Big data on the farm
This is an English translation of the article ‚Big Data auf dem Bauernhof ‘ written by Sebastian Balzter and first published [in German] on 25 October 2015 in Frankfurter Allgemeine Sonntagszeitung (page 30). © All rights reserved. Frankfurter Allgemeine Zeitung GmbH, Frankfurt. Courtesy of Frankfurter Allgemeine Archiv.
Big data on the farm
Driverless tractors, cows milked by robots – this is how agriculture works today: thanks to accurate data
By Sebastian Balzter
The scene could be set for a tearjerker: an old farmhouse, horses in a paddock, mild evening light. The view stretches far beyond the fields up to the Harz mountains with its forested slopes. In the barn, you can see the benches and tables from recent thanksgiving celebrations.
Or is it science fiction? The landlord sits in front of a computer screen. The results of recent soil samples arrive from the laboratory. Drones with cameras are buzzing through the air to examine farmland while the tractor drives autonomously in the field.
Agriculture in the year 2015
We are talking about agriculture in the year 2015, so it’s a bit of both worlds. On the Derenburg estate in Saxony-Anhalt a motorist’s dream has come true: taking the hands off the steering wheel, focusing on the control screens in the cockpit, leaving the steering process to the computer. The tractor is pulling the four-meter-wide seed drill over the canola field on the outskirts of Magdeburg in straight lines, row by row, exactly in parallel, minimizing overlap.
Three different displays show the current power usage of the tractor, how much seed per hectare is being distributed by the seeding machine, and how much farmland still remains to be cultivated. The slot for the USB stick – which was previously used to upload field map data – is no longer used as Klaus Münchhoff, senior partner at Derenburg estate, has already emailed the data package from his office to the tractor in the morning.
Mr. Münchhoff grows canola, barley and wheat on approximately 1,000 hectares. His farm thus is one of the largest in Germany. And one of the most modern – because Klaus Münchhoff collects data.
Few farmers know as much about their land as he does. Mr. Münchhoff uses drones to shoot aerial images of the 40 fields on his farm. Specialists measure the electrical conductivity of the soil. He has divided each of his fields located around the town of Derenburg into up to a hundred smaller parcels so as to capture the specific varieties in the soil, even within one and the same field. Finally, Mr. Münchhoff has compiled hundreds of data entries for each of these parcels into digital maps: infrared images, biomass distribution, fertilization cycles.
"Cars find the road, tractors find the parking slot"
Around 9,000 different data sets are now on the hard disk. Together, they are about three gigabytes in size. The investment was considerable, it was certainly not cheap. But Klaus Münchhoff is convinced that it will pay off. According to him, he can maximize the precision of the modern machines (on which he has already spent a lot of money) only with the help of these maps. In other words, only by so doing, the Derenburg estate becomes a farm 4.0.
Today, auto-steered tractors are already a standard item on large farms, tractor models above 200hp are barely produced anymore without an integrated automatic guidance system. For these tractors to remain more precisely on track than conventional GPS, special receivers have been installed all across the countryside during the past 20 years.
Apart from the lower average speed and lower traffic density on the field when compared to a highway, these receivers are the decisive advantage which tractors have when compared to passenger cars in terms of autonomous driving. A margin of error of up to two centimetres is just about acceptable for agricultural machinery aficionados. With standard GPS, this margin is at best 5 metres, oftentimes a lot more. According to the German Engineering Association VDI, in the agricultural machinery value chain, sensor technology, electronics and software already make up 30% of the value chain. By comparison, this share is only about 10% in the car industry, which already likes to celebrate the expected future breakthrough of automatic driving. "Cars find the road, tractors find the parking slot", is how experts summarize the difference.
Keyword Precision Agriculture
Mr. Münchhoff is a great believer in all things traditional. Otherwise, he would not invite all of his employees, neighbours and friends to the annual thanksgiving party in the rustic barn. But he is not religious about sticking to well-trodden paths. That’s why, apart from the automated driving system for his four tractors and the combine, he has also bought a fertilizer spreader with an integrated camera. When fertilizing, the sensor detects in real-time the amount of plants in the field and adjusts the required amount of nitrogen fertilizer accordingly. The data from the soil maps is added to the process: in a furrow, there is no point to fertilize because the sandy soils cannot support the growth of many crops. Accordingly, the machine will stop. The sensor for the fertilizer spreader has cost 40,000 euros, says Mr. Münchhoff. Is it worth it? “We need six kilos less nitrogen per hectare. The crop now grows in a more uniform manner, which is an advantage during threshing. And the yield has risen by 100kg per hectare. In the end, this results in a surplus of around 25,000 euros – in just one single year.”
