TDWI | Training & Research | Business Intelligence, Analytics, Big Data, Data Warehousing

Q&A: How Location Intelligence Helps Make Farming More Environmentally Sustainable

Location data isn't just about plotting the best delivery route or where your customers live. In the use case described in this interview, we learn how Argentinian farmers are profiting from it.

• By James E. Powell
• September 23, 2019

Location data is often associated with plotting the most fuel-efficient route for delivery vans or analyzing where your customers reside. As this case study shows, location data can also help farmers increase their yields. We spoke with Horacio Balussi, CIO of the Argentine Cooperatives Association (ACA), about his project to help farmers using satellite data.

For Further Reading: Emerging Practices in Location Data Management and Analytics  Location Intelligence and the Conquest of "Inside Space" CEO Perspective: Future Trends in BI and Analytics

Upside: What kinds of information was the ACA monitoring before you started your project and how was this accomplished?

Horacio Balussi: As one of the largest grain operators in the country, the Asociacion de Cooperativas Argentinas is a primary contributor to Argentina's export market. Before starting our project, we were monitoring each stage of farming in the fields, from seeding to harvesting, but we could not analyze the satellite imagery manually on such a large scale. In the past, these farmers were not using tools from the ACA. Instead, they used spreadsheets and other types of software.

Why was this method no longer optimal?

Because we could not analyze the satellite imagery manually, we lacked insight into the health and crops of the fields, which resulted in increased costs and caused negative effects on environments in Argentina. For example, there was wasteful use of water and pesticides due to lack of visibility on crop health.

What alternative methods did you examine before choosing your project's direction?

Last year, we were looking for a platform or product to obtain different indexes or images from satellites and drones. There are many options in the market, but SAP S/4HANA Special Services processes the enormous amount of information coming from satellite images and gives us the index we need. We selected SAP because they provide us with processed information from the satellites. Without SAP's processing capabilities, we'd be unable to take actionable next steps from the satellite information.

Tell us about the solution.

Our project for digital farming was centered around giving our farmers a tool to improve crop production. This project is centered on intelligent information and intelligent planning using forecasts and machine learning to create models that predict different situations. We built an intelligent spatial platform, combining SAP Cloud Platform, SAP HANA, SAP Leonardo Machine Learning Foundation, and SAP S/4HANA. Using satellite and drone imagery, weather data, and machine learnings, the system provides daily end-to-end monitoring of crops to determine abnormal areas to inspect for possible pests and diseases before they spread.

The solution provides a set of automated alerts and advice based on AI classification of several radiometric indices from weather IoT and more. The infrastructure enables speed and agility, and is also lower maintenance for IT teams.

Farmers can manage the costs of the inputs and the date of harvest to make an appointment to move the grain in time for the harvest. For the input, in the beginning of the season, the ACA uses machine learning to provide the farmers with the requirements for the materials they'll need, such as seeds, fertilizers, chemicals, operators, and services.

How long did it take to build this solution? Did you do all the work in-house or did you also use outside consultants?

There were many different stages to this project. In total, it took two years to complete the project. The first stage took about six months and the other phases of the project were each three months long. We worked with the SAP HANA Special Services team based in Germany to create this project. We have our support and weekly meeting with the team of developers.

What does location intelligence provide you? Was this information collected at the ACA level or provided to individual farms as well? How did each group (the ACA and farmers) use the information?

The location intelligence provides us with analysis and insights into the conditions of crops, fields, and soil. With this project, we finally have a GIS and geodata interface build. This innovative system provides information from the fields using satellite and drone imagery, weather information, and other sources. This, combined with business data from the ACA, provides all-in-one, day-to-day monitoring of crops to determine anomalous areas to inspect for possible pests and diseases.

The system calculates all possible radiometric indices, then generates vector maps for the entire coverage area of the field. The process is carried out every five days, whereas before using SAP we were not able to get this information at all.

What were the project challenges you knew you would face and -- more important -- what challenges hadn't you anticipated? What proved to be easier than you expected and what proved to be harder?

One of the biggest challenges was creating a team of people that work together efficiently with different skills. We were working with data scientists, agri-business traders, developers, and user experience designers. We needed to create a model to work together for all of these different users.

Running results from the use of machine learning and IoT information from remote sensors proved to be very powerful and these results exceeded our expectations.

What lessons did you learn? If you had to do it all over again, what would you do the same and what would they do differently?

One of the best lessons we learned with this project is that we are dealing with customers, dealers, workers, and operators -- and we needed to create and engage with these people. It takes time to adjust to using new technology. Something else we would change is the order in which we did the development and research. If we had to do it again, we would do the research phase first instead of doing it in parallel with the development phase.