Farming is one of the oldest industries in the world, with evidence of agriculture being found as far back as 21,000 BC. However, with an increasing population and environmental disruption through climate change, farming needs to find new solutions to improve food security.
One way more food could be grown is through using data analysis to improve the efficiency of farming. By analysing data on weather patterns, soil conditions, pest invasions, sunlight levels, market forces and other factors, farmers could identify the optimum times to sow, treat and harvest crops, as well as identify the best crops to maximise returns for particular conditions.
Making decisions based on data is not new for farming. Since 1792, the Old Farmer’s Almanac has provided US farmers with information regarding weather forecasts, planting charts and astronomical data. Meanwhile in the UK, Farmer’s Weekly provides a regular source of information and updates designed to aid farmers. These publications, and others like them, have provided regularly updated information to enable good decision-making.
Giving farmers more data and sophisticated tools with which to analyse it will enable better-informed decisions regarding how to manage crops. There is already a vast amount of historical weather data that can be used for predicting weather patterns. Although climate change tends to increase the likelihood of extreme weather events, such events are being incorporated into existing climate models.
There are many sources of data that relate to farming. Stock market data can be collected to identify which crops are performing better than others. Sensors can be used to collect data on soil quality and moisture. Drones can be deployed to monitor pests and weeds, and satellites sent into orbit to collect data on the amount of sunlight in a field.
WWF and TechUK recently published Crops to code: The role of data in fostering sustainable agricultural trade and responsible supply chains. The report highlighted the role of data and technology in promoting sustainable agricultural practices and responsible supply chains globally.
The study emphasises the importance of data monitoring using mobile technology and digital platforms at the production level, to ensure supply chain visibility and sustainability. Additionally, the authors made strategic recommendations to the UK government to scale up technological innovations.
The report concludes: “Data harmonisation provides numerous opportunities – from enabling farmers to better understand the impact of their production and the supply chains they are a part of, to supporting financial institutions to make informed decisions, to verifying claims of sustainable production or distribution.”
Trying to get labour on a farm is a big problem, so farmers are open to using technology in all different forms David Ross, Scotland’s Rural College
One of the key benefits of data analysis is the ability for incoming data to be continually monitored over a period of time and alerts transmitted when issues arise. For example, sensors could identify the optimum times to irrigate or fertilise crops.
Through identifying trends within existing data, data analysis systems could predict the likely outcomes. Using predictive solutions, such as a Monte Carlo simulation, would enable farmers to identify market trends or probable growing conditions for different crops.
“We are now collecting data to not only support decision-making in growing agricultural products, but it has moved into the area of predicting,” says Mark Wolff, advisory industry consultant and chief health analytics strategist for the Global Internet of Things (IoT) Division at SAS.
“If I do these things in this combination at this time, what should I expect given a certain assumption about irrigation?”
Crop trials and tribulations
Computer simulations could be used to predict what happens when different crops are grown or a new routine is tried. Previously this would have required the farmer to trial growing the crop for a season to learn first-hand the return on the crop, which is time-consuming and could lead to losses.
“The next level of analysis is the digital twin – a simulation of a workflow,” says Wolff. “Once you have a mathematical set of relationships between inputs and outputs – the crop, the genetic composition of that crop for a particular geography, a set of inputs for chemical and biological agents, and an outcome – you can then simulate that.”
There are more than just financial issues that farmers need to consider in agriculture. Often farmers will choose crops for reasons other than immediate financial gain. Crop rotation is a common practice in agriculture, which is where different crops are grown in sequence for pest and weed control and to improve the soil.
“Rather than telling us which crop to grow, there is scope for identifying emerging outbreaks of a disease in part of a field, like an early warning system,” says David Ross, principal consultant for arable services at Scotland’s Rural College.
As with everything, financial viability is key. Few farmers could afford a comprehensive data analysis software suite. Although this might be a one-off payment, it would take time for the returns on investment. If the product then becomes obsolete or no longer supported, further significant investment would be required.
For these reasons, agricultural technology companies are looking into the practicalities of offering data analysis services as a subscription service to farmers. This overcomes the problem of the initial high cost of the software.
However, investment would still be needed for a variety of sensors. Many agricultural companies have already generated large datasets they could use as a baseline model for agricultural data analysis.
A data collection network would be required. This incorporates the sensors for collecting data into a network that transmits the information for storage and analysing. Installing such a network could also be a significant one-off expense, similar to the process of connecting a business to telecommunications services.
Overcoming connectivity issues
Furthermore, network connectivity remains a problem in rural areas, especially in the more remote regions of the country. There are even rural locations that still do not receive mobile phone signals. For these data analysis solutions to be effective, there needs to be further investment in the rural telecommunications infrastructure.
One potential challenge to analysis-driven farming is the lack of information-sharing between different providers. There is little to no interoperability between devices of different manufacturers, meaning farmers have lots of different datasets, but no way to combine them together to see how the they all interact with each other.
“We have dozens of pieces of software that the farm uses to input and output data, but I refer to them as the data silos,” says Ed Harris at the Data Science Agriculture Research Centre at Harper Adams University. “They don’t speak to each other and you only get information about your farm.”
Being able to combine multiple datasets together would provide a more holistic overview of a farm’s operations, rather than a series of independent data inputs. However, sharing this data while also protecting each farm’s personal and financial information would be essential for such a service to remain compliant with data protection regulations, such as the UK’s Data Protection Act 2018.
Being able to compare data with other farms would provide farmers with a greater understanding of how their farms operate. Independently, this information would be worth very little, but by being combined it would enable baseline estimates of how a typical farm operates. This would enable farmers to identify areas on their farm where they need to focus their attention.
“A pet peeve of mine is that if you get those numbers from a single farm, the farmer will have that one number,” says Harris. “What I think is needed is a business case to create benchmarking for carbon and for other farm operations. I then have a little more information in the context of the other farms.”
There are also regional elements to consider. A data analysis system developed using farms in a particular part of the country will not necessarily work in another.
There are environmental and geographical issues that need to be adjusted for each data analysis solution to be effective. “We worked on a project from Australia and it worked perfectly, but you bring it into Scottish conditions and it’s just a disaster as it was a different environment,” says Ross.
Alongside data analysis, automation is being used to address recent labour shortages. These are typically repetitive and mundane tasks such as milking cows, which allows the workforce to focus their attention on more complex tasks.
“Trying to get labour on a farm is a big problem, so farmers are open to using technology in all different forms,” says Ross. “A robot on a dairy farm costing £30,000 a year has the potential to save the farmer £85,000 a year in labour.”
Although agriculture is not seen as an IT-driven industry, it is one that is reliant on needing reliable and accurate data to enable informed decision-making. Data analysis could be used to identify emerging trends within the datasets and alert farmers to when intervention may be required to protect crops and increase their yields.
“Farmers are cynical buggers. It needs to be proven and there needs to be a return on their investment,” concludes Ross. “That’s not always cash, as it can be time. It needs to be relatively simple, robust and work on a messy farm and in dirty environments, because there’s just no getting around that.”