Current Research

Zania Pothen and Stephen Nuske

Harvest dates play a crucial role in determining the quality of grapes. The choice of harvest dates and locations, particularly in red table-grape grape varieties, is planned based upon the color development of the grape clusters which indicates the maturity of the fruit. The traditional process for evaluating the amount of ripe, fully-colored fruit is done manually by visual assessment. But this process is subjective, prone to errors and the number of locations where a grower will evaluate the fruit development is statistically insufficient given the size of commercial vineyards and the variability in the color development. Therefore, an automated approach for evaluating color development is desirable.

We, at CMU, have developed a vision based algorithm that can categorize clusters of grapes into different stages of fruit maturity and measure the rate and variability of color development across large commercial vineyards. For instance, we measured the rate of change of the grape from its initial immature green colouring to the ripe reddening of its skin.

We collected images from 3 consecutive growing seasons of the Flame Seedless table-grape variety, from 2014 to 2016. We used color development measurements on these images. Doing this, we were able to measure high-resolution current state of color development and also we have constructed a prediction model that maps the current state of the fruit maturity to its future state, within a growing season. We repeated this process for the Crimson Seedless variety for a single growing season, that of 2016, and were able to develop a similar model.

Figure 1: Grape clusters categorized into 4 grades depending on their color development.

We categorize the clusters of fruits by their color development into four grades, A to D, with grade A being least mature fruit and grade D being harvestable (Fig 1). For our initial calibration, we randomly select 30 plots in the vineyard. Each of these calibration plots is visually assessed by an industry expert – a viticulturist or grower – and a color development value is assigned for each plot for each of the four grades. The images of vines from the calibration plots are paired with the corresponding color development value and grade assigned by the industry experts (Fig 2).

Figure 2: Automated grading of grape clusters. (Left) Raw image of grape clusters in the field. (Right) Image of detected grape clusters and their associated grade based on color development. Berries that are well-colored (Grade D) are highlighted by red, immature berries (grade A) are highlighted by green.
Figure 3: Correlation between Human measurements and Image estimates for the (a) Flame Seedless and (b) Crimson Variety.

We combine the percent of clusters of individual vines for a given grade across the vineyard to generate a spatial map as shown in Fig 4.

Figs. 4b and 4c present maps of the color development on two separate days for Grades B and D obtained using image measurements. These spatial maps are compared with the spatial yield map shown in Fig.4a.

Figure 4: Spatial Maps of the Flame Seedless Vineyard: (a) Yield Map of the berry count across the field. Spatial Map of percentage of clusters for Grades B and D on the 18th June (b), 24th June (c).

For computing the color progression, we collect images of the same location in the vineyard on different dates. We then develop a linear regression model that estimates the rate of change of maturity by comparing the color information of the same berries in the images collected on different dates.  Figure 5. shows the average rate of maturity (or color development) in the Flame Seedless grape variety is consistent over all grades/hues of the fruit for the three growing seasons. Albeit our model does not include weather conditions and admittedly the past three seasons have been similar in weather patterns. In future years we will attempt to introduce weather modelling to correct maturity prediction in abnormal weather years.

Figure 5: Forecasting model of rate of color development. We have been building this model over the past three years. A) shows the rate of color change has been relatively consistent. B, C, D) show the raw data used to fit the color development model.

Terry Bates, Cain Hickey, Rhiann Jakubowski

For several years, we have been using ground based NDVI sensors pointed at the side wall of Concord canopies to estimate canopy growth across vineyard blocks.  Sensors are mounted on a farm vehicle and connected to a GPS antenna and field data logger and data are collected as the vehicle passes through the vineyard.  For our purpose, we are using the NDVI sensor in its most basic function of indicating if it senses the presence or absence of actively functioning leaves.  To try and capture the variation in canopy growth across a vineyard block, we mount the sensor to point at the actively growing region of the canopy; therefore, in Concord high cordon systems, the sensor is mounted high early in the season and is moved to lower positions as the season progresses.

Concord canopy development throughout the season.  Scanning is typically done with a single sensor which will be mounted around the cordon early in the season then moved lower as the canopy develops to capture regions of high and low vine growth.

