Opinion: The Missing Link in Precision Agriculture

Opinion: The Missing Link in Precision Agriculture

The agtech boom over the last two years is indeed astonishing. Farmers are becoming more open to new, emerging technologies and are more willing to adopt them.


Yield maps, VRA (Variable Rate Application), satellite imagery, and in-field sensors have been around for a while. Drone technology is a relatively new player that holds the potential to make a substantial transformation in the way in which data is being collected.

However, with so many new technologies and agtech companies that sprout up, there is still a missing link. Actually, there is more than one, but I would like to focus on one of the major missing links — decision making.

Many of the ag technologies today focus on precise application of fertilizer, pesticides, and water. Many others provide good solutions for obtaining real-time information from the field, such as growth status, water conditions, nitrogen status, stress caused by pest or disease, etc.

These can potentially help farmers make better decisions, by providing a higher quality or more accurate data. But in fact, the majority of today’s technologies provide only partial decision-making solutions and leave the decision to the farmer. Although this may sound very logical and straight forward, the end result of this situation is that the ability of such technologies to make a substantial impact on yield is limited.

For instance, growing in greenhouses is a well-established practice. The greenhouse is a controlled environment. Existing technologies allow for high precision in the application of water, fertilizers, and pesticides. Because of the small area, scouting and identifying problems in time is done easily and most growers scout the plants on a weekly basis. Moreover, growers have a wide range of flexibility in splitting the greenhouse area to smaller sections and control each section individually.

And yet, there is still so much room for mistakes. Very often, operating and calibration errors result in conditions that are far from optimal. Because of the vast number of parameters that must be taken into account when having to make a decision, growers tend to “go back to basics” and usually do not take full advantage of the technologies they already have at hand.

Furthermore, crop management decisions are complex and multi-dimensional, while most of the existing solutions only deal with one aspect of crop production.

To summarize, many good new ag technologies are becoming available today. The clear majority of them provide either sensing solutions or precise application of inputs. The missing link is what comes in between — decision-support tools — that can provide farmers and their advisers with both actionable and validated advice.

It seems that it will take some time and substantial research before such tools are developed and adopted by farmers.

Leave a Reply

JP NEL says:

Agreed, we have got wonderful tools to identify a problem , but to correct it in time is not so easy. Not enough experienced people willing to travel and examine the differences spotted. Produces not so willing to pay these specialists.

Donald N. Baker, PhD says:

Guy, you are absolutely right here! In the 70s & 80s my research team built a controlled environment facility and characterized the relationships between environment and crop physiological process rates. These were combined in predictive dynamic crop simulation models. For cotton the model was embedded in an expert (decision support) system that was used by approximately 150 farmers and extension people to manage irrigation, fertilization and crop termination. After retirement as a research scientist I consulted with large cotton farming operations in the application of this system. As you have noted decision making in managing a crop system involves too many interactions to be done without a predictive model. The mental models (used for thousands of years) can be greatly enhanced by the digital computer!

I couldn’t agree more. In fact, I think it’s important for precision ag, in general, to admit that VR alone, does not hold all the data that is required to make decisions. In a retail custom application operation, there are multiple straight-rate passes across the field that applying products that carrying higher margins and boost yields just as much or more than traditional VR applications (fungicide or foliar feed are 2 examples). I’ll warn you, I’m about to plug a product – HighQ (www.agworks.net) – sorry about that, but folks need to check it out and start pulling in all relevant application data when they start thinking about decision making!

Steve Watts says:

Guy, you are quite correct about this gap…and it is not a gap that is going to be closed soon. An even greater gap exists between the often overhyped value of new technologies (by the purveyors) and the actual results obtained (by the users). The fact is most agricultural technologies are still little more than decision support tools in most cases. We are still a long, long way from rendering human judgment unnecessary or obsolete. When technology reaches the point that it can take into account a myriad of potential variables and can repeatedly pick the next Super Bowl champion before the season begins, or maybe even the score of a single game to be played in the season, we’ll still only be closer to perfecting a decision making system for agriculture.

Guy, I have my doubts. What do you mean by “decision-support tools “? If you mean algorithms (AI) that analyze data and make decisions, then in my opinion and little experience, it might be dangerous because in agriculture we can control only a few number of variables. And, what’s more important, artificial intelligence lucks common sense, necessary to deal with varying environments. In my opinion the missing link in PA is: “Coaching” in order to train the farmer to make decisions using and interpreting information. On the other hand, if by “decision-support tools”, you mean generating recommendations by using algorithms as said before, I fully agree with you.

Guy Sel says:

Gabriel, I do refer to generating recommendations by using algorithms. I do believe, though, that AI will become much more dominant in agriculture in coming years, when data will be better validated and algorithms will have the power and the “know-how” to analyze it.

Noel Magnin says:

More brain than tech needed but tech is easier to provide.