Transforming the agriculture industry using IoT and predictive analytics

IoT is enabling the digital transformation of the agriculture industry. But infrastructure constraints, limited talent and digital skills, and the need for regulation around the capture and exchange of data present some challenges.

By Tas Bindi, Feb 1, 2017

Agriculture is commonly cited as an industry that has much to gain by harnessing the Internet of Things (IoT), with some even saying that it could be “the first big industrial IoT market”. Advancements in technology mean that farm equipment, soil, waterways, and even livestock can now be instrumented.

This gives farmers the ability to automatically collect objective information about the status of their soil, water, crops, and animals. But the potential of IoT to transform agricultural efficiency, improve financial performance, and boost yield is best achieved when it’s combined with data analytics and machine learning.

Australia has strong productivity levels in areas such as cropping and red meat — in 2016, Australia’s crop productivity was 19.2 tonnes per hectare compared with the world’s productivity of 14.6 tonnes per hectare. However, its overall agricultural productivity growth rate sits at 1.4 percent, below the global average of 1.7 percent, according to a report by StartupAus, KPMG, Commonwealth Bank, and the Queensland government.

Added to that, rising production costs — including prices of seeds, fertilisers, water, electricity, and machinery — are placing increasing pressures on Australian agribusinesses to more efficiently utilise their resources and increase output.

But manual data collection processes make it difficult for farmers to achieve optimum levels of efficiency, especially given the geographic dispersal of their farms. After all, there are more than 123,000 agribusinesses in Australia occupying 50 percent of the nation’s land, according to the StartupAus/KMPG report.

This is where technology is playing an increasingly important role. The sophistication of today’s sensors, internet-enabled devices, software applications, and cloud data storage facilities are allowing vast amounts and types of data to be captured, stored, manipulated, and fed into decision-support tools to guide business decisions.

According to a 2016 report by the Australian Farm Institute (AFI), what we’re seeing is the initial stage of “a fundamental change away from the skills-based farm management systems that have prevailed until present times towards a more industrialised model of agriculture where decisions are based to a greater degree on objective data.”

As an example, Tasmanian salmon producer Tassal has been using IoT technology — provided by Dell EMC and Intuit — to improve efficiency and reduce operational costs in the business.

Sam Boyes, senior manager for ICT at Tassal, told ZDNet that a typical lease area from which the company operates will consist of 20 pens holding approximately 20,000 salmon each. The pens have been fitted with smart cameras above and below the water to get more accurate information about the water’s oxygen levels, temperature, and current. The software to which the cameras are connected can then identify the optimum time to feed the salmon and release food accordingly. This ensures the company maximises the salmon’s growth patterns, while also minimising food wastage, said Boyes.

“That’s the single biggest cost in our business — the feed. So the last thing we want to do is waste any of it or not get the return on it through [the salmon’s] growth.”

Boyes added that a reduction of one kilo of food every day saves the business AU$25,000 per year.

“Any small improvements we can make there are quite significant. If we can reduce any [losses] by understanding what’s happening in the environment — for example, if there’s a fast-running current — then we can make decisions about when and how we feed those fish so that even a small amount of food isn’t taken away by the current,” he said. “These are the types of insights we’ve been able to get to reduce the cost of operation.”

“Likewise, when the fish want to eat and the feeding conditions are good, we can make sure we are maximising the opportunity for fish to grow well,” he added.

Tassal isn’t using predictive analytics yet, but Boyes admitted that it’s an exciting area and that the company is headed in that direction.

An agtech company that’s looking to bring predictive analytics to the forefront of the Australian agriculture industry is The Yield, founded by Ros Harvey, who was previously the Sense-T program director at the University of Tasmania. The initiative uses data, sensing technologies, and analytics to solve problems for Tasmania’s agriculture ecosystem.

