IoT and satellites combine for sustainable irrigation
Image credit: Nasa
Eyes in the sky and sensors in the ground combine to make sure crops don’t dry out.
Agriculture is the world’s biggest consumer of water, using 70 per cent of the Earth’s freshwater supplies. Irrigation alone accounts for some 1,200km3 (2.6 quintillion gallons) of water per year. Much of this consumption - three times more than 50 years ago - is only set to increase, raising the global water use of agriculture a further 19 per cent by 2050. With overpopulation and climate change just two stress factors, the Earth’s stocks of freshwater are diminishing. According to the World Agricultural Report, 52 per cent of the world’s population will live in areas where freshwater is under pressure by 2050.
A recent study led by the University of California found that a third of the world’s biggest groundwater sources are already running out. In fact the problem is already so bad that the World Economic Forum classified water availability as the biggest threat facing the planet.
California has become something of a poster boy for the problem. Having suffered from five successive years of drought, the largest producer of food in the US and the fifth largest in the world, has been in a state of emergency since 2014. With groundwater levels depleting and crucial snowpack at record low levels, the sunshine state is in a position where, according to meteorologist Eric Holthaus, “wasteful agriculture is literally sucking California dry”.
Perhaps it is fitting, then, that a water-saving solution should come from one of California’s many small producers, and that, coming from the home of Silicon Valley, it should involve the Internet of Things (IoT).
Five years ago Kurt Bantle bought a 12-acre avocado farm in Fallbrook, southern California, as a weekend escape from his day job as a solutions architect for telecommunications firm, Spirent. With farming time limited to the weekends, Bantle soon realised he would need a way to automate the process of irrigating his trees. “I started looking online,” he says. “I found lots of stuff for the big row crops but for something of 12 acres there wasn’t a lot of stuff out there. So I started making some of my own stuff along with some off-the-shelf components.”
Bantle divided his farm into 22 zones of 50 trees and in each zone planted two moisture sensors at different depths. “The one at 8in depth tells me when to water and the 24in one tells me how much water to put on,” explains Bantle. He then set up an irrigation system connected by solar-powered radios so that all the irrigation valves could be controlled remotely. Every ten minutes the sensors send their data to the cloud where it can be accessed and visualised by an IoT platform, GroveStreams, on a browser or smartphone app. When the 8in sensor detects low moisture levels, irrigation is triggered until water reaches the 24in sensor, which switches it off. Bantle even receives a text to let him know it’s happening.
Bantle knew this would economise on time but he didn’t realise just how much it would cut down on irrigation. The system has slashed his water use by 60 per cent. “What creates water saving is that we’re not following typical irrigation schedules,” says Bantle. “People typically put 12 litres of water on every four days and you ask the question, why - what if the trees don’t need it? I have zones that can go 10 days without water.”
Bantle’s is a small-scale solution for a small farm. Satellites have traditionally provided water monitoring on larger scales, but as imaging technology improves, even these distant objects could provide the kind of granular data to help individual farmers save water.
Nasa’s Landsat satellites have been measuring the rate at which the surface of the Earth loses water since the early 2000s. Landsat’s thermal infra-red detector measures the surface temperature of objects on the ground, combined with visible and near infra-red sensors to measure vegetation cover. “The variable that’s of interest is evapotranspiration,” says Jeffrey Masek, project scientist with the Landsat-9 mission, “which is the amount of water that leaves the land either through direct evaporation or because plants are pumping it back into the atmosphere as part of their metabolism.”
Measuring evapotranspiration gives an important insight into how stressed plants are and therefore how much nutrients and water they need. Crucially, whereas this kind of data is also available from other satellites, Landsat provides it at a granularity that is useful to farmers. “Meteorological satellites have similar thermal IR bands but they tend to be at kilometre or half-kilometre scales,” says Masek. “We’ve got it down to 100m.”
Landsat data is currently used to monitor water consumption at US state level. Getting the information to farmers is a challenge. “That’s the roadblock with satellite technology,” says Masek. “There are a few hundred people who can download the data from the provider and come up with algorithms to provide useful information, but you’re a long way off the farmer in northern India.”
However experts are working on a solution. Researchers from the universities of Nebraska and Idaho are currently developing an application called EEFLUX that will enable farmers to access Landsat information on their smartphones, using Google Earth Engine to produce field-scale maps of water consumption.
While Landsat measures evapotranspiration, two other satellites, one from the European Space Agency and one from Nasa, monitor soil moisture content directly. ESA’s SMOS satellite uses an array of radiometers to measure the brightness temperature of the Earth’s surface at the microwave level. From this, soil moisture content can be calculated. Nasa’s SMAP uses a similar ‘passive’ radiometer combined with an ‘active’ radar, although the active part has unfortunately died. Both satellites aim to use soil moisture content readings to improve knowledge of the Earth’s water cycle and hence better predict meteorological events like droughts and floods. However, both satellites operate at resolutions of 30-50km making them unsuitable as yet for individual farmers.