Comparison of Concord vine size and NDVI signal.  If the NDVI sensor is mounted correctly to capture difference in canopy growth there tends to be a positive linear response between vine size and NDVI between 0 and 3 pounds of dormant cane pruning weight (8’ vine x 9’ row spacing).   In large canopies (above 3 pounds pruning weight) there is not a strong relationship between vine size and NDVI.  Under this standard vineyard configuration, vine canopies completely fill the allotted trellis space at around 3 pounds so it makes sense that the NDVI signal would also saturate at 3 pounds.  Our use of NDVI sensors essentially measures external canopy fill, not canopy density.

When is the best time of the season to collect NDVI information?

In 2016, we set out to answer the question of which NDVI scan best related to vine size and vineyard yield.  Eight commercial Concord vineyards were NDVI scanned at four time points; approximately two-weeks pre-bloom, bloom, two-weeks post-bloom, and veraison.  For validation, manual pruning weight sample locations were identified which targeted low, medium, and high NDVI values.  In addition, continuous yield monitor data (where available) were collected and compared to the continuous NDVI values.

The relationship between NDVI at different times of the season and vine pruning weight (left) or crop yield (right).  Manual pruning weights were collected at 30 locations across 5.9 acres and yield monitor data were collected continuously on 2.8 of the 5.9 acres.

In this Concord vineyard example, there was a positive correlation between NDVI and pruning weight at each scanning time.  The weakest relationship was in the pre-bloom scan and the strongest was at veraison; however, the last three scans (bloom, post-bloom, and veraison) were similar.  From a practical management perspective, a positive relationship at the bloom and post-bloom timings has the advantage of giving useful information in spatial vine growth for in-season management decisions, such as for differential floor management or crop thinning.

Similarly, there were positive relationships between NDVI at each scan time and harvest yield.  The pruning weight data indicate that the vines in this vineyard ranged from approximately 0.5 – 2.75 pounds/vine; therefore, we would predict a positive linear response between NDVI, pruning weight, and crop potential (i.e. the vines, in general, had not reached maximum potential light interception nor saturated the NDVI signal).  As with pruning weight, the strongest relationship between NDVI and yield was at the veraison NDVI scan time; however, the earlier scans also gave positive correlations and could be used for practical in-season management applications such as crop estimation or fruit thinning.

To see Dr. Terry Bates’ full Updated Project Overview click here.

By Kaan Kurtural, Luca Brillante, Johann Martinez, Runze Yu

University of California Davis

Spatial variability in wine grape vineyards is a major limiting factor in achieving maximum fruit and wine composition. Wine grape growers and winemakers have recognized this for decades and have devised a wide range of management practices to overcome this problem. Despite their best and ongoing efforts, variability still exists in most vineyards and continues to contribute to reduced economic returns to growers and wineries. Many crop production attributes contribute to this variability. Despite our understanding of this, it has been difficult to eliminate the effects of spatial variability on vineyard management. Soil nutrient status, water holding capacity, irrigation management, and climate are some of the major contributors to variations in vegetative growth, yield and fruit composition.

The lack of uniformity in grape yield and composition within vineyards may influence the management of grape delivery to the winery. Particularly, spatial variation in fruit composition can limit the opportunity to maximize wine composition either by unconscious mixing of low and high quality grapes or by losing the opportunity to ferment separately the existing high quality grapes.

The application of precision farming practices to viticulture is relatively recent, but has taken advantage of technologies applied in other crops such as yield monitoring.  The variability in yield revealed by this technology, although not unexpected, is still typically enlightening to growers and winemakers alike. In our recent work variability in both yield and fruit composition were reported in several vineyards over three years. The variability in yield showed some spatial consistency within a vineyard, but the range in yield was significantly different between years. We associated this temporal variability with temperatures during the bloom period. This tendency towards spatial consistency was encouraging with respect to the long term objective to understand the source of this variability and hence to develop more precise management practices to improve vineyard uniformity.