Leveraging technology from Microsoft, Bosch, and Intel, The Yield’s microclimate sensing system is able to analyse components such as salinity levels in the water and predict when oysters need to be pulled out. For example, if there was a storm and the farmer knew contaminated water was going to flow towards the oyster farm in the upcoming days, then they can take action to pull the oysters out.

“Regulators used to open and close oyster harvesting in Australia based on rainfall as a proxy for runoff. But the problem is that rainfall can be coming from 100km away or more,” Harvey said, adding that about 30 percent of rainfall-driven closures are unnecessary. “What we do is put in real-time salinity, temperature, and depth sensing into the water that oysters drink. Effectively, what that means is the regulator is able to open and close harvesting using that information. If the salinity goes down that’s because freshwater is coming in.”

Harvest closures, especially during peak holiday periods, can cost farmers up to AU$120,000 in lost revenue per day, Harvey said. To date, The Yield’s microclimate sensing system has been able to reduce unnecessary harvest closures by 30 percent, and is currently relied upon by 300 oyster farmers, with both the Tasmanian and New South Wales state governments being customers.

“It’s not helpful to know something unless it’s an actionable insight,” Harvey said. “This is why IoT is going to be massive; effectively, we have the ability to know what we couldn’t know before, and we have the ability to use that data to get insight in a way that we haven’t before.”

Standards and data-sharing

When it comes to data, the general consensus is that open standard data platforms and data sharing protocols will add significant value to agribusinesses as greater intelligence can be drawn when datasets are combined with other datasets.

This is why Tassal is exploring partnerships with data providers such as the Commonwealth Scientific and Industrial Research Organisation (CSIRO).

“[They] can give us more wide-scale environmental data and we can overlay that with the data we’re capturing on our sites to find additional patterns that can help us,” Boyes said.

There are a number of use cases, as presented in the AFI’s report, that show how combined datasets are delivering greater value to farmers. For example, data collected from soil sensors and crop monitoring systems are being combined with local and hyperlocal climate data, enabling farmers to make better decisions to increase crop yields. On some farms, sensors and monitoring systems are connected to automated watering and fertilisation systems to maintain an ideal soil environment, facilitate healthy plant growth, and maximise crop turnover.

There are livestock identification and monitoring systems that can not only provide almost real-time information about the geolocation of the livestock, but also biometric information such as activity levels, weight changes, blood composition, changes in heart rate, milk characteristics, and ruminal parameters that can be fed into a business intelligence platform to help with decisions around animal health, welfare, and reproduction.

Ben van Delden, partner, head of markets and head of agtech at KPMG Australia, pointed out that Australian farmers are operating about 24 hours behind North American farmers in terms of the speed with which data from the device can be overlaid onto farm management systems and publicly available data sources such as the Bureau of Meteorology.

“Our issue is that there is a poor backhaul of data to bring intelligence back to the devices for decision application — for example, dependence on satellite time passes. This restricts the ability for real-time gains from integrating the collection of data on crop or livestock performance with farmer interventions,” he said.

“Farmers and growers are incredible risk takers when you think about what they do. [They’re] constantly making decisions about when to plant, when to irrigate, when to protect, when to feed, when to harvest.”

According to a Commonwealth Bank Agri Insights survey of 1,400 Australian farmers, 76 percent see the value of sharing on-farm production data with others, with 58 percent of farmers already doing so.

The survey, published in October 2016, also shows that the agriculture sector is generally receptive to potential productivity gains associated with the deployment of digital technologies, with 70 percent of respondents agreeing that the technology available would bring significant value to their operations, and 20 percent expecting to increase technology investment in the next 12 months.

Harvey said, however, that some farmers are sceptical of new technology.

“I don’t think it’s because farmers aren’t innovative — they are. Farmers and growers are incredible risk takers when you think about what they do. Growing something is an incredibly complex thing to do. It’s not like a factory. You’ve got all these different variables to take into account. You’re constantly making decisions about when to plant, when to irrigate, when to protect, when to feed, when to harvest,” Harvey said.

