IBM Develops Platform to Fulfill on the Promise of Digital Agriculture
If precision agriculture solutions are going to help farmers, a chasm must be crossed.
“By nature, agriculture is a very fragmented and siloed business, which has prevented precision agriculture from advancing as fast as it could have over the past two decades,” says Mark Gildersleeve, vice president and head of business solutions, Watson Media and Weather, IBM. “We haven’t made it simple enough for a grower because we’re not connecting the dots to give him better insights.”
Gildersleeve says there are three important reasons why we need to tackle this issue.
- Razor-thin margins. “Growers are at razor-thin margins right now,” he says. “They need a little bit more of a financial cushion underneath them. We believe we can help give them some financial cushion, which in turn helps improve their profitability.”
- More mouths to feed. “There are going to be a lot more mouths to feed over the next 30 years globally,” he says. “We really have to improve yield if we’re going to feed 2.2 billion more people on the planet.”
- Consumer demand for quality food. “Consumer demand is really pushing food companies to become much more focused on producing quality food sustainably,” says Gildersleeve. “We have to work back up the supply chain to help growers deliver what consumers are demanding.”
With the introduction of the Watson Decision Platform for Agriculture, IBM is bringing together data and artificial intelligence to help growers make better decisions. This new platform is an innovation that draws on IBM’s most advanced capabilities in artificial intelligence (AI), analytics, IoT, Cloud, and weather to create a suite of solutions that span the farm-to-fork ecosystem.
How it works
The platform begins by creating an electronic field record (EFR) as the single source of truth for each farm. Similar to an electronic medical record, the EFR is populated with information like:
- Weather data from The Weather Company, including historical data, near-real-time observations, and forecasts 15 days in advance as well as seasonal and subseasonal trends.
- Soil data like moisture at multiple depths, nutrient content, fertility, and type.
- Farm practice and workflow data gathered from cooperative growers (e.g., planting and harvesting dates, fertilizer and pesticide application rates, and harvest outputs).
- High-definition visual imagery from multiple satellites, drones, and airplanes.
Once the data for the EFR is gathered, Watson applies AI, machine learning, and advanced analytics to extract valuable insights and automatically generates recommendations to help farmers make smarter decisions. A unified dashboard enables growers to easily visualize data and alerts related to critical elements such as weather forecasts, soil conditions, evapotranspiration rates, and crop stress.
For example, AI visual recognition of drone footage may be used to automatically identify certain types and the severity of pest and disease damage. With this field-specific insight, growers can save time and money while reducing the impact on their fields by better understanding how and when to spray.
The solution can deliver a variety of benefits across a farmer’s fields including:
- Improved crop protection by leveraging AI to better understand and proactively alert growers to critical daily crop stress levels, identify signs of pests and diseases, and more effectively assess current risk levels of crops.
- Increased yield optimization with benchmarking and validation against yield models for comparable soil and weather conditions as well as support for better decisions around irrigation, product application, and planting and harvest timing.
- Smarter in-season trading with productivity assessments and decision guidance as well as varying weather conditions that feature detailed analysis of subseasonal and seasonal forecasts.
Putting the Technology into Practice
“Until now, no one has tackled putting all of this information into one place,” says Roric Paulman, a third-generation Nebraska farmer. “I have 40 different ag apps on my phone. It just stops being useful.”
Paulman generates 1 terabyte of data every month across the 10,000 acres he covers. IBM’s new platform allows him to bring all of that information together on his phone, so he has a unified view of his farm.
For Paulman, applying artificial intelligence to his data provides startling new powers. He can now capture a field of corn from a drone and use Watson-enabled visual recognition analysis to identify crop disease or a pest infestation. The app also allows him to photograph plants up close, so the technology can identify exactly what is stressing the plant.
On his farm, an agronomist currently visits once a week to analyze infestations and blight. Now, with a simple photo, Paulman can instantly find out what type of pest is affecting his plants and he can take action.
“That means I can react in real time and won’t lose yield waiting for the agronomist,” he says. It also allows him to better target pesticide use, which reduces the environmental impact and lowers cost.
With this platform farmers can anticipate problems before they start. “The platform helps farmers understand critical factors such as soil temperature and moisture levels, crop stress, pest and disease risk and identification, yield predictions, and alerts,” says Gildersleeve. “That information helps the farmer make informed decisions on irrigation, planting, fertilization, worker safety, trading, and pest and disease eradication.”
One of the biggest challenges farmers like Paulman face is knowing when to sell their crops. “Prices fluctuate constantly, and this platform offers a tool that marshals huge amounts of pricing data – from the local grain elevator to the futures markets – and recommends the best time to sell a crop in order to maximize profit. It’s the type of data gathering and analysis that would be impossible without AI and analytics.”