The spatial variability in fruit composition was not as consistent as yield and furthermore was not well correlated with spatial yield variability. This strongly indicated that the factors controlling fruit composition are more complex than yield and hence the latter cannot be used as criteria upon which to determine fruit composition management. The advent of variable fertilizer applications, foliar nutrient programs and drip irrigation all help to minimize variability in vine growth as well as fruit composition. But since there are other factors such as slope, aspect, pests and disease, influence of soil texture, canopy size, and sunlight, composition at harvest is still difficult to predict. To help overcome this problem, it has always been desirable to practice differential (both temporal and spatial) harvests. However, this is generally too expensive for most large scale operations. There is a need for development of on-the-go wine grape composition sensing technology, which would enable quality zone delineation for effective harvesting and management of vineyards.

In our work in 2016, we worked with a producing vineyard in Sonoma County. The vineyard was planted to Cabernet Sauvignon clone 7/110R on a high quadrilateral system. Based on anecdotal evidence the vineyard manager informed us that the fruit composition varied greatly at this site. We laid a spatially dense grid after sensing canopy reflectance and soil electrical resistivity (Figure 1).

Figure 1. The research site in Sonoma County

The workflow at this site is presented in Figure 2.

Figure 2. Workflow at the research site.

Our terrain analysis revealed that the topography of the site was quite varied. The absolute elevation of the site ranged from 64 m to 76 m. More interestingly, the slope of the site ranged from 2o to 6o, hence resulting in a catchment at the southern end of the vineyard. (Figures 3a, b and c).

Figure 3. Terrain analysis of research site

When plant water status was modeled by clustering analysis, the vineyard was delineated into two water stress zones: Higher and Lower (Figure 4).

Figure 4. Water stress clustering at the research site

This indicated that between the two water stress clusters there was 70% variability. The water stress was directly related to elevation change, but inversely related to total wetness index of the research site. The difference between the two water stress zones drove quite striking difference in primary metabolism such as net carbon exchange, Brix accumulation during the last 40 days of the ripening period (Figure 5), and titratable acidity of the berry.

Figure 5. Clustering of Brix accumulation at the research site.

As previously mentioned we were unable to determine a relationship between yield and the fruit composition values we monitored. Furthermore, we were unable to find a significant relationship between water stress and yield. However, the differences in primary metabolism drove differences in anthocyanin and proanthocyanidin content of the berry at harvest. We associated the differences in anthocyanin content to degradation due to greater water stress as delineated by our cluster analysis. Furthermore, the differences in proanthocyanidin content of the berry responded in similar manner to anthocyanins as modulated by the water stress model at this research site. (Figure 6).

Figure 6. Total anthocyanin content of the berry was affected by the water stress clustering at the research site.

In the initial year of the study we have shown that vineyard variability affected harvest composition. In cases such as this where variability at the research site is too large to coalesce, selective harvest can be a useful management tool. Water status allowed us to effectively discriminate between harvest zones. We can now easily model  and sense water stress deliniation using canopy reflectance with less of a need to take repeated measurements in vineyards.

By Claudio Piccinini and James Taylor

The “spatial data processing and decision support system” team are tasked within the program to develop a web-based Spatial Decision Support System (SDSS). But what is a SDSS and why is this important for precision viticulture?

Precision viticulture is based on the collection of raw spatial data, its transformation into spatial information and the generation of good viticulture decisions for spatial or site-specific management.  Site-specific management is not a new concept, but there is still a lack of adoption by commercial producers and one of the reasons for this is the absence of SDSSs. SDSSs are needed to provide users with the right decision-making environment that allows for the management of spatial data in a flexible manner. Decision makers with a complex spatial problem often have multiple conflicting objectives when determining a solution. They need a system that lets them define the problem and articulate the objectives for its solution.

In general, Decision Support Systems must support a decision research process, rather than a more narrowly defined decision making process by providing the decision maker with a flexible problem-solving environment. Such an environment empowers the decision maker in 2 ways:
– first the problem can be explored to increase the level of understanding and to refine the definition and, secondly, the generation and evaluation of alternative solutions enables the decision maker to investigate possible trade-offs between conflicting objectives and make a decision based on the grower’s attitude to risk.

