Editor’s Note: Dr. Naira Hovakimyan is cofounder and chief scientist at IntelinAir, a digital agriculture and aerial imagery analytics startup focused on precision agriculture for the row crop MidWest. Here she describes some of the inherent challenges in applying data analysis to agriculture and offers some thoughts on the solution for the digital agriculture industry going forward.
The digital transformation of the US economy is beginning an agricultural revolution that’s likely to revolutionize arable agriculture to a similar extent that mechanization and biotechnology did last century.
There are now a growing number of startup companies working to provide farmers with digital technologies to help them operate their farms, and the majority of these are doing so using a data-driven approach. This approach involves collecting various data including data that can be collected at arbitrarily high resolution using equipment sensors, satellites, aircraft, drones, and multiple other data sources; data from sensors that can measure local weather events, moisture level and other important parameters with high accuracy; and data from financial markets or seed and chemical companies.
The startups are then using the increasing availability of affordable computation to process algorithms overnight on terabytes of this data. With that processed data and continuous monitoring, they are then trying to predict various outcomes such as annual yields or disease outbreaks, using machine learning.
The challenges of a data-driven approach
The problem with a data-driven approach, as Dr. Joseph Byrum pointed to in a previous article on AgFunderNews, is that in any farm environment, no two cases are identical. This makes the development, testing, validation, and successful rollout of digital technologies a lot more challenging compared to other industries where the data are usually static.
The data collected with high-resolution sensors changes rapidly through the season; it is non-stationary, unstructured, heterogeneous and highly sensitive to the zone, soil, weather, pests, among many other uncontrollable factors.
Also, compared to other industries such as the consumer technology sector, access to useful data in agriculture is usually restricted by privacy concerns and corporate confidentiality, and in some cases, the data has just not been collected.
The data-driven and computer science approach taken by many digital agriculture startups has meant that the results of their efforts often do not meet expectations, because they have focused on developing sophisticated, predictive models of field crop growth on the assumption that open field crop growth and health can be managed in a controlled manner, when it most certainly cannot. The consequence is slow adoption by farmers.
A systems theory approach
It is clear that digital agriculture needs a new paradigm to succeed, where collaboration in between scientists and scholars from different disciplines can yield the type of results acceptable by farmers. System theorists take into account the underlying dynamics of processes they want to control. General models of photosynthesis and crop growth are available from many years of research but have not been integrated into processed sensor and image data and predictive methods for individual farmers. System theoretic principles can help to discard some of the sensor data and focus only on important inputs, thus possibly simplifying the application of machine learning in challenging scenarios with noise and other artifacts; the methods and tools from systems theory can help to choose when to react and how to react only to particular events.
By minimizing the need for data collection, the analytics engine can be used more efficiently, saving the computation time for delivering the actionable insights to the farmer in a timely manner. Being selective with the analyzable data can help to find the sweet spot between the two competing notions in information sciences: “data-rich, information-poor” and “data-poor, information- rich.” Typically, a systems theory approach will take longer to produce “results” than a data-driven approach that produces results based on a small subset of recent data without looking into the prior history of the farm or field, or into the particularities of the crop growth dynamics.
A broader vision
We believe that a combination of the two approaches is necessary to create successful digital agriculture solutions. Interdisciplinary collaboration between academicians and industry—involving mathematicians, computer scientists, agronomists and system theorists—will be instrumental.
It is fundamentally important to account for the particularities of crop growth dynamics, analyzing the response of crops to various inputs as fertilizers, nutrients, weather changes, and so on, to try and control and manage the whole process more scientifically. The data collected by various sensors should be viewed from this perspective: namely, to use it in the best way to build more sophisticated models and be able to analyze the data in relation to the models and their evolution.
But there remain two challenges still: the cost of taking such an approach in a startup ecosystem where venture capital investors are bound by specific timelines; and the need to involve farmers.
Will data-driven artificial intelligence methods achieve a level where the data can be analyzed and learned on-the-go at a speed and cost that’s appropriate for farmers and repaid by improved crop yields? Or could a more fundamental, specific approach achieve the much-desired cost effectiveness by accounting for the crop growth models and helping to react only to particular events, minimizing the load of the analytics engine and the response time?
The longer road to a commercial product implied by a systems theory approach requires a substantial investment to develop the integrated models and methods that venture capital investors might not be willing to bear, despite the fact that down the road, this investment could reduce the costs for the data collection, storage, analytics, and optimization.
The participation of the farmer in all of this is essential too. The farmer has the empirical knowledge of a specific environment from first-hand observation which he or she can combine with data to decide when to plant, how to monitor growth, when to apply fertilizers, how to manage weeds, diseases, and pests, and when to harvest.
A farmer’s motivation to cooperate in the development and deployment of digital tools, and the ability to create value at the farm gate to achieve profitability, will be the key to revolutionizing farming in the era of digital industrialization.
Acknowledgments: The author would like to thank Professor Matthew Hudson of College of Agriculture of UIUC for his insightful comments and editorial remarks.
Image: IntelinAir’s dashboard