Home Sustainability

How can climate models and forecasting help winemakers beat climate change?

Albert Soret Miravet and Marta Terrado from the Barcelona Supercomputing Center recently discussed with me, how computer-based climate and weather modelling and predictions can help agriculture tackle volatile weather patterns and manage against climate change.

The audio interview with them is below, and there’s a short video on what they are doing for grape growers in Southern Europe at the moment. Here’s a fact sheet about their work.

Underneath the audio, and the video, there’s a transcript of the interview if you prefer that format. If you can’t be bother with either, or both, here’s some quotes from the interview that I think are pretty interesting:

“Our objective is to demonstrate that climate predictions are useful. We don’t want to provide only the forecast. Our objective is to do an assessment, taking into account the specific needs of the users, and then also to provide the forecast”

“At the stage we are now, what we are doing with the users for example, in the wine sector, is to work with data from the past. We ask them to identify some particular years in which the yield was very good or very bad. Then we show them the prediction that would have been available at that moment, and we ask them, “which decision would you have taken if you had this information before?”

“….They (grape growers) are also very interested in knowing the number of pest treatments that they should apply, because also precipitation is changing, and pests are very related to precipitation and temperature. Then this is something that climate predictions that we provide can advise you about”

“The users in the wine sectors have told us for example, “I need to know when I need to prune my plants.” or, “I need to know where I need to start the pest treatment that I was mentioning before, or also when they need to start fertilizing or, when do I need to irrigate?”

Here’s the interview in audio. If getting this via email here’s the link.

VISCA PROJECT from UT-SEMIDE on Vimeo.

Here’s a transcript of the interview:

Toby: Welcome to another podcast with me Toby Webb, and I am delighted that joining me in today’s podcast is Marta Terrado and Albert Soret, who work for the Barcelona Supercomputing Center. Good morning to you both, how are you today?

Marta: Good morning, we are very well, thanks.

Toby: Thanks for making the time to speak to me on the podcast. You were introduced by a friend of mine, he told me that you guys are doing some really interesting work around climate modelling and agriculture, particularly with regard to wine. Perhaps you could tell us a bit more about the work you are doing.

Albert: The Barcelona Supercomputing Center there is a science department and we have about 96 people working in air quality and climate. We have four different groups.

In our group our main activity is to do applied research. Taking into account the needs of the society, of private companies, of administrations, we try to provide useful information taking into account the knowledge that we have in our department.

For instance, for the air quality we work with the Barcelona City Hall, and we help them to define their air quality plans, for climate we work with energy companies, and agriculture, we help them to provide forecast for the forthcoming weeks months and years.

Toby: What are your forecasts showing you at the moment that’s relevant for agriculture and for the wine sector?

Albert: We cover different forecast windows in agriculture and we address them using different models that combine different forecast windows. For instance first, when we talked about seasonal predictions we provide information for the forthcoming months, and it’s really useful for them in April to start defining the whole growing season to know if in summer there will be a heat wave, or to whether plan different task that goes from April to October.

Toby: How far ahead do you feel that you can credibly forecast the weather? I know there’s been a lot of debate about this in the past, and of course we also know that climate change is leading to significant unexpected weather volatility, which is causing all sorts of problems in agriculture. I’m just wondering, how far do you guys feel you can forecast ahead of time?

Albert: Actually now, the way to know what’s going to happen in the forthcoming month, it was to assume that the past is going to represent the future, we call it a climatology approach. To analyze the last 5 years or 50 years, assemble all the data, and then to say, “More or less this are the rains that are going to happen this year.” It includes some shortcomings. The first one is the length of the sample. It’s important to include more years. The second is that every year it is different, and now we also have climate change that implies climate trend but also climate variety, every year we have extremes and these extremes are even more extreme. 

Our objective is to demonstrate that climate predictions are useful. We don’t want to provide only the forecast. Our objective is to do an assessment, taking into account the specific needs of the users, and then also to provide the forecast. We see it as a way of developing the science, because now it’s the point that this technology seems to be quite mature and it’s important to take all this information and provide this information to the users, to be used.

Marta: Regarding the time scales, because you asked how far away you are able to predict, and I think here it’s important to make a differentiation between what we call very short-term, which is a prediction for the next following days, then what we call medium-term, which we refer to predictions that are for the next weeks and months and maybe years ahead. And also the long-term predictions which is the weather that we are going to have at the end of the century, for example.

