The system integrates data on weather, soil conditions with satellite imagery to determine how much water each section of the vineyard needs.
A California winery has turned to the Internet of Things to grow high quality grapes while optimising the amount of water applied to the vineyard.
E. & J. Gallo Winery, a family-owned winery headquartered in Modesto, California, has been working since 2012 on improving water use efficiency using remote sensing. Using NASA satellite imagery, the winery determines the vine canopy size and water status, as well as vine water use. The Gallo team noticed a high degree of variability in vine water use across their vineyards, which led to a partnership with IBM Research.
For its part, IBM applied its expertise in IoT, physical analytics and cognitive computing technologies to co-develop a precision irrigation method and prototype system and install it in a 10-acre vineyard. This work provided the first large scale scientific evaluation of grapevine response to hyper-local precision irrigation, and showed that that this method could reduce water usage by 25%, according to IBM Research.
The system integrates data on weather and soil conditions with satellite imagery and other sensor data to determine precisely how much water each section of the vineyard needs to produce the highest quality grapes. Before the prototype was developed, irrigation levels could only be adjusted at the vineyard block level and did not account for individual vine requirements, according to IBM Research. A common farming practice prior to this finding was to irrigate all vines in the vineyard at the same level, leading to some vines getting too much irrigation and others not enough.
IBM scientists came up with an advanced irrigation system built with a series of sensors and actuators in the 10-acre plot that communicated with a central vineyard control system. Using geo-spatial data such as soil, climate and imagery from hyperspectral satellites, the system predicted vine irrigation needs that sent signals to open valves and released the precise amount of water to each vine, according to IBM Research. Physics-inspired machine learning technologies were used to establish the irrigation schedules.
The irrigation system has been installed in six different vineyard blocks encompassing nearly 250 acres in California. Gallo said these blocks are experiencing a 20% increase in water use efficiency, which means that they use 20% less water for each pound of grapes that they produce.
IBM, meanwhile, is applying what it learned from the project to an experimental research effort that involves aggregating, indexing and analysing terabytes of open geo spatial data.