Precision Agriculture is the key word for the development in recent years, and for Precision Farming you need accuracy. The fertilizer bill of the Derenburg estate reveals the logic behind this thinking. Urban dwellers tend to think that, in the countryside, there is so much space that one or two centimetres more or less don’t matter. But the opposite is true: since the area is so large, the accuracy is worth the while. An example: the canola field in which the tractor pulls his tracks is 800 by 200 meters in size. At four meters working width, the tractor drives the seeding machine 200 times back and forth, each time for 200 meters. An overlap of 10 cm will thus result in 4,000 square metres of overlap area. That’s half a football pitch. A big waste. Moreover, diesel, time and seeds are expensive.
Four times more wheat than in 1950
According to the Bavarian State Research Centre for Agriculture, digital farmers manage to work with 10% less herbicides and 20% less fuel than its counterparts without sensors and automatic driving systems. On this point, not even farm romantics will find modern technology a silly thing. Apart from that, they still find sufficient things to criticize about the technological progress. The risk that farmers’ data could fall into the wrong hands – such as speculators, which could thereby obtain an undue advantages for their stock exchange transactions because they can assess from the data how the harvest will turn out – , is not even the most prominent concern. A catchier point of criticism: if you reduce duplication of tasks, there will be fewer jobs. And a lot of capital is put into new equipment, so larger farms have an advantage. Not everybody likes that.
However, the same development can also be expressed in a very different way: in no other sector of the German economy has labour productivity increased as much in the past five years as in agriculture. And in no other industry today the capital investment per job is higher. According to the German Farmers’ Association DBV, around 465,000 euros are behind every job on the farm, twice as much as in the manufacturing sector. The result: farmers today harvest four times more wheat per hectare than their ancestors did in 1950. With potatoes, yield levels have doubled. And cows produce three times more milk than before.
Cows check-in at the milking robot, just as skiers check in with their lift pass at the lift
Technological progress has not only changed arable farming. Computers and big data have also arrived in cow and pig stables.
The 75 cows on Torsten Ebeling’s farm in Wolperode in the German region of Lower Saxony wear a sensor around the neck. This sensor allows them to check in at the milking robot at the end of the barn just as skiers check in with their lift pass at the lift station. The machine will then know whether the cow has already exceeded its target and has only come to the milking parlour to get an additional piece of the delicious power feed with which the cows are rewarded after milking. Cheaters are denied entry and reward. However, if the cow still is allowed to have another turn, she may well enjoy the reward. Meanwhile, lasers detect the position of the udder and teats and bring the milking arm of the robot in position. Milking and milk yield per teat are recorded and stored in real time. And after milking, the udder is automatically sprayed with a cleanser.
Cows get used to this treatment. The robotic milking services in Torsten Ebeling’s stall are in high demand. The tiny cleaning robot which collects the manure and drops it into an underground storage through the gaps in the stable ground does not interfere with them. At the moment, Antonia is being milked. After four and a half minutes milking time, the display reveals a yield of 7.6 kilograms, a decent average.
Additional comfort as a welcome side effect
Mr. Ebeling, a farmer from a small village near Bad Gandersheim does not refrain from giving his animals names, even if the computer assigns each cow an individual identity number. Antonia carries number 3 860 630. As soon as the milking system detects an irregularity, Mr. Ebeling receives a message on his mobile phone. Then he leaves his farmhouse in the village to go to the stable amidst the pastures to investigate. Modern livestock farming does not work without people. “But I do not have to get out here every morning at five o’clock and spend hours in the barn. Instead, I can usually have breakfast with the family before the children leave for school,” says Mr. Ebeling. “That's a big difference.” Even a week of holidays in autumn is now within reach for the farmer.
The gain in comfort is a welcome side effect, as also the Dutch manufacturer of livestock equipment Lely underlines during its sales events. However, as is the case with combine harvesters with an email address, the main focus is operational efficiency. The standard version of a milking robot costs round 100,000 euros. That money needs to pay off. All data are immediately available in digital form, which facilitates the calculation. In addition, the milking rhythm can be reduced to the largest possible amounts at the lowest possible cost. And the recorded movement profiles of the animals support better decision-making: if a cow in the barn moves around more than usual, it's probably in rut; on the contrary, if it barely moves, it should be seen by a veterinarian. The latest development is a specific algorithm that calculates from different datasets the exact date when the next calf will be born.
As was the case with the auto-steer tractors, the first milking robots became fashionable about twenty years ago. “But during the past five years, there has been a real boom in demand,” says Daniel Stove from Lely’s office in Germany. The reason for this development lies in the opportunities that accurate data has opened up for the use of the technology. One gigabyte per hundred cows is the usual order of magnitude. In this way, Lely promises, you can reduce production costs of typically around 30 cents for a litre of milk on average by two additional cents. That does not sound like a lot. But it's like in the canola field: size makes a difference. And anyone who cannot be convinced by additional comfort or commercial considerations might simply enjoy working with state-of-the-art technology: recently, Lely already presented the next conceptual step: data transfer from the stable to the display of an Apple watch or Google Glass. The positive reactions from farmers show they are definitely more interested in advanced technology than in tearjerkers…
© All rights reserved. Frankfurter Allgemeine Zeitung GmbH, Frankfurt. Courtesy of Frankfurter Allgemeine Archiv.