration platforms (www.siti4farmer.co.uk) for agriculture, with automation, insights and predictive analysis with soil at the core. When working in real business cases with farmers and food industries, the internal, upstream and downstream processes take the lead.

Avatar for Orlando Saez Orlando Saez says:

Great writeup, especially calling out #7 – stress detection. Disease detection cannot happen without the verified presence of pest or pathogens. Many believe that detecting crop pests can be modeled from environment, stress maps and host susceptibility, but this would invalidate the disease triangle. At http://www.aker.ag, we offer a FREE (no strings attached) crop scouting app (AkerScout) and focus ground truthing the old fashion way, by walking fields, and leveraging technology to automate the process. I hear that this is your first article. I suggest you continue your writing work… We need more voices supporting innovation in agriculture! Nice work.

This is excellent Michael! We’ve using the term “digital twin” to convey the same message and I certainly agree on the seven factors as well.

Avatar for Michael Collins Michael Collins says:

Charlie, First it has been awhile and thank you. I have been an enterprise architect the last twenty years but for the entire time before that I built digital twins of real world situations. I called them simulations. Most of them were for Dept of Defense but some were for things like bus systems, manufacturing capabilities, and others. The simulation of the real world has gotten much more sophisticated and complete. It is also now possible to do virtual live with the integration of IoT. The enabling of DF with the fabric of a Digital Twin is this next step. In these articles I am trying to stimulate the dialogue and address the remaining issues to ensure value creation and risk mitigation. This article and my subsequent ones are generating a lot of feedback and long discussions. Again, thank you. Michael