It seems like the agriculture sector produces more data than commodities these days. Realizing the full potential of that data, however, requires more robust programming.
In other words, growers need better tools to manage more aspects of crop production, while simultaneously knowing which ones require more focus.
Why it matters: More refined computing could remedy the lack of return on investment (ROI) from digital tools.
Raj Khosla, professor of precision agriculture at Colorado State University, made that case during the 2020 Virtual Precision Agriculture Conference.
In an extrapolation of the 4R concept of nutrient management, Khosla says farmers need more precise programs (algorithms) to achieve the “5Rs” of digital agriculture management — the right input, at the right time and place, in the right amount and manner.
“Now we’re at a point where we’re collecting numerous data layers,” he says. “There is a spatial as well as temporal component. This is precisely what I mean by farming the data… Using algorithms which will let farmers make better decisions.”
Khosla says big-data tools have already been applied to run a completely autonomous farm in the United Kingdom, the Hands Free Hectare initiative at Harper Adams University. This shows farming exclusively via data is possible, although improvements are needed.
Wider dissemination is now largely a matter of making computing processes more effective and efficient to bring costs down. Considerations such as the relationship between moisture and soil texture, which factor should be given more weight, and how to account for seasonal anomalies in the data, all need to be refined.
“I think this is going to be part of the many tools and technologies for us to use in field to continue to reduce our footprint, increase production, be profitable, and do all that in a sustainable manner,” says Khosla.
But the subjective knowledge and connection farmers have to their geography is still desirable, as well as valuable in production. That means ensuring farmers have the ability to override AI-generated prescriptions is critical.
Overcoming labour issues is also an ongoing challenge.
“We really lack those skilled individuals that understand agriculture and the computational (side),” he says. “Universities need to update our curriculum. That prepares our next generation of professionals.”
Khosla says producers should still collect data despite not being able to use its full potential. Data that currently has no use could be useful later, whether via the Internet of Things or individual digital tools.
The potential availability of alternatives such as edge computing — in which digital tools are stored in computers on the field edge — is one example.
When asked what he believes are the most valuable data layers available, Khosla says anything related to time and topography (since landscapes tend not to change much) are often cited as bringing the best returns. Yield data, too, can be very useful in drawing conclusions over time.
Khosla says the number of acres is not fundamentally relevant, although some digital and precision tools are more easily adopted by larger farms. Even African farmers working less than half an acre could use precision tools to improve their businesses, he says, provided the right limiting factors are identified.
“We need to challenge ourselves to identify those limiting factors. Will there be growing pains? Absolutely. But yes, it can be done.”