A support system should facilitate the introduction of new factors into analyses and also the relative importance of factors in analyses, both to evaluate the sensitivity of solutions and to reflect different opinions and objectives for the solutions. Finally a system should be able to display the results of analyses in a variety of ways that help users to understand them.

For this project we decided to develop the SDSS for the US viticulture industry with open source Web technologies because they are free to use and they allow easy management of the diffusion of decision support tools, their update, and communication with centralized databases. The Figure below shows an overview of the SDSS under development. This is a prototype, therefore the software components may be replaced as the project proceeds.  Basically the prototype is modular and will contain a presentation layer, a business layer, a GIS layer, and a data/analysis layer as independent components.

  • The presentation layer will include the interactive web application that runs on the user device; both desktop and mobile devices will be supported.
  • The business layer management system will supervise the modelling and analytical capabilities of the system including the generation of web pages for the presentation layer.
  • The GIS layer will contain the tools necessary to distribute spatial data using industry standards; in particular the interaction with the data/analytical layer will employ “Web Processing Services” which define standard rules for exchanging spatial data between different software.
  • The data/analysis layer will contain the tools for storing, processing, and analysing the user’s spatial data. The main tool for this layer will be the spatial database which allows permanent storage and query of spatial data; however other tools may be employed to enhance the analyses of images and maps, the storage of large quantities of data, and the conversion of raw data into meaningful and useful information for spatial business analysis purposes.

2016 Efficient Vineyard Project Overview (click image below to open)

To see Dr. Terry Bates’ full Efficient Vineyard Project overview click here.

by Stephen Nuske, Ph.D.

Research article – Steve (adapted from “The Grapes of Nuske” by Olivia O’Connor, Carnegie Mellon Today January, 2014)

The USDA estimates that there are nearly 800,000 fruit-bearing grape acres in California, an area equivalent to some 605,000 football fields. This vast spread of land, encompassing table grapes, wine grapes, and raisin grapes, takes an equally vast upkeep effort. Grapes must be given the proper amount of water and fertilizer. They must be sprayed with insecticides to deter insects and fungicides to prevent disease. They must be harvested at the proper point in their growing cycle and transported at a temperature that disallows fermentation or freezing. A considerable amount of the work that goes into maintaining these conditions happens at night, when temperatures cool.

stephen-1-betterStephen Nuske, a “Systems Scientist” at Carnegie Mellon’s Robotics Institute is working on a robotic system that could revolutionize the routine in vineyards. As part of an USDA/NIFA Specialty Crop Research Initiative grant titled Efficient Vineyard, his team is working to improve both the quality and quantity of ripening grapes through something called High-Resolution Spatiotemporal Crop Load Measurement and Management.

In essence, Nuske is conducting a trial of an automated grape-counting system that will allow vineyard managers to track variations in their fields and more accurately estimate their harvest sizes. Ideally, the system will allow managers to make changes to their growing practices during the season to improve their overall yields as well as the quality of the fruit.

In 2012, he collaborated with Research Professor Sanjiv Singh on a project designed to test “Automated Crop Yield Estimation for Apple Orchards.” The program caught the attention of the National Grape and Wine Initiative, a coalition that coordinates grape research and growing practices across the United States. The NGWI reached out to Nuske to see whether the technology he was working on could be adapted to the grape industry, and Nuske jumped on board, partnering with Cornell researchers as well as his CMU colleagues to pioneer the new grape-counting technology.

Of course, “grape counting” is a simplification of what Nuske’s system actually does. The rig involves a camera (to take photos of the vines, from which a computer system will later detect and count the fruit), a laser scanner (which takes a three-dimensional measurements of vine foliage), a computer system (which tracks the data as it comes in), and the vehicle itself (which allows the team to quickly cover large areas of the vineyard). One person can operate the system in the field and then data is taken back to office to process and extract the vineyard performance metrics. Although if needed the system can be configured to output processed data live on the vehicle. The result is an odd-looking contraption, futuristic in function but humble in design. The rumbling vehicle rolls through the vineyard, sending out bright camera flashes at a rate of five per second.

