“Wine is the only artwork you can drink”, as the famous quote goes, but where’s the border between art and science? As AgriTech (Agriculture Technology) is developing it is starting to encroach on this ancient art form. From the darkest reds of Bordeaux to the most crisp whites of Sancerre (and maybe even the Rose’s in the middle) vine yards are starting to explore the use of technology to help them get more consistent, higher, and healthier yields in order to maximise their efficiency and ultimately profits. A single bad year for a wine maker can put huge pressure on finances and even with high yields supermarkets demand low prices and consistent taste each and every year (P.S. don’t buy wine from supermarkets if you can help it) so getting this is really key for many of the larger grape growers.
As with pretty much every industry there’s some key developing technologies that are really starting to provide benefit. For wine making there’s two emerging as potential game changers: Drones & Satellites and Artificial Intelligence.
Drones & Satellites
I put Drones and Satellites under the same category as they’re both providing similar outcomes - high quality imagery across a broad range of the EM Spectrum, although there is recognition that they of course do this in different ways with their own benefits and disadvantages.
With the average vineyard in Europe measuring 1.3ha (13,000m2), and over 10ha when considering France on it’s own, knowing what’s going on in every corner can be extremely challenging and of course across that area conditions can vary greatly due to geography, soil, rainfall, sunshine, temperature, disease and so on.
This is where drones and satellites come in. Using these enablers with multi-spectral and other sensors allows growers to get measurements that are often beyond easy reach using traditional methods. For example, measuring soil moisture at varying depths across a whole vineyard, measuring temperature and humidity, measuring density patterns (typically referred to as the Normalised Difference Vegetation Index (NDVI)). While these have generally always been possible using human labour and some tools they are clearly time intensive tasks and still may not be at the resolution that’s really needed.
By making use of drones and satellites these measurements can be taken very accurately at potentially cm resolutions. This makes the spotting of patterns so much easier and therefore gaining actionable insight also so much easier. Where previously a grower would have to walk the vines and spot problems such as disease with their eyes they can now see problems perhaps before they’re visual and can see them on a smaller scale. This then means actions such as the use of fertilisers and pesticides can be used preventatively and on a much smaller scale, perhaps on just one vine rather than a whole area - reducing costs and environmental impact.
By measuring the vineyards NDVI more accurately and to a greater resolution growers can also harvest grapes at varying times to ensure that they have the best uniformity. Where typically a vineyard would pick everything at the same time, or perhaps split it in to large chunks, they can now see exactly which vines are ready for picking and which need more time. This allows wine makers to get grapes that are all at relatively the same maturity.
While measuring NDVI using satellites isn’t particularly new we are now seeing the use of higher resolution and cheaper satellites such as Cube Sats which is opening up the technique to ever more growers. The use of drones can be even more approachable, particularly for smaller vineyards where satellites don’t make economic sense.
Looking to the future we’re probably going to see the use satellites more and more in this area as they become much more economical and their capabilities continue to increase. We’re probably going to find that companies offering this ‘as a service’ will be the most successful with growers wanting insights and actions rather than just raw data. On the drone front we’ll probably see similar development. We’ll likely see the use of drone swarms so that more area can be covered quickly. We may also see the use of drones for spraying but realistically this needs significant improvements in battery technology first - 10-15 minute flight times at the moment make it not very viable unless you’re using a lot of drones at once.
Artificial Intelligence
Now I use the term AI in the loosest possible definition to basically mean data analysis with a bit of prediction. As grape growers make more use of technology the amount of data they collect and hold only increases. This data though can be hugely valuable when it’s used in Machine Learning algorithms, either as training data if historic or as data to form the basis of predictions if current.
Making use of the data collected from drones and satellites as discussed above is proving hugely valuable.
Predicting the yield of a vineyard each year is extremely important for growers and for the makers who rely on the grapes. Knowing not only the amount of grapes but also the quality can really help with predicting financials for the year and bad predictions can wipe out multiple years of profit as growers overstock on resources such as fertiliser and human capital or the other way not having enough barrels ready or people available to pick.
Unfortunately predicting yields is probably one of the hardest things to do in agriculture (and we must not forget wine making is a part of this industry). There are so many variables such as weather which in itself includes rainfall, temperature, humidity, wind that it can be difficult to predict early in the season what the yield will be at the end. Currently growers take samples from across the vineyard and assess measures such as weight, number of grapes per bunch and canopy growth, then extrapolating this across the whole vineyard to predict themselves what the yield may be, based on their years of experience. This gives a roughly 60-70% accuracy (surprisingly high I thought).
By taking historic measurements such as weather, soil data and yields, which most vineyards will have, they can use machine learning algorithms and data analysis to better predict yields from early measurements of the current season, based on what’s happened in previous years. By also making use of better long term weather/climate forecasts it’s been shown that you can accurately predict yields for the whole season with up to 90% accuracy.
As more and more data is collected, stored and analysed these algorithms will become ever more accurate and useful. Looking further towards the future we’ll see AI moving in to the production facilities with makers being better supported by data in their decision making as we learn more about what properties make wine taste good.
While many believe wine making is an art form, it is in fact much more like science.
—
Sources:
https://pursuit.unimelb.edu.au/articles/five-ways-technology-is-changing-the-wine-we-drink
https://thegrapevinemagazine.net/article/drones-in-the-vineyard-uses-benefits-concerns-key-players/