“I’ve been waiting for something like this,” Paulman says. “IBM has independence. They’re not trying to sell me more fertilizer or machines. They don’t have a horse in the race. It’s a trust thing.”
Grower Is One Piece of the Puzzle
The grower application is just one piece of IBM’s larger effort to improve agriculture.
“The platform can help a full range of ecosystem participants such as agronomists, input providers, equipment manufacturers, traders, lenders, crop insurers, and governments make more confident decisions specific to their own roles,” Gildersleeve says.
For example, he believes agronomists are spending 80% of their time trying to get their arms around farmers’ data.
“They’re only able to spend 20% of the time actually making a recommendation,” Gildersleeve explains. “We need to reverse that trade-off so the agronomist can take action earlier. What we’re trying to do is automate the data collection process and get the information presented to the growers and the agronomists, so they can make decisions faster. The best artificial intelligence isn’t supplanting the agronomist; it is just doing the preliminary work, so he can make more decisions faster.”
Varying Degrees of Risk Aversion
In farming, there is a wide range of farmers who are in varying degrees of risk aversion – in other words, their level of willingness to take a risk and try out new technologies or practices. So how do you convince farmers to take a chance on yet another ag tech tool?
“I think that’s a completely fair question,” Gildersleeve says. “I would approach it this way. A lot of the information we’re delivering through both the mobile application and web application is real-time information. As an example, we’re delivering weather that’s precise to the nearest, essentially, to each square kilometer across the globe. We are updating that weather every 15 minutes to every hour. We are providing real-time data that allows growers to take action with their mobile device, so they can decide whether to spray or not. They can really understand the most updated weather information that they can get anywhere.”
Second, he says, is with their crop stress imagery, you can identify areas in the field where you need to scout faster and take action earlier.
“The third area where I think we will deliver benefit to growers is we’re going to deliver answers or recommendations based on all the data we’ve collected from all of our grower clients and not just from their particular field,” he says. “They can learn what the data is telling them about how to improve yield or how to lower costs across the nation. That will be, I think, a better learning experience than things that are just focused on their individual field performance. I think we have to learn how to help growers learn from each other with data and be focused on profit, not just yield.”
IBM is a firm believer in data cooperatives as a way for growers to contribute data. Your personal data would not be shared without your permission, but the cooperative would be able to run analytics on a combination of growers anonymously, so we can understand, for example, for all the corn growers in Nebraska, what is common among them that drives 20% over the average as opposed to the grower that had 20% below average.
“As a community, it would enable us to help growers try to understand what practices are truly driving better yields from a different data perspective and which practices are not contributing,” Gildersleeve says. “Growers would be able to learn from one another. They don’t have to just learn from their own bios on their own farms year after year, which is how many of them are learning today.”
What is this going to cost farmers to use?
There is a wide range of options for growers. The base level package is running between $500 and $750 a year depending on volume of the partner. There are add-on options that are more expensive than that when we get into higher level analytics that are driven off of drone imagery, for example.
Offering a relatively low entry price, he says, will allow more growers – no matter what size operation they have – to reap the benefits of this platform.
“We think we can help growers that are at average yield or even slightly below average yield to improve their level of profitability,” he says. “This is not just for the early adopter. This is not just for the people that are already way outperforming their neighbors. This is for typical growers who need our help to improve their profits. That’s what we’re all about right now – how do we help these growers eke out more profit. This is not just about driving better yield, this is about driving better profit.”
A Glimpse into Data’s Future
In the next five to 10 years, shame on us if we haven’t figured out how to automate 100% of the data collection, Gildersleeve says.
“Shame on us if we haven’t put artificial intelligence to task to be cleaning up every piece of data that we’re collecting from the field. Whether it’s from a machine, whether it’s from the grower or whether it’s from the environment of that field, we need to have a much cleaner data set, which will make the analytics for growers better,” he says.
With thousands of growers’ experiences contributing to its data set, IBM will have a much richer data set to be able to give growers evidence-based recommendations.
“We’ll be able to enable the agronomist to make decisions with greater confidence, because right now having enough confidence is one of the biggest limitations,” he says. “I still expect the agronomist to be a part of every one of the key decisions preseason and in-season, but we just need to make it way simpler for them.”
Lastly, Gildersleeve believes we must do a better job of connecting the silos across agriculture. “From the input providers to the insurers to the lenders to the food companies, we need to collectively figure out how to help make growers more profitable for the entire system to work. Ultimately, our mission is to connect these silos, so we can make growers more profitable.”