Deep Planet – VineSignal: an innovative vineyard decision support platform

Executive Summary

Deep Planet’s goal is to harness the power of AI to help improve vineyard management efficiency and sustainability while improving productivity. To achieve this, we have developed VineSignal, an innovative vineyard decision support platform offering cost efficient and scalable precision agriculture tools to winemakers, growers and regional associations.

VineSignal is currently monitoring over 80,000 hectares under vine and providing regular insights to over 200 users. Deep Planet has commercial and R&D partnerships with over 60 producers, researchers and industry bodies, helping improve vineyard and soil health, harvest logistics and provide vineyard monitoring at scale.

SDG Sustainable Development Goal

Climate Action

Reason Why/Motivation/Background

Deep Planet was born out of the founders’ vision to leverage technology to improve agricultural productivity and sustainability. Wine is often referred to as “the canary in the coalmine for climate change”. This is because wine grapes are extremely sensitive to weather and environmentalconditions. In fact, many of the world’s vineyards are at risk of continual decline in yields as climate change brings less favourable conditions. Further, there isexpected to be 75% shrink in production area by 2050. Temperatures have risen so much that harvests are now beginning on average 2 weeks earlier than theydid prior to 1988. Deep Planet has its roots in sustainability with a goal to harness AI for sustainable precision viticulture. The company uses satellite and machine learningtechnology to help growers to improve vineyard management decision-making and to reduce their impact on the environment while improving productionoutcomes.

Project Description

Deep Planet’s VineSignal product helps growers, winemakers and industry associations to improve vineyard management efficiency and sustainability in the faceof climate change. It achieves this by providing accurate and actionable insights on:

Soil Health and Input Management

• Remote detection and monitoring of soil organic carbon and nutrients (NPK) allows growers to visualise the variability in SOC and nutrients available in the soiland to track changes over time. This helps target compost and fertiliser applications, optimise inputs, cut down on costs and improve production outcomes.
• Soil moisture interpolation and prediction using satellite imagery and ground sensors and evapotranspiration monitoring inform decisions on irrigationscheduling and help improve water management and reduce water waste.

Vineyard Health

• Vine health monitoring using satellite imagery at 10m, 3m and 0.5m resolutions and dynamic insights into weekly and monthly changes in vine health allowgrowers to monitor their vineyards remotely, streamlining the time and effort of vineyard management.
• Automated anomaly detection alerts the vineyard management team of potential problems in the vineyard, including frost, disease, hail, heat damage andirrigation leaks. This tool helps growers identify and address problems faster than by walking/driving through the vineyards.

Harvest Logistics and Grape Quality

• Accurate harvest date predictions per block and variety using historical and current season data help winemaking and vineyard teams to coordinate harvestlogistics more efficiently, save €2-€20 per bottle in reworking costs due to consistent sugar levels and reduce team stress around harvest.
• Maturity and quality variability maps visualise the variation in sugar levels across the vineyard and in individual blocks, allowing better identification ofsampling locations and selective harvesting opportunities and reducing costs of labour and machinery planning
• Yield prediction using satellite imagery and historical data; and
• Yield variability maps allow growers to improve winery intake and supply planning and to identify areas of the vineyard with poorer productivity.

Large Scale/Regional Monitoring

• Vineyard detection and mapping shows regional associations and cooperatives where their member vineyards are located for easier certification and auditing.It also makes it easier to promote technology uptake by individual members by having maps readily available.
• Regional scans detect damage to vineyards from frost, disease, heat, hail, flood, irrigation leaks and other environmental factors. This enables regional bodiesto provide quick assessment and high-level information to their members.
• Together these tools provide a holistic approach to getting data insights for more informed decision-making in the vineyard.

Benefits/Achievements

VineSignal is currently helping more than 200 growers and winemakers in Australia, New Zealand, France, Italy, Portugal, UK and the US to support theirvineyard management decisions. It has enabled savings of €45 per tonne of grapes and helped upgrade wine quality by €2 to €20 per bottle. VineSignal hashelped growers identify problem areas in vineyards, reduce input application, identify irrigation leaks, improve sampling and harvest decisions, identify areas ofremediation and manage variability, supporting more efficient and sustainable vineyard management practices.

Deep Planet has also conducted vineyard scans for regional bodies such as Wine Australia and Riverland Wine, providing vineyard monitoring at scale andenabling faster and easier technology uptake by individual growers.

Lessons Learned

Precision Viticulture with AI can be impactful both financially and environmentally for producers. However, the rate of technology adoption is slow. Largeproducers are more open to experiment with new digital technologies. However, for the industry to be sustainable and continue to thrive, it’s critical for smallervineyards and growers to adopt technology. This is why the involvement of regional associations and cooperatives is key in making technology accessible tosmall growers and to increase adoption in collaboration with industry suppliers.

The use of precision viticulture and AI-based tools in agriculture and the wine industry is relatively new. As a result, many producers do not have the kind of datanecessary to take advantage of the full potential of AI. This is why we have developed certain tools which allow growers to start getting value from ourtechnology immediately. Concurrently, we work with them to collect data in the right amount and quality so that they can get more advanced data insights andpredictions in the future.

Next Steps

Deep Planet continues to increase its product feature set, improving commercially available features on VineSignal and moving new features through its R&Dpipeline. As the user base of VineSignal increases, the overall accuracy of predictions also improves. In the future, in addition to predictive insights, we will alsoprovide prescriptive information to make it easier for growers to incorporate insights from data into their vineyard management practices.

Challenges

1. Access to quality data: Deep Planet has data from 80,000 ha of vineyards to tailor its solutions to the wine industry. However, to continue improving anddeveloping more tools, more quality data is necessary. We collaborate with producers, research institutions and regional bodies who provide us with the kind ofdata which allows us to develop reliable solutions. Increasing the number of these partnerships will increase the speed of new tool development and theaccuracy of existing tools.
2. Another challenge common to other precision viticulture solutions is scepticism from the industry. We overcome this challenge by showcasing and validatingthe technology in trial vineyards, by conducting joint projects with research institutions and by running pilots with larger producers who are more experimental.

Source of Information/Support 

Deep Planet is supported by the European Space Agency, Copernicus Incubation, H2020 and Innovate UK. To understand industry needs and region-specificchallenges, we consult and follow the representatives of the industry, including winemakers, viticulturists, regional associations and non-profit organizations like the Porto Protocol.