Is it possible to increase the yield with a smarter use of available data? The IBM T J Watson Research Center has developed a platform for the use of Big Data in precision farming. The results speak for themselves: 20 percent more output and 10 percent more efficient use of water.
Three years ago, E & J Gallo, a large winemaker based in California, found that despite steady drip irrigation they were not growing grapevines evenly. Some vines produced more, others stayed behind, and this affected the harvest. The solution would be to give the branches that stayed behind additional water, but how could you select the right ones?
Satellites collect a lot of information about crops.
The research team of Levente Klein accepted the challenge. Research resulted in the platform PAIRS (Physical Analytics Integrated Repository and Services), where Big Data is used to control irrigation. IBM’s research division is the largest private research institution and employs 3,000 people worldwide. “We are investigating many different subjects,” outlines Levente. One of the research areas is the use of Big Data in agriculture.
The technical side of the project is quite complicated. In addition to the collection of data, you need to find a way to spatially align it; furthermore, that information must be converted into advice, for example, for irrigation. The PAIRS platform makes use of public satellite data. By looking at all the information, lessons can be learned. Levente illustrates this with an example.
Simply with a detailed relief map of the area you can already see which parts of a field are higher than others. On that basis, you can find out where rainwater is collected and where it just flows away quickly. “When you combine that with data on the vegetation index, you get a first insight into the yield,” explains Levente. “If you then add the data about the soil to it, then you have a third indicator. In this way, the context becomes richer and you get a better understanding of what happened on the field without the need to inspect it physically.”
The bottleneck in the use of Big Data is to combine all the data and process them in a way so that they fit together. “The satellite images cannot always be directly superimposed.” These images are obstructed by sun position, clouds and aerosols, which cause small irregularities in the images acquired by satellites. “We have to correct the data in order to calculate exactly what happens on the field.”
IBM also wants single sensor point measurements combined with the satellite images. The challenge is that single point measurements contain data on a small area, but many times, while satellite images are the exact opposite, as those images cover a large area and are less frequently taken.
With all the data on, for example, the surface, location, altitude, weather, historical data, etc., the PAIRS platform can plan an irrigation or fertilization schedule for the next ten days. “That limit was chosen because the weather forecasts of the different services are reliable for up to ten days. Additionally, a grower will only need to load the data into the computer once a week. Shorter intervals are also possible and the system is updated automatically.”
With this system, E & J Gallo managed to increase the yield by 20 percent. “But the larger harvest is not the only result,” concludes Levente. “The quality of the grapes also improved. By adjusting the irrigation, the sugar content and brix of the grapes became more uniform, and this is an important factor to determine the quality of the wine.”