Aug 11, 2016If you are a fan of the avocado, you know that your desire for the fruit often comes at a steep price. That's partly because it is a very thirsty crop. In California, farmers typically expend 74.1 gallons of irrigated water to grow a pound of avocados, according to Mesfin Mekonnen and Arjen Hoekstra, researchers at the University of Twente, in the Netherlands. That amount of water is far greater than what is needed to grow the same amount of tomatoes, strawberries or lettuce.
Five years ago, Kurt Bantle, a senior solution manager at Spirent, a company that provides evaluation and development services for telecommunications networks, devices and applications, decided to try his hand at avocado farming. He quickly learned about the burden that water footprint takes on growers.
"I farm 12 acres, which is a small size for an avocado farm," Bantle says. When he purchased the farm, Bantle considered cutting the cord completely from his tech job. But then the financial reality of managing a 900-tree avocado orchard, during what has become a historic drought in California, set in.
Most avocado growers, Bantle says, use 4 acre-feet of water to irrigate each acre of trees per year. But at $1,200 an acre-foot (1 acre-foot equals 326,000 gallons), water consumption can make avocado farming a losing proposition for small growers. "We could never afford to operate the irrigation system for 12 hours every four days," he says, referring to a baseline consumption of 4 acre-feet of water per acre (a total of 48 acre-feet) per year.
As luck would have it, Bantle was also tasked with helping to develop an IoT network offering at his day job. So he decided to turn his avocado-growing hobby into a work project.
Bantle started by surveying the marketplace for IoT systems designed for agriculture. "There are IoT solutions for large-scale operations," he explains, "such as big soybean farmers growing thousands of acres, but there was nothing for small scale." He began by testing the use of soil-moisture sensors that communicate via a ZigBee mesh network, but found that managing the network quickly became difficult. Due to the nature of the mesh network, every data packet was transmitted over every node, which meant he was "sending a lot of data over a small pipe," and consuming a great deal of battery power. "I probably didn't have the most efficient topology set up," he admits. "I had every [sensor] acting as a router. Nothing ever slept."
Rather than trying to optimize the ZigBee sensors, Bantle decided to evaluate different technologies. At the time, low-power, long-range radio networks (LPWAN), operating in an unlicensed ISM band, were starting to hit the market and Bantle was intrigued by LoRa, a LPWAN protocol developed by IBM, Semtech and more than 10 other technology and telecommunications providers.
The LoRa network that Bantle deployed has been running for roughly one year, and comprises 20 moisture sensors that measure the soil's water level based on its electrical resistance, as well as 10 thermistors. Two moisture sensors (one positioned near the topsoil and another set more deeply) and one thermistor are wired to each microcontroller, which is paired with a LoRa radio, provided by Multitech Systems.
Every 10 minutes, the sensors transmit their readings over the LoRa radio to a gateway, provided by Multitech. The gateway contains an embedded SIM (eSIM) provided by Spirent partner Oasis. Embedded SIMs are chips that are soldered into devices and programmed, over the air, with a SIM profile that allows the device to be commissioned to a specific cellular network but which can later be overwritten with a different operator's profile. (Remote SIM provisioning is an approach that the mobile phone industry association GSMA developed in order to improve the versatility and lifecycle of machine-to-machine hardware that may be deployed in large quantities or remotely. This would make the task of physically replacing SIM cards impractical in the event that the organization responsible for the network would like to switch to a different cellular operator.)
The moisture data is collected in a cloud-based software platform that Spirent is preparing to commercialize, modeled on the system that Bantle developed. When the moisture level falls below a set threshold, the platform sends the message back down to the gateway, over the cellular network, and the gateway relays a command to the LoRa-based valve controllers on the irrigators. The sensors continue to transmit moisture and temperature data to the cloud-based software, via the gateway, and once a sufficient amount of water is applied, the system closes the irrigation valves.
To determine his irrigation schedule, Bantle previously relied on a local weather station and the California Irrigation Management Information System (CIMIS), which calculates evapotranspiration rates for zones across the state, which farmers use to determine how much they should irrigate. But now he uses the sensor network to allocate the water.
So far, the network has paid significant dividends, Bantle reports. His goal was to reduce his water consumption by 50 percent, but he is actually using only a quarter of the water he had been using prior to deploying the sensors, because he also decided to rip out the mature trees in the orchard and replace them with new ones, which require less water overall. Still, he is bullish on the technology and believes that—even as the trees grow and demand more water—it will enable him to maintain a far more water-efficient operation than he ran prior to the network installation. "The trees are not at full maturity yet, but when they are, we'll be at 50 percent of the water we used to use, or maybe even better."
All told, Bantle spent $8,200 on the hardware (the LoRa radios, the sensors, valve controllers, a LoRa gateway and a cellular backhaul), plus his monthly cellular subscription cost of $60. Yet, by reducing water consumption by 75 percent, he was able to achieve a return on his investment in just six months.