“I think the issue is we’ve not been good in the technology sector in delivering reliable, accurate technology. Because of that, often the experience hasn’t lived up to the promise and so there’s a fair degree of scepticism.”

Infrastructure issues

While the extent of challenges vary across different agricultural production systems, infrastructural constraints apply across the board when it comes to IoT implementation.

Many rural and regional areas have poor electrical and network connectivity, with relatively small download allowances, and at a higher cost and slower speed than the services available in metropolitan cities. As such, many Australian agribusinesses are operating with broadband connections that are inferior to those available in countries competing for similar export market opportunities such as Canada, Denmark, and the US, the AFI report states.

The AFI said lack of access to mobile telephone and data coverage can be a “major impediment” to technology adoption, advising the government to increase available funding to ensure equitable and affordable access to telecommunications infrastructure in rural and regional areas and to investigate the potential for public/private co-investment models to further enhance data coverage in those areas.

“Most farms will have some form of 3G or 4G connectivity, but in many cases, it’s pretty poor. In many cases, there’s congestion. And in many cases, they’re frustrated. Clearly, the faster we can get broadband and reliable connectivity to them, the better,” said Kevin Bloch, CTO at Cisco ANZ.

“But there’s a whole area of technology that’s coming through now that we’re providing which is what we call low-power, wide-area network (LPWAN) technology. In other words, we can now put in the capability by using an access point that can get you 5 to 10 kilometres of coverage, which is pretty cool because a lot of sensors don’t send a lot of data. You don’t need a lot of bandwidth, but you need bandwidth to be kilobits per second.”

In 2015, the University of New England (UNE) began transforming a 2,900 hectare commercial farm called Kirby-Newholme into a SMART Farm (Sustainable Manageable Accessible Rural Technologies Farm). The Kirby Smart Farm, an initiative led by the Australian Centre for Broadband Innovation in collaboration with CSIRO and the UNE’s Precision Agriculture Research Group, was one of the first farms in Australia to be connected to the National Broadband Network (NBN) fixed wireless service and is often referred to as an example of what quality broadband can enable.

In addition to using a low-cost wireless tracking system and video monitoring technologies to monitor animal behaviour and location, the farm has deployed sensors to collect weather, light, and soil data. Local wireless networks allow the sensors to send a continuous stream of data to a remote cloud-based computing and analytics service.

While rural and regional access to fast, reliable IT infrastructure is considered a “vital precursor” to building a strong agtech sector, van Delden said the question agribusiness leaders need to ask themselves right now is not: “How do I get data when and where I need it when connectivity is disparate?” Given there are more and more sophisticated sensors coming to market, van Delden said the right question is actually: “What am I going to do to use the sensors and data to augment my business and decision making processes?”

“This is not about adding IoT solutions to your business as a bolt-on; it is about fundamentally thinking about how do I embed IoT into my business systems and reinvent my processes,” van Delden said.

“It’s critical that you reimagine your business enabled by IoT and think through how the supply chain you are part of can be reorganised and made more efficient. You need to realise that IoT creates a dynamic environment that can speed up the way decisions are made within the supply chain … and enable real-time interventions to optimise production.”

James Kavanagh, CTO at Microsoft Australia, communicated a similar sentiment, saying that digital transformation should not be thought of as simply being the automation of an old process; rather, it’s about changing those processes.

Kavanagh also said that it’s possible to work around infrastructural constraints.

“We do have some infrastructure constraints in Australia from a networking perspective. But it’s possible to design with those constraints in mind,” Kavanagh said. “There are many places in the world that would regard Australia’s infrastructure as fantastic and yet still have IoT capabilities.”

“We’ve undertaken projects in Africa and Asia where there is almost minimal wi-fi or any kind of connectivity and satellites required. We just design around those constraints.

“Sometimes it requires a different architectural approach — such as more power at the device end to manage cache and perform the analysis — but there are ways around it.”

Boyes said Tassal has been able to work around those constraints, but warned that it can be challenging from a financial and technical standpoint, as well as distracting.

