Drones, Sensors, Satellites and Cameras: New tools for managing a vineyard for sustained production
August 05, 2019
|This soil electrical conductivity sensor generates data about a vineyard’s soils as it is pulled through the vineyard rows by a gator.
|A map of the Clearview Farm vineyard soils was generated from data produced by a soil electrical conductivity sensor to define the soils underlying their vineyard|
|The rotational speed of the shoot thinning heads attached to this tractor changed as the tractor passed by cones marking the different zones on a prescription map|
Many grape growers across the country understand the concept that balancing vegetative growth and cropping levels on their vines can result in sustained production from those vines. There’s one problem: what do growers do in vineyards that vary on a vine-by-vine basis? What tools can help them so that this principle can be applied in those variable vineyards?
The Nelson J. Shaulis Symposium, held in conjunction with the American Society for Enology and Viticulture-Eastern Section meeting in Geneva, NY, presented a program July 17 and 18 on “Digital Viticulture: New Tools for Precision Management.” The first day included a vineyard tour and demonstrations of new technology in action and the second day featured a series of talks by researchers and other experts on the process of implementing precision management in vineyards. The overall goal of the symposium was to introduce attendees to new tools and agricultural technologies that are making it possible for growers to identify the variability within a given vineyard and then adapt viticultural processes on a vine-to-vine basis to achieve management objectives for that entire vineyard.
Demonstrations of digital technology for vineyards
On July 17, two large buses took Symposium participants to Clearview Farms, located west of Branchport, NY above Keuka Lake and operated by the third and fourth generations of the Tones family. The farm was founded in 1951; when the third generation – Don, Harold and Jim – became involved, production shifted from cows to grapes, and the family currently manages 1,000 acres of farm and woodlands, with 500 acres in vineyard. In 2012, the Tones acquired the neighboring winery, Stever Hill Vineyards, which is now run by Don Tones’ daughter Elizabeth Sprague and her husband Jay, both of whom graduated from California Polytechnic State University’s wine and viticulture program. Liz’s brother Mike and cousin Jon manage the vineyards.
Since 2015 the Cornell Lake Erie Research Extension Laboratory (CLEREL), under the direction of Dr. Terry Bates, has been working on a precision viticulture project known as the Efficient Vineyard project, which is funded by the USDA-Specialty Crops Research Initiative program of the National Institute for Food and Agriculture. The project was designed “to deliver an innovative, science-driven, and approachable precision viticulture platform to measure and manage sources of vineyard variation.” Accordingly, there are three parts to the program:
Measure: where proximal sensors are used to collect spatial vineyard soil, canopy, and crop data which are validated with manual in-field observations;
Model: in which multi-layer spatial-data processing is used to visualize vineyard performance and generate grower defined management maps; and
Manage: where spatial management maps are integrated with precision agriculture technology and vineyard machines such as tractors and sprayers for variable-rate field applications.
In the past, soil core samples have been taken manually in order to determine the physical, chemical and biological properties of a vineyard’s soil. Now, according to Jackie Dresser, a research technician at CLEREL, a GPS sensor on top of a gator can log its position while a soil electrical conductivity sensor is pulled behind the gator (Photo 1). This procedure had been done on the Clearview Farm, and the data collected was used to map the soils underlying their vineyards. The maps created then can be used for vineyard management decisions such as where to apply lime – more in clay soils and less in sandy soils (Photo 2).
Heather Barrett, the newest member of the Efficient Vineyard team, reported on the use of proximal canopy reflectance sensors located on farm equipment such as a tractor, ATV or gator to track how the vineyard is growing. Two weeks before bloom, shoot counts can be used to determine shoot thinning; at bloom, leaf area measurements can be recorded; two weeks post-bloom pruning weight can be measured. Barrett did note, however, that while drones and sensors have improved, the growth of cover crops can interfere with the sensor data for vines, and growers will still need to go into the vineyard to get pruning weights from actual vines to compare with the mapping data.
Jennifer Russo, appointed viticulture specialist at CLEREL in May, spent the past year developing and testing an automated grape counting machine to help estimate and forecast crop levels in the vineyard. By using available spatial soil, canopy, and crop information, stratified crop sampling can be used to get a mid-season crop estimate that is better than traditional random sampling. A machine has now been developed that incorporates automated cluster weighing with grape counting for rapid and accurate automated berry counts and weights that will help with crop estimation.