One way of achieving finer resolutions with satellite technology is by bringing it much closer to Earth. A Nasa-funded project to adapt SMAP is currently being conducted by drone company Black Swift Technologies. Originally developed under a Nasa fellowship at the University of Colorado to use SMAP technology on manned aircraft, Black Swift obtained a Nasa small business grant to continue scaling it down to drone size. Black Swift Technologies’ sensor uses a microwave radiometer similar to SMAP’s combined with an NDVI sensor operating over visible and near infra-red bands to measure vegetation density, alongside a thermal sensor to correct for varying surface temperatures. “The system is effectively the same as the Nasa SMAP sensor,” says Professor Albin J Gasiewski, the original head of the project at the University of Colorado, “but of much greater spatial resolution since we fly so close to the Earth - about 15m resolution, versus 40km for SMAP.”
Scaling down such complex instrumentation has come with lots of challenges, not least in developing drones that can carry it securely while taking off and landing in remote areas. Black Swift Technologies is experienced in creating drones for extreme conditions such as those found in the Arctic, and has even designed drones capable of chasing tornadoes. One of the specific challenges of building a soil moisture-mapping drone was keeping the device at the same height above the ground to correlate all the measurements. “Flying at a constant 15m faces a lot of obstacles,” says Black Swift founder and CEO, Jack Elston. “We had to make a terrain-following controller to keep it a uniform distance from the ground, so we’re not running into hills or things like that.”
Then there is the amount of data that needs to be processed. Currently post-processing takes about a day, according to Elston, with the goal of eventually getting data to the user in near-real-time. Black Swift’s soil moisture drone has been testing over fields in Colorado all summer. Next season Elston hopes to start data testing and move to commercialisation. “We need to find 10 growers interested in using the system to see how we can increase their yield,” he says.
Another entrepreneur bringing satellite technology closer to Earth is Javier Marti, an expert in electromagnetics who worked on the SMOS satellite. Marti uses sensors mounted on drones or fixed installations at heights of 2-10m above the field, providing sub-metre resolutions. He employs a technique invented by scientists of ESA called GPS reflectometry. It taps into the near-universal coverage of GPS and exploits the energy of its signals to measure their reflection from the ground. “By looking at the direct and reflected signal,” says Marti, “you can derive how much soil moisture in particular is affecting the reflection from the ground of the GPS signal.”
Marti’s company, Divirod, is currently testing prototypes over two farms in Colorado and in semi-arid farms in the south of Spain, where they are gathering data and validating the technology. Marti hopes to launch the first commercial agricultural product at the beginning of 2017, before extending its use to water-level monitoring to study climate change. “If you think of our concept installed in harbours, you can have a network of harbours and be data monitoring the water level difference on a continuous basis,” he says.
Satellites, drones and soil sensors connected to the IoT - three technologies with the potential to solve the problem of water use in agriculture, but which one will win? Marti believes that drones are the best technology for monitoring soil moisture, pointing out the lack of resolution of satellites and the assumptions that need to be made with soil sensors. These ground-based sensors, according to Marti, “can only tell you the soil moisture content at that particular point, from there you have to make an extrapolation through modelling what sort of composition the soil has and how the hydrology will work”.
On the other hand, as Tom McKinnon, founder of agricultural drone company Agribotix, points out, soil sensors give a better depth profile of soil moisture and thus a better idea not just of where to water but, crucially, how much.
However you do it, McKinnon says, measuring soil moisture is only one part of the problem. Delivering the water is a whole other challenge. “The problem isn’t so much sensing it as being able to take action,” says McKinnon. “There are different kinds of irrigation systems. The most common is still flood irrigation and with that what are you going to do?”
Another common method, especially in the US Midwest, is centre pivot irrigation where a line of nozzles revolves around a centre point. But adapting the system to variable rate irrigation, where individual nozzles operate independently, can be prohibitively expensive - around $70,000 for a single pivot, according to McKinnon. It means that, until irrigation technology catches up, ultra-accurate soil moisture sensing might only be practicable at small scales like Bantle’s avocado farm.
Despite the challenges, everyone agrees on one thing - the importance of continuing to push the technology. “I think it’s critically important,” says Nasa’s Masek. “There are areas where agriculture is currently available where it probably won’t be in a hundred years, so to maximise yields we have to look at techniques like these.” Bantle agrees that the application of water-saving technology is critical although the solution holds a certain amount of irony for him. “I tried to get into farming to get away from technology,” says the avocado famer, “but the more I get into it the more I’m like, wow, we need more tech in agriculture.”
Satellites and sensors: water hotspots
Nasa’s Landsat satellite calculates the amount of water being used by vegetation down to resolutions of 100m. It does this by estimating evapotranspiration from infra-red emissions.
Landsat’s thermal infra-red sensor measures the surface temperature of the Earth and its visible and near-infra-red bands measure the amount of vegetation cover. Since the solar energy used for evapotranspiration is what is left from the energy used to heat the ground and air at the surface, calculating evapotranspiration is relatively simple.
“You know the amount of radiation coming in from the Sun,” says Landsat project scientist Jeffrey Masek, “and you can make assumptions about the energy transferred into the ground. You know the wind speed and you can use the surface temperature from Landsat to calculate the sensible heat. Evapotranspiration is the latent heat left over from that energy balance calculation.”
Finding the energy used to heat the air was the difficult part of the calculation - it couldn’t be done using Landsat’s measurements directly. This was solved by going over historical Landsat images to find examples of near-zero and near-full evapotranspiration. Once these two extremes had been found, air temperature could be calculated without directly measuring it.