“You have to tackle [those challenges] yourself, but at the same time you want to make sure you’re not distracted from your core business,” he said.

Boyes added that it’s not always worth agribusinesses building their own mobile phone networks, and it can be helpful to understand the roadmaps of infrastructure providers to see if there’s an opportunity to work together.

“Quite often there’s an opportunity to influence where [the providers] might start expansion or bring improved coverage to. Having good relationships is definitely key,” he said.

“The centre of excellence for food production is here in Australia and that’s where we want to grow the agricultural technology sector.”

“If you are embarking on it yourself, you’ve got to make sure you don’t go too far left field and make sure you cover all your bases like security … and put the complexity into a small area that you can easily compartmentalise and then transfer and communicate the key bits [of information] that you need.

“We identified that implementing these sorts of strategies would provide us with a competitive advantage, so we invested a lot of time and a lot of money to get that competitive advantage.”

Skills gaps

Another barrier to IoT adoption is the skills and talent required to manage and get maximum value from technology deployments.

Dr Mike Briers, CEO of Food Agility, a consortium of commercial companies, universities, and farming industry bodies looking to facilitate the digital transformation of agriculture industry in Australia, said improving digital literacy would help reduce the fear that might be preventing farmers from adopting IoT. Food Agility is currently working to understand what education is required.

According to the StartupAus and KPMG report, Australia is seeing an increase in enrolments for agricultural university degrees and there is an opportunity to upskill the agriculture sector in STEM subjects and digital literacy.

“Agtech requires people with skills that are not traditionally connected to agriculture including engineers, software designers, data analysts, and business support services,” the report states.

Meanwhile, van Delden noted that the average age of Australian farmers is increasing — currently, it’s 57 years old.

“This creates an opportunity for the large change in ownership and shift we are seeing to corporate farming where increasingly young technology-driven farmers are rushing into the sector,” van Delden said.

“The top farmers are seeing the need to embrace the opportunities IoT solutions can bring them to access and utilise the best data in their businesses faster. We are seeing interest from family offices and impact investors in agriculture and agtech businesses, as they can see the productivity gains and improved sustainability technology brings.”

He added that IoT highlights the need for collaboration across the agriculture and food processing supply chain.

“Digitisation of landscape allows a whole-of-landscape decision-making process which enables decisions down to a square metre level or scaled up to a national level,” van Delden said. “IoT means that knowledge will become common knowledge. Market dynamics will shift as a consequence of the supply chain getting access to insights on regional crop yields, infestations, and quality. This will change the speed of which agriculture commodity trade prices will move.”

An agricultural ecosystem

A collaborative agricultural ecosystem is gradually forming. Food Agility, which raised AU$160 million and is seeking a further AU$50 million in public funding to be spent over 10 years, will be working with partners to build an ecosystem that connects all stakeholders along the value chain. The organisation is a major proponent of IoT and is particularly keen on creating a common knowledge base via open standard data platforms and data sharing mechanisms.

“[Australia’s] got excellent brand reputation internationally for high quality green food … We believe the centre of excellence for food production is here in Australia and that’s where we want to effectively grow the agricultural technology sector,” Briers said. “For the food and agriculture sector to reach a higher level of maturity from a digital transformation perspective, you’ll need to have a stable and reliable underlying measurement system.”

Innovation Central Sydney (ICS), an innovation centre focused on agriculture, is another example of various parties coming together to support the industry. Via ICS, Cisco, Data61, and the University of New South Wales, in partnership with the National Farmers’ Federation, the NSW Farmers Association, the NSW Department of Primary Industries, and ATP Innovations, are identifying and developing new uses for IoT within the agriculture sector.

Kavanagh said collaborative efforts, particularly between large enterprises and small niche providers, is becoming more and more commonplace in the world of IoT, citing The Yield as an example of what can be achieved when different parties come together to solve a problem.