How to turn spatial sensor data into useful information for vineyard management was the topic addressed by Dr. James Taylor, senior researcher at the Institut National de Recherche en Sciences et Technologies pour l’Environnement et l’Agriculture at the University of Montpellier, France, who was introduced as the “spatial data processing guru.” Raw data from multiple sensors can be processed by computer to smooth the data observations by removing data from signals that are outside those associated with what the sensor is being used to detect. The data file is then shaped to fit the border of a vineyard and the file is programmed to generate in-field rate trials and variable rate prescription maps. The integration of spatial data and decision support with precision agriculture technology and vineyard mechanization has led to mechanized variable rates for shoot thinning, fruit thinning and fertilizer applications.
For example, Ag Leader hardware and software can be used onboard a tractor to import and show prescription maps, and flow rates are assigned to the different treatment zones on the prescription map. GPS tracks where the tractor is in the vineyard and on the prescription map, and when the tractor enters a new management zone, the signal to the liquid flow controller changes. That signal is sent to a variable-rate hydraulic valve, which controls the rotational speed of the shoot thinning head. A speed sensor on the thinning head allows the translation of the flow rate to RPMs and shoot thinning rate.
A tractor in the Clearview Farms vineyard was equipped with shoot thinning equipment and an Ag Leader computer monitor; as the tractor passed by cones marking the different zones on the prescription map, the rotational speed of the shoot thinning heads changed according to the instructions of the prescription map (Photo 3 and 4).
After lunch at Stever Hill Vineyards, the attendees went by bus to Anthony Road Wine Company’s vinifera vineyards on western side of Seneca Lake just south of Geneva. The first demonstration was presented by Hunter Adams, a fourth year Ph.D. student in mechanical and aerospace engineering at Cornell. Adams has designed the prototype for a weather “satellite” called the Monarch that can gather localized data on light, moisture, temperature and other factors that would help vineyard managers to maximize their harvests (Photo 5). A total of 20 of these in-canopy environmental sensors have been placed in one of the vineyards at Anthony Road. He hopes to design the final version of the Monarch during his last year at Cornell and then establish a company to sell the Monarch sensors to vineyard owners.
Dr. George Kantor, senior systems scientist at the Robotics Institute at Carnegie Mellon University in Pittsburgh, PA, brought two versions of his 3-D cluster imaging cameras (Photo 6). They each have a built-in computer and can take “really good photos with really good images, even when the vehicle is driving on rough ground.” The images produced were used to make predictions about harvest in California, with only a 2% inaccuracy rate. The more recent version of the camera costs approximately $8,000 to build.
A more inexpensive vision-based yield-prediction camera program is being developed by Dr. Justine Vanden Heuval, associate professor of viticulture at Cornell University, and a student, Jonathan Jaramillo. They used a smart phone camera in a camera holder to take videos in vineyards to increase the accuracy of yield estimations and cluster density mapping early in the growing season. Photos are taken in late May through early June, before heavy foliage occurs and the canopy closes.
In between rain showers, attendees went into Anthony Road’s vineyard to watch a self-driving Monroe tractor go down a row while applying sprays at variable rates (Photo 7). While billed as “self-driving,” an operator needs to be on board to take over if necessary. The automatic driving feature does allow the operator to pay more attention to other operations such the spray program rather than just concentrating on driving the tractor (Photo 8).
New tools for precision management
The second day of the “Digital Viticulture: New Tools for Precision Management” symposium began with a session on vine balance and precision viticulture, which was followed by sessions on the three major parts of the Efficient Vineyard project: Measurement, Modeling, and Management.
Dr. Andy Reynolds, professor of viticulture at Brock University in Ontario and a former student of Dr. Nelson J. Shaulis, began the program with a talk on “What Nelson Shaulis Taught Us.” The Symposium carries Shaulis’ name for very good reasons. As professor of viticulture at Cornell University’s New York State Agricultural Experiment Station in Geneva, NY from 1944 until his retirement in 1978, he had a major influence on students for more than 30 years. Shaulis continued to attend meetings of the ASEV-Eastern Section and other conferences, and his questions and comments were often those of a major professor working with a student (even though that “student” was by then also a professor).
Shaulis, and others, developed principles of vine physiology that have formed the basis of modern precision viticulture for more than 50 years. He was especially known for the development of the Geneva Double Curtain trellis system. Reynolds noted that “the first divided canopy was made by [Thomas] Munson, but the Geneva Double Curtain was the first that worked.” Shoot positioning was easier to do, more fruit was exposed and, importantly, the vineyard could be mechanized.
“Balanced Pruning” was another concept that Shaulis developed over the course of his career. He came up with pruning formulas for different varieties that were based on the weight of cane prunings that would bring each vine into balance. Vanden Heuval in her talk on the physiology of vine balance noted that researchers after Shaulis continued to refine the definition of balance in vines. One of the more recently developed equations for crop load is the Ravez Index, which is the ratio of yield from the current harvest compared to the pruning weight of those vines during dormancy following the harvest. Vine balance is the ration of fruit weight to leaf area, or the ratio of vine yield to vine size. “Pruning weight is easier to measure than leaf area,” she said. “Basically, vine balance is the balance between carbon sources and carbon sinks.”
