Crop load, the ratio of vine size to mass of fruit harvested, is fundamental to viticulture. Measuring vine size and crop yield, the components of crop load, has historically been a labour intensive exercise that has limited the use of crop load information to improve management in vineyards. Recent advances in assessing vine vigour, size and yield using geo‐referenced sensors are starting to make high resolution crop load mapping possible. In this paper, the concept of crop load is revisited with an emphasis on how vine size and yield can be mapped in vineyards. Existing literature is reviewed on how vine size and yield vary spatially and temporally within vineyard blocks and the inference this has on the spatio‐temporal variability of crop load. An example of crop load mapping using sensor technology is presented to illustrate recent advances in sensor technology in viticulture. Finally, some emerging technology and knowledge gaps for implementing spatial crop load information into vineyard management are discussed.

https://www.eprints.ncl.ac.uk/file_store/production/251759/F2C3D68B-FE25-4770-A7E7-C278CD0C3536.pdf

Considerations on spatial crop load mapping

Phenotyping in grapevines is the assessment of qualitative and quantitative traits including growth, development, tolerance, resistance, architecture, physiology, chemistry, ecology, and yield. Traditionally, phenotyping techniques relied on measurement of visual, chemical, physiological, or other characteristics by experts, often at low-throughput. The use of standardized OIV or phenological descriptors and scales to phenotype grapevine traits has provided a good foundation for international adoption of phenotyping standards and cross-comparison of results. However, many of these descriptors are subjective, fail to capture complete trait variation, or may not be relevant to some studies. Phenomics, the future of phenotyping, brings opportunities and challenges in increased throughput, objectivity, precision, dynamic measures, and integration that demand new approaches for standardization, data management, and analysis. Here, with a focus on large-scale genetic studies, such as QTL mapping, we describe current phenotyping approaches and their limitations and introduce some future opportunities in phenomics, including the promotion of FAIR data principles of Findability, Accessibility, Interoperability, and Reusability.

From Phenotyping to Phenomics: Present and Future Approaches in Grape Trait Analysis to Inform Grape Gene Function

Canopy sensing in viticulture is widely associated with the term NDVI (normalized difference vegetation index). However, there are many other vegetative indices (VIs) that can be calculated from information captured with visible/near-infrared (NIR) sensors. A proximal canopy sensor was used to survey 27 vineyards in the Lake Erie Concord belt and stratified to collect pruning weights (PW) at a density of ~25 samples per vineyard. Seven VIs were derived from the sensor data and the first principal component (PC1) extracted from a principal components analysis of the seven VIs. The VIs and PC1 were regressed against the local PW measurements and ranked in terms of their goodness-of-fit. Over the 27 vineyards, there was no single VI that outperformed the others, although VIs that used the red-edge band had a slight advantage over VIs using the red band. It is therefore recommended to use the normalized difference red edge index (NDRE) in place of the NDVI when predicting PW from terrestrial-based proximal canopy surveys. The PC1 derived from the decomposition of all seven VIs did appear to convey some benefit to PW prediction compared with a single VI approach, particularly with just NDVI. More research into the potential for multivariate approaches is recommended.

Comparison of Different Vegetative Indices for Calibrating Proximal Canopy Sensors to Grapevine Pruning Weight

In this memoire I will present my journey and the knowledge that I have gained within the Precision Agriculture domain over the past two decades. The first section broadly introduces variation in cropping systems and serves as a very general introduction to the research domain. 
Section 2 provides a more detailed overview of PA and develops some of the main issues and challenges that have been historically faced by PA practitioners. This includes clearly defining the domain and a recognition of the diversity and multi-disciplinary nature of PA. The role and importance of variability in production systems is then introduced along with the challenges in correctly and properly defining production variance. The final three parts of Section 2 turn the focus to how information, particularly multiple layers and multiple types of information, start to be translated into decisions. This shifts PA from a technical question on measuring and managing variation to a socio-technical innovation question on how technology is perceived and adopted and the potential disconnect between what scientists consider important and what commercial users want.
Section 3 further develops two key areas – the disconnect between academic research and commercial application (innovation) and the disconnect in the data that we have and what we want (in terms of data type and the quality). There is a deliberate focus on how PA technologies and methodologies are effectively translated into successful commercial services and what I have learnt about this from my various placements and from my attempts to define industry-facing protocols at various stages of my research career. 
Section 4 provides an overview of how my research activities and publications that have (I hope) helped to address the issues raised in Sections 2 and 3. A focus is given to a large body of work that provides a descriptive reference base to observed spatial variation in yield and crop quality parameters in various annual and perennial systems. This is followed by the role and the evolution in zonage approaches over the past 2 decades and how these zones or management units have been used as a basis for more advantaged analysis to improve crop production knowledge and ultimately crop management decisions. The latter parts of Section 4 shift from natural, agronomic applications to more recent research that has started to interrogate and to build models to understand socio-technical aspects of PA. It develops ideas on how PA translation and adoption can be enhanced by better considering barriers and drivers of adoption.
The last section, Section 5, outlines my own vision for the evolution of Precision Agriculture over the next decade. It outlines how Precision Agriculture needs to develop so that producers have a more ‘personalised’ agricultural decision system. It highlights key areas of research that I would like to pursue in my future career, and how advances in digital technology will help to achieve this research and translation. There are some concluding remarks.
Throughout the memoire, I have deliberately avoided the incorporation of mathematical notation. I have preferred to keep the discussion more general and not too specific for a broader audience. The cited references contain more specific details for those wishing to understand the derivation of the geostatistical approaches and models discussed. By convention, I have indicated in bold the references that are self-citations within the document. These are all listed in the Bibliography. A more detailed publication list is appended with my full publication list. The publication list contains work that is not cited in the document, particularly research that I have performed in soil science, rather than in PA, and teaching and industry-oriented publications.

https://hal.archives-ouvertes.fr/tel-02548498/document

Towards Incorporating Within-Field Variation into Spatial Agronomic Decision Processes

Combining Spatial Data Layers using Fuzzy Inference Systems: Application to an Agronomic Case Study.

Summary Goals: This report aims to present a clear protocol for (a) deploying proximal canopy sensors into single high-wire trellis Concord ( Vitis labruscana cv. Bailey) vineyards and (b) converting the canopy sensor response into an indication of vine size (pruning weight). The protocol is designed to be robust and practical for easy adoption in commercial systems. Evidence will be presented of the efficacy of vine size prediction using the protocol in multiple research and commercial vineyards. Key Findings: Using different vineyards and pruning crews, the protocol performed well in >80% of vineyards and permitted growers to generate maps of actual vine size within vineyards. These maps provide a valuable indication of current site-specific production potential and a baseline to assess changes in vine size over time. In a few vineyards, the proposed simplified calibration process did not generate a clear relationship between canopy response and vine size, which may be due to changes in vine shape in highly mechanized systems. Impact and Significance: Managing vine size is critical to the long-term sustainability of cool climate viticulture and to managing quality in all viticulture systems. However, convincing growers to routinely measure vine size for more effective management has been difficult historically because of the time involved and the difficulty of translating the data into a decision process. The proposed protocol uses technology and targeted sampling to minimize the effort required, and presents more coherent information that allows fast grower responses. Grower adoption of this protocol should promote continual vine size measurement, with the goal of decreasing vine size variation within vineyards.

/pdf/a-protocol-to-map-vine-size-in-commercial-single-high-wire-4ebglz3m4j.pdf

Kelly Link

Papers

A Protocol to Map Vine Size in Commercial Single High-Wire Trellis Vineyards Using “Off-the-Shelf” Proximal Canopy-Sensing Systems