“The magic of it is it’s a partnership between a small Australian company called The Yield, Microsoft, the Tasmanian government — each bringing their unique elements of capability, but together providing this platform that solves a problem,” Kavanagh said.

Van Delden said IoT can and should make it more attractive for banks, private investors, and superannuation funds to finance the agriculture sector, which has historically struggled to attract sufficient domestic and foreign investment.

According to the StartupAus and KPMG report, the Australian government provided AU$306 million of R&D support to agriculture between 2015 and 2016, of which AU$252 million was allocated to rural research programs. This represents 1.3 percent of gross farm receipts in 2015, compared with the US at 9.4 percent and the EU at 18.9 percent.

The report also highlights that in 2015, 0.3 percent was invested in the agriculture sector by Australian superannuation funds. Like Canadian superannuation fund PSP Investments, which acquired a significant stake in agtech company Allflex Group, the report states that Australian superannuation funds have an opportunity to play a bigger role.

“Growers (borrowers) will need to be open to lenders having access to more real-time insights on farm performance,” van Delden said. “Financial institutions will need to think about how they can enable sharing of performance insights and analytics they collect to help their farming clients lift their performance.”

“For way too long, [farming has] been a human-oriented business, with a lot of guesswork involved. With technology, we can really help farmers produce more and be more profitable.”

While all state governments in Australia are supporting the agriculture industry to varying degrees by providing funding, driving collaboration efforts, supporting research, and building agtech solutions, the general consensus is that they can do more to encourage the advancement and adoption of IoT in the industry, particularly from a regulatory standpoint.

“With food production now Australia’s largest manufacturing sector there is good reason for government to support the technology adoption in this sector,” van Delden said. “Key areas for government to support are in embracing regulation of landscape, farming practices, transport and logistics, biosecurity, and import and export requirements. Embracing an IoT-based risk management framework has significant potential to improve efficiency and profitability to the businesses involved in the food production supply chain.

“Reimagining how we regulate food production and environmental compliance will help accelerate the uptake of agtech and IoT as well as give the consumer greater confidence in the provenance and food security environment that our produce has passed through to reach their plate.”

Van Delden added that regulation built around the capture and exchange of data from the paddock through to customs clearance can release significant speed and efficiency dividends through the supply chain.

He also pointed to the need for data standards and the interoperability of devices so that stakeholders can have confidence in the quality of the data that is collected and how it is shared.

At the moment, there is little clarity around data ownership such as whether farmers or technology vendors own the data that is being captured, stored, and shared.

“Data ownership and use rights become more opaque in the event that the cloud storage or farm software facility transfers the data to a third party — even in the event that such a transfer is permitted by the farmer whose land the data was originally generated,” the AFI report states. “Ideally, data use and access permissions should ‘travel’ with the data, although whether this is legally enforceable by the original data owner is unclear, and the extent to which this applies to anonymised data is also unclear.”

Van Delden pointed out that as data moves through the supply chain, an additional challenge emerges around leverage.

“As we know, knowledge is power, and in the supply chain context, this insight can change your negotiation leverage. Contractual arrangements are and will change significantly because of IoT,” van Delden said. “Not simply due to technology like blockchain to enable the transaction, but in how yield optimisation dividends can be shared in the supply chain. Done right, this could and should be the opportunity for producers to pull more margin back inside the farm gate.”

While IoT is still a maturing field, with many technologies and information systems still undergoing development, most agree that agribusinesses can still secure productivity gains from adopting it even if it requires significant initial monetary and time outlays. The maturity of cloud from multiple vendors, coupled with the cost of sensors, routers and hardware coming down, means developing and deploying an IoT solution is easier now than it was as little as two years ago.

“[The agriculture industry] can definitely do with some technology enablement,” Bloch said. “For way too long, it’s been a human-oriented business with a lot of guesswork involved. With technology, we can take a lot of the guesswork out of farming and really help farmers produce more and be more profitable.

“I think the farming community really get that now.”