Sensors and big data provide farmers with decision-making support
January 24, 2018
By Maria Fernanda Ziegler and José Tadeu Arantes | Agência FAPESP – Researchers in the field of e-agriculture like to repeat an anecdote according to which farmers used to tell their children they should study if they wanted to leave for the town but now tell them to study if they want to stay on the farm.
Changes are under way not just in production but also in management. Big data combined with the Internet of Things makes plants and animals themselves the key decision-making support for rural producers. Precision agriculture already uses information technology to analyze climate, soil and other variables. Now, the producer’s management decisions can also be based on the signals emitted by crops and livestock.
“It’s no exaggeration to say that in the next ten years, we’ll see plants and animals responding to each stimulus in the production system. Decisions about crop and animal welfare will be based on data collected by biosensors implanted in each plant and each head of livestock,” said Iran Oliveira da Silva, a professor in the Biosystems Engineering Department of the University of São Paulo’s Luiz de Queiroz College of Agriculture (ESALQ-USP) in Brazil.
Silva was a speaker at the eScience Workshop on Agriculture in the Digital Age held in FAPESP’s auditorium in São Paulo to examine changes in agriculture due to the use of information technology.
“In the future, animals will talk, and you’ll remember me,” he said. “The ability to make decisions based on signals emitted by animals is already a reality in scientific research in various parts of the world. Studies in bioacoustics and animal vocalization enable us to use sound frequencies to understand an animal’s expression. Belgian vets can use technology to find out from ambient sound whether parts of a herd have diarrhea or tuberculosis, for example. And the error rate is under 1%.”
The workshop was attended by researchers affiliated with institutions throughout São Paulo State, who discussed the future of agriculture in Brazil in a setting where soybean fields, coffee groves, poultry farms and cattle ranches will be full of sensors that will help produce data and information on the need for more irrigation, ventilation, soil treatment or medication.
“Besides sensors, chips and biomarkers, many farmers are also using microdrones rather like mosquitoes equipped with cameras and sensors. They collect masses of data and can also carry out pollination,” said Jansle Vieira Rocha, a professor in the University of Campinas’s Agricultural Engineering School (FEAGRI-UNICAMP).
For Rocha, futurology is a much-needed activity, especially in sectors that are undergoing major changes, such as agriculture. “We have to think ahead and develop innovations,” he said. “We have to take advantage of seminars and meetings like this not just to present academic results but also to think about solutions to problems we don’t yet have.”
Brazilian agriculture is at the forefront in several ways, according to Rocha, including no-till cropping and integrated crop-livestock-forest systems, implemented in 11.5 billion hectares across Brazil according to the Brazilian Agricultural Research Corporation (EMBRAPA). “Much of our success is based on the scientific research supported by EMBRAPA, but in information technology, sensors and imaging, we still lag behind other countries,” he said.
A key issue in e-agriculture is how to combine the various kinds of data continuously generated by sensors in plants, animals, vehicles and machinery with climate and production data in such a way that all these data can be converted into useful information for farmers.
“This is a typical discussion in e-science, allowing researchers in computer science to dialogue with and understand the problems of researchers in other fields. It’s important to guarantee this dialogue,” said Claudia Bauzer Medeiros, one of the head coordinators of special programs for the FAPESP Research Program on eScience and Data Science and organizer of the workshop.
For the researchers who attended the event, multidisciplinarity is crucial to implementing e-agriculture effectively.
“It’s most important to consider the question of professional multidisciplinarity in agriculture,” Silva said. “We’re now experiencing the Internet of Things, high technology and data analysis in cropping and animal husbandry. For this reason, it’s fundamental to get people like animal scientists, vets, agronomists, biosystems engineers, agricultural and computer engineers and so on sitting around the same table to solve problems.”
Connectivity in the countryside
One of the obstacles to e-agriculture in Brazil is lack of connectivity in rural areas. “Insufficient connectivity in the countryside hinders technological progress in agribusiness. Vast areas of Brazil have no signal at all. In Mato Grosso State, for example, which is a major soybean producer, you can travel 400 km without being able to use a cell phone. Here in São Paulo State, network coverage misses out areas within 36 km of important cities,” said Carlos Lorena Neto, an engineer at CPqD, an information and communications technology (ICT) company, during his presentation.
“Telcos aren’t interested in putting up antennas in rural areas. It isn’t profitable for them to invest in network coverage there. But other networking solutions are available to farmers,” he told Agência FAPESP.
In his presentation to the workshop, Lorena Neto described an Internet of Things project implemented at São Martinho, a leading player in the Brazilian sugar and energy industry. Located in the municipality of Pradópolis, São Paulo State, the group’s rural properties produce 10 million metric tons of sugarcane per year in an area with a radius of 84 km, mostly without cellular network coverage.
With financing from BNDES, the national development bank, under the Plan for Joint Support for Agricultural Technology Innovation in the Sugar & Energy Industry (PAISS Agrícola), the project included development of an integrated broadband mobile communications and remote sensing platform to enable applications such as real-time control of farm implements and sugarcane traceability during harvesting and processing.
“We designed a base station similar to those used by telcos, except that it operates at a different frequency to ensure better propagation so that the radio signal goes further. The base station is connected to sensors [which monitor the location and functioning of the sugar mill’s machines] that send the data to this network,” he explained.
Lenira El Faro Zadra, a researcher at the Animal Husbandry Institute of the São Paulo State Agribusiness Technology (APTA), presented the state of the art in wearable and remote sensor systems for animal health, applied in this case to dairy cattle herds. Multiple variables such as feed consumption, temperature, hoof and udder health, milk composition, and methane emissions can be monitored using wearable sensors and remote sensors fitted to equipment at different places on the farm.
“A specific sensor made by an Austrian company to monitor the movement and temperature of cows with very high milk production is available on the market,” Zadra said. “It’s inserted via the esophagus into the cow’s reticulorumen and measures the animal’s temperature every ten minutes.”
Using the data collected, the cow management system detects not just heat stress or the occurrence of fever due to infection, but above all the temperature variations associated with the reproductive process. “Temperature rises during estrus, about 24 hours before mount acceptance, and falls prior to calving,” Zadra said.
The cylindrical sensor is 13 cm long and 2-3 cm in diameter. It can store data for 50 days, downloading when the animal approaches the base station. Its battery lasts four years.
The same company also makes a sensor to monitor changes in ruminal pH due to diet. This device can operate for 150 days. It enables the farmer to be sure the cow is coping well with the special diet administered to support very high milk production in terms of digestive tract acidity. Neither sensor can be removed from the animal.