When the predicted harvest yield (gathered by the grape-counting system) is compared to the actual harvest yield (determined months later), the overall error is less than 5%; in comparison, current industry standard predictions may be in error as high as 20-30%, says Nuske.

steve-image-3It’s an impressive improvement, and it’s not the only use of Nuske’s system. The most important aspect of the technology is its ability to show variation within the vineyards. He explains that viticulture (the study of grapes) is all about vine balance: the amount of leaf area as compared to the amount of fruit. The idea is summed up in a concept called “crop load,” which refers to both the amount of fruit and the health of the vine. Terry Bates, the director of Cornell University’s Lake Erie Research and Extension Lab and Effective Vineyard project director, compares crop load to the process of losing weight. A large number of leaves mean that a large amount of photosynthates are in the vine, available to the fruit. These photosynthates, Bates explains, are the “calories in.” And the grapes drawing the photosynthates from the vine are the “calories out.” In a healthy vine, the calories in and calories out are balanced. Too little or too much leaf area and the grapes suffer, leading to consequences that can go far beyond a bad glass of vino. According to the NGWI, grapes are the sixth-largest crop in the United States, supporting an industry that is valued at $4.9 billion and contributes $162 billion to the U.S. economy (and $33 billion in wages) each year. These numbers illustrate why the NGWI is concerned about maximizing crop yield.

Once all the data are collected, Nuske creates a spatial map of the parts of the vineyard through which his equipment has travelled. The spatial maps are constructed with a combination of each area’s crop load number, determined by the mathematical relationship between the amount of fruit and the amount of leaf area. The maps show which parts of the vineyards are doing well, and which aren’t. Growers can then adjust their methods to improve the balance between each area’s foliage and fruit. “This is kind of the cutting edge in viticulture,” Nuske explains. “A hundred years ago, if you wanted to grow a really good bottle of wine … you’d want to actually reduce the amount of fruit.” So farmers would remove grapes from the vine. But, in fact, Nuske explains, the past 20 years have shown that “you can produce a really good-quality crop with a lot of fruit with really big plants. … It’s not the old case of less gives you better quality.” As long as leaf and fruit weight are balanced, a large crop can be a quality crop.

This is good news for growers, Bates explains, because competition from domestic and foreign markets means that the price paid for grapes has not increased very much during the past 40 years. The result, he says, is that growers have been forced to become more efficient to offset rising expenses such as fuel and labor over the years, so they can still make a profit. And Nuske is putting the tools for greater efficiency in the palms of their hands, literally. His spatial maps can be downloaded to smartphones, so that growers can walk through their fields along the course of the maps, look at the high- and low-yield areas in person, and determine how to improve the crop load: water here, prune here, fertilize here. In an industry that is endlessly variable, the information that Nuske’s system offers insight that enables growers to adapt with the season, rather than rely on the conditions of previous years.

Multiple fields intersect to make Nuske’s work possible: biology, computer programming, engineering, robotics, agriculture. The system is complex, and Nuske acknowledges that the process of research and development can be tiring, even tedious.

“I think what we’re doing is probably not anywhere near as tough as what they’re doing,” he says referring to the vineyard laborers. Yet, despite the difficulty of farm labor, some critics question the future of agricultural technology, worrying that the human component of farming will be phased out, that jobs will suffer. But Nuske points out that his system isn’t taking over anyone’s livelihood. There aren’t any workers counting grapes full-time. But he’s still conscious of the ethical implications of precision agriculture. In general, he believes it will do far more good than harm.

by Cain Hickey, Ph.D. and Terry Bates, Ph.D.

The Concord marketplace has not changed much in the last couple decades – profits are attenuated by stable crop prices as input costs continue to increase.  The Efficient Vineyard project seeks to increase profitability by way of reducing input costs, and improve ripening and crop production uniformity across vineyards.

The focus of the Efficient Vineyard project is to develop and evaluate geospatially referenced management technologies to improve vineyard management efficiency, focusing on variation throughout vineyards.  The project started by adapting off-the-shelf technologies, originally used in row crop production systems, to measure and map variations in vineyard canopy; these sensors are called normalized difference vegetative index (NDVI) sensors and low NDVI equates to lower canopy size (see Fig. 1a).  These measurements are then coupled with maps of soil electrical conductivity/magnetic susceptibility (See Fig. 1b).  Together, these maps provide researchers and growers information on how vine growth patterns change within a  vineyard, and if this  variation in vine growth is due to inherent factors (such as changes in soil type) as opposed to management factors (i.e. differential pruning).

