Google Play icon

Scientists employed machine learning to predict smoke contamination in vineyards

Posted September 2, 2019

Winemaking is a very technical craft, heavily dependant on nature. You can be the best winemaker in the world and you will still be hopeless if the conditions of the vineyard are less than ideal. Now scientists from the universities of Melbourne and Adelaide have started testing a non-invasive model for detecting smoke contamination in grapevines.

Smoke contamination can alter the taste of grapes and consequently the wine, making it lesser quality. Image credit: Amanda Slater via Wikimedia (CC BY-SA 2.0)

The problem in this case are bushfires. In the light of climate change they are becoming more and more common. Vineyards can be safely away from the fire, but smoke might still reach it. This may result in a smoke taint in the wine, which is, of course, not ideal. In fact, even a minute taste of smoke can reduce the grade of wine and make it less valuable on the market. It may sound like a trivial issue, but that kind of detail is what separates the best vineyards from the mediocre ones.

Scientists now used machine learning algorithms and infrared and near-infrared thermography data to predict whether the grapes were contaminated with smoke. The initial trials revealed that these predictions are 96 % accurate. The best thing is that this method if non-invasive – grapes don’t have to be picked and destroyed to employ this method. This is also quicker, which is important in these situations.

As climate change progresses, Australia is experiencing more and more bushfires. However, smoke contamination in vineyards is not just Australian problem – Greece, Chile, parts of California and South Africa are also suffering from the same. In fact, there are probably no vineyards in the world where smoke from wildfires couldn’t do some damage. This algorithm could help mitigating the damage.

Smoke forms carbonic acid, which reduces the pH in the stomata (the pores in the outside of leaves), forcing them to close, which reduces the yield. Then the grapes themselves absorb the smoke-derived volatile phenols, which can be extracted, but reduce the quality of the wine significantly. Professor Sigfredo Fuentes, lead author of the study, said: “It is important to be able to detect smoke contamination because the removal of taint from wine isn’t selective and can inadvertently remove important compounds, affecting the taste and sensory experience of the wine”.

Australian wine is one of the best in the world and it is quickly gaining the recognition globally. However, changing natural conditions are threatening its quality. By having means to avoid at least part of the damage, Australian viticulture could continue to thrive and grow. Hopefully, this algorithm and near-infrared thermography data will help avoiding that faint taint of smoke in the wine after bushfires.


Source: University of Adelaide

Featured news from related categories:

Technology Org App
Google Play icon
84,875 science & technology articles