Dr. Nick Dokoozlian, vice president of viticulture, chemistry and enology at E&J Gallo Winery in Modesto, CA, spoke on the “New metrics: examining grapevine response to crop load through a different lens.” He noted that canopy size determines the vine’s ability to ripen fruit; digital viticulture allows the measurement of leaf area to fruit weight and yield to pruning weight; crop load impacts sugar and color.
Then he asked, “What else matters?”
The explanation that followed was complex, deserving of a separate followup article.
The three sessions on measurement, modeling and management included talks by many of the professors and researchers who had participated in the vineyard tour the previous day.
Measurement talks included:
Jackie Dresser discussed mapping vineyard soils using proximal sensors;
Dr. George Kantor spoke on 3-D image processing for cluster and berry counts, and noted that the current 3-D camera counts the visible grape berries on VSP trellis systems and predicts the tons per acre with close to 99% accuracy, but whether it will work on other trellises hasn’t been determined;
Dr. Jim Meyers talked about remote sensing by satellites, proximal sensing on the ground, and aircraft and UAVs such as drones in between;
Dr. Alan Lakso discussed the development of nanotensiometers for measuring vine water status and noted that the best single measurement comes from stem water potential. He hopes to release a continuous-reading water potential sensor for use in vineyards by spring of 2020;
Dr. Tim Martinson spoke on behalf of Dr. Andrew Landers about ultrasonic sensors for variable rate spray applications. He noted that infrared sensors can monitor growth stages and gaps in the vines, and the information generated can be used to apply spray where it is needed and in appropriate amounts, thereby reducing spray volume.
In addition during the session on measurement, Katie Gold, a Ph.D. candidate in plant pathology and an MS candidate in biometry at the University of Wisconsin-Madison (she will be joining the Cornell Agritech faculty as assistant professor for grapevine disease ecology and epidemiology in February 2020), reviewed the five main types of sensors, which include digital cameras, multispectral, hyperspectral, thermal and LiDAR (laser sensors). She noted that digital cameras and multispectral sensors can be used to determine the stage of crop growth and the precision timing of important sprays, while sensors for detection and differentiation of specific stress factors are not yet available commercially. For example, hyperspectral sensors are not yet ready for use in vineyards.
The session on Modeling included two talks:
Dr. James Taylor reviewed how viticulturists process spatial data and validate what the sensors measure, and how spatial data is then turned into “management zones.” According to Taylor, the most critical point is determining what the user wants the support systems to do, then the data can be processed appropriately.
Dr. Hakim Weatherspoon, associate professor of computer science at Cornell University, started his talk with a goal: “How do we feed 10 billion people by 2050 -sustainably?” To do so, it will be necessary to transform agriculture, including vineyards, into Software-defined farms (SDF). Sensors will produce data, that goes into local computers, then to the Cloud; that data will be analyzed and put into programs that the farmer can access on his SDF. He concluded, “A SDF is an integrated systems approach, a system of systems, designed to comprehensively and flexibly incorporate many different models and inputs.” However, currently such a comprehensive system has yet to be created for any complex agricultural system.
The Symposium concluded with two sessions on Management:
Richard Hoff, director of viticulture at Mercer Ranches in Prosser, WA, manages more than 3,300 acres of vineyard, which he describes as an on-going experiment. The company decided to mechanize for economic reasons – “we’re out in the middle of no-where,” he stated. Hoff reported that Mercer uses a mechanical pruner, does shoot thinning, sucker removal and mowing mechanically, uses a suck-‘n-cut deleafer, and harvests with a mechanical harvester and destemmer. However, after doing trials with their equipment, they continue to move wires on the trellis by hand. Hoff added, “Our sensors [for counting buds or clusters] are people.”
In the final session, Dr. Terry Bates reported that the Efficient Vineyard project in California and New York is wrapping up its work on integrating spatial crop load and soil mapping into practical management plans. He concluded that there are is plenty of technology available, but you have to collect the data so that you can use it. “Large operations may be able to use it; for smaller growers, it depends,” Bates said.
|A monitor in the tractor cab displays the prescription map for the tractor operator|
|Hunter Adams talked about the Monarch vineyard sensor that provides in-canopy temperature, humidity, wetness and light data|
|A current version of a 3-D cluster-imaging camera is mounted on the back of a gator to produce data on the crop yield of a given vineyard|
|The self-driving Monroe tractor can navigate vineyard rows while the operator pays attention to the spraying process|
|The Monroe tractor drives down a vineyard row while the operator applies sprays at the rate appropriate for that point in the vineyard|