In the very simplest sense, these maps could direct growers to “problem areas” of the vineyard (see red areas in Fig. 1a), perhaps to focus management in these areas for tasks such as vine renewal or more intensive nutrient or pruning management.  Larger problem areas mean less actual production acreage, which means less income for the grower.  The sooner the grower acts to manage these areas, the sooner they will see an increase in net returns.

Figure 1. NDVI map (a, left) and shallow soil EC map of a Concord vineyard in the Lake Erie region (b, right).

The focus of the project has switched from canopy and soil sensing, to a more proactive approach of implementing variable rate management strategies to manage against inherent vineyard variation.  Small pilot projects were initiated this summer to test the efficacy of GPS-driven variable rate crop load management via crop and shoot thinning.  An NDVI map was used to make management classification zones based on vine size (see Fig. 2a for an example of three NDVI-based vine-size management classification zones).

In short, computer programs were used to geospatially change the rate at which crop and shoots were mechanically thinned throughout the management classification zones.  Without any manual adjustment by the operator, relatively more crop or shoots were thinned as the tractor was driven into small-vine-regions, and vice-versa for large-vine regions.  Our goal is to im­prove ripening uniformity and increase vine capacity by reducing crop load in parts of the vineyard with small vines.  These vineyards have been monitored for differences in canopy development, and will continue to be evaluated for fruit maturation rate and dormant season cane pruning weight.

Recently, the Cornell Lake Erie Research Extension Laboratory (CLEREL) research team has been out taking weekly fruit maturity samples from vineyards that have been scanned with NDVI-canopy and soil sensors throughout the growing season.  The eventual goal is to evaluate if and why fruit maturation rate (and, eventually, crop yield and pruning weight) is different between sensor-derived management zones (See Fig. 2a).

For now, we are interested in characterizing fruit maturity. Fruit maturity (juice soluble solids) samples taken in several vineyards within a 50-mile stretch of the Lake Erie Concord region ranged 11.0 to 17.9 °Brix last week.  The 6.9 °Brix range was likely a function of several factors, broadly encompassed by differences in vineyard site and management practice.  However, our project is interested in characterizing and managing the variation within a vineyard.

Using the management zone classification map in Fig. 2 as an example, soluble solids concentration ranged 11.0 to 14.7 °Brix across the vineyard, and the mean value was 12.3 in the purple zone, 12.7 in the green zone, and 13.2 in the red zone.  Berry weight ranged 1.27 to 2.85 grams, and the mean value was 2.2 in the purple zone, 2.5 in the green zone, and 1.5 in the red zone.  Thus, the small-canopy vines in the red zone also have smaller berries, which may be partially responsible for the relatively greater juice soluble solids concentration in this compared to the two other zones.  The red zone is likely under some soil water or nutrient stress, potentially due to physical limitation of root growth.

Since this vineyard is still in “diagnostic phase”, the take home for now is that vine size, berry weight, and fruit maturity are different across NDVI and soil sensor-defined zones; crop yield and pruning weight will likely also be different.  This means sensors are effective at characterizing vineyard variation.  The next step is to work with the grower and develop a variable rate management plan to to either improve capacity and yield, and/or rate and uniformity of fruit maturation.  Plans will begin to take shape over fall and winter, and be put into action by spring.

Figure 2. An NDVI-derived management zone classification map (with data plots for fruit maturity sampling, in black) of a Concord vineyard in the Lake Erie region (a, left), and a CLEREL post-doctoral research associate sampling berries (b, right).

by Terry Bates, Ph.D.

As part of the Specialty Crop Research Initiative project on vineyard spatial crop load management, we have been investigating ways to use sensors and spatial data to (a) improve the accuracy of mid-season crop estimates and to (b) test the ability to perform variable rate fruit thinning in NY Concord vineyards.  Last week, the CLEREL team performed spatial crop estimates in seven different commercial vineyards.  The following update summarizes a portion of the research done on crop estimation and variable rate thinning in cooperation with the Betts family in Westfield, NY.

Crop Estimation

Characterizing spatial vineyard variation:  The first step in improving crop estimation through directed vineyard sampling is to understand the spatial growth patterns within vineyard blocks.  This project uses mobile soil (DualEM) and canopy (CropCircle – NDVI) sensors to measure and map both soil and vine growth patterns (left).  The objective is to identify healthy regions of the vineyard with higher production potential as well as identify regions of weak vine growth that may need additional management for improvement.  For example, the research shows that early- and mid-season canopy sensor data relates to harvest yield potential (right).  Similar to pruning weight measurements, vines with a low NDVI sensor readings because of lower canopy growth and, therefore, lower sunlight interception have lower fruit production potential.  In contrast, big vines with large canopies and full light interception have higher yield potential.

Stratified Sampling:  In contrast to picking random sample locations across a vineyard block, directed or stratified samples may be more accurate by taking into account known sources of variation identified by the sensor data.  The continuous spatial sensor maps are simplified into management classification maps with 2, 3, or 4 classifications that make sense to the vineyard manager.  In our crop estimation example, the vineyard was broken into classifications of low, medium, and high NDVI (and potentially low, medium, and high yield).  Crop estimation sample locations were generated so that the low, medium, and high vineyard regions were all sampled.  In each sample location, vines were cleaned picked with a harvester and the fruit was weighed.  In this case, the sample size was one percent of an acre identified with a rope on the ground (left).  The harvester is also equip with a field computer and grape yield monitor (right).  The field computer runs precision agriculture software (AgLeader, SMS) which shows the spatial management classification map (generated by Rhiann Jakubowski at CLEREL) and the location of the harvester in the field.  The yield monitor records the weight of fruit as it goes over the cross conveyor belt.

Recording Fruit Weight:  The stratified sample locations were clean picked with a grape harvester and the green fruit was discharged into a bucket on a platform scale and weighed by Dawn Betts (left).  At the same time, the grape yield monitor recorded the fruit weight as it was discharged over the cross conveyor belt.  The yield monitor data was then compared to the actual scale weights to test the performance of the yield monitor (inset).  For each management classification, sample fruit weights were multiplied by a berry weight factor to give the predicted harvest yield.  In this case, crop estimation was done at 30 days after bloom with the assumption that the berries were 50% of the final berry weight; therefore, sample weights were multiplied by 2 to give the predicted harvest yield.  For the whole field estimate, a weighted estimate was calculated by multiplying the management classification yield estimate by the size of each management classification.

Crop Estimation

Variable Rate Crop Adjustment:  Once an accurate crop estimate is calculated, the fruit thinning or crop adjustment procedure starts with the vineyard manager making an educated decision on if the crop needs to be reduced in any particular management classification and how much the crop should be reduced within that management classification.  A manager may decide that no fruit thinning is needed, or only that the weak vines need to be thinned, or that all the zones need to be thinned at different rates.  In this case, Thom Betts set up his harvester so that the shaker rod RPMs could be adjusted through the harvester hydraulic system and controlled by the AgLeader software.  With a little testing in each management classification, shaker speeds and ground speeds were determined to reach the desired yield levels in each vineyard area.  These values were entered into the field computer software.  The harvester was then driven over the whole field while the shaker speed and thinning rate were adjusted on-the-fly by the field computer.  Check plots where no fruit thinning was done (seen as the small light blue boxes in the left image) were incorporated into the prescription map.  At harvest, the grape yield monitor data will be used to measure the effect of this variable rate thinning trial.  This work on crop estimation and variable rate fruit thinning is just a portion of the larger specialty crop research initiative project on spatial crop load management.  The project is national in scope with research in juice, wine, and table grape industries and multi-disciplinary with project leaders in engineering, viticulture, precision agriculture, economics, and extension.  The project is also made possible by the innovation and effort of industry cooperators, such as the Betts, in directing and integrating research into real-world practice.