All these different time scales can be predicted quite robustly. It depends on the variable. Of course, youris not the same predicting temperature or the change in precipitation or wind each variable has its characteristics. An important point here is that the BSC, for example, our group is specialized in this medium-term prediction, which is for the next weeks, months, years. Even if science can’t predict at all time scales we are mostly producing these medium one.

Albert: So usually when we talk about climate, people thinks about what is going to happen by the end of the century. There are many companies that they need this kind of information to know what’s going to happen this year or in the forthcoming years.

Toby: When you say medium-term, what’s the upper limit there in terms of years, is it 10 years out? What would be the upper limit of the time scale?

Marta: The upper limit would be like 10 years.

Albert: Yes, 10 years.

Marta: It’s true that when we talk to users from…it depends on the user as well. Even if we can predict 10 years maybe some users from the wine sector are only interested in the next months. We can provide both types of information with the model we are using.

Toby: Okay, that’s very interesting. It seems very clear to everybody that the long-term picture on climate change is not looking good at the moment unless we make very radical cuts to greenhouse gas emissions in the next 10 to 12 years, I think that’s pretty clear.

Equally, it seems to me, talking to wine makers, there’s a lot of very short term volatility, which I’m sure you guys are very helpful for them, in terms of planning against. When we talk about say 10 years time, from running a vineyard and I’m planting vines now, it’s going to take me about 10 years to start getting the kind of fruit that I want. What do your models say about how it’s looking in the next 10 years in the Mediterranean region in terms of climate change?

Albert: The first important point for us is to really understand what’s the information that is needed. For instance, talking about agriculture, the main variables are precipitation and temperature. Different farmers can have different questions, taking into account the specific crops. For instance, we did a project with the Joint Research Center working with…Maize in this case. They had specific questions for Southern Europe and Eastern Europe. Our greatest task was to assess the predictability of the predictions.

We saw that we have some predictability for the coming five years. Between two and five years we can provide information about precipitation, temperature, and also we can define some indicators that are useful for them. This was the result of the project, to assess for very specific needs, we can provide this kind of information.

Toby: If I was a wine maker and I said to you, “in the area closest to Barcelona, let’s say Priorat, it’s a hot region, what kind of temperature rises should I be planning for in 10 years time? Is that a question you could answer?

Albert: I don’t have now the results, but yes, this is the kind of questions that we can provide. First, we have to be sure that the information we trust…and we provide- we call it quality assessment. We evaluate previous forecast in the past, and then we can also provide the specific forecast the people is asking.

sdr

Toby: What kind of tools and resources are available to growers and vineyard owners? I imagine it must be quite daunting for them to engage with a Supercomputing Center. What are the kind of things that they’re looking for? What are the kind of tools that they’re using, other than very short term weather forecasts?

Albert: This is changing a lot in the last year. Now we have, at European level, we have Copernicus, it’s a European effort to put all the forecast, and all the observations available to everybody, and there isn’t a specific program that’s climate change. Copernicus, it’s to provide specific tools and services. Some people is getting familiar with all this information.

I think that now the users they are not directly using all this information, but they are aware and they are testing it. For instance, for any specific field there are some management procedures taking into account this information, and at the end of the season they check the results.

Marta: At the stage we are now, what we are doing with the users for example, in the wine sector, is to work with data from the past. We ask them to identify some particular years in which the yield was very good or very bad. Then we show them the prediction that would have been available at that moment, and we ask them, “which decision would you have taken if you had this information before?” Then we start understanding how they take decisions, and how do we need to show them the predictions that we produce. In this way, they also get familiar and aware that these predictions exist, and can be useful for them.

Albert: There is something that is important here it’s that we have to learn their needs, they have to learn also about the limitations of the simulations, because if they use the simulations in the same way that they are using meteorological forecast for tomorrow, for the coming days, there will be an error. It’s really important to really define the specific problems and then to develop the science to answer these problems.

Toby: Okay, thank you. I’m curious how do you connect with wine makers and vineyard owners, and what do they tell you that they are worried about? Are they worried about rising temperatures, meaning too much alcohol in their wine? The wine that’s produced locally around Priorat, it’s already up at around 15% alcohol. Once you get above, sort of 16% it gets actually very hard to classify or to continue classifying it even as wine. What are the things that wine makers and grape growers are saying to you that they’re worried about?

Marta: Yes. Actually they are worried about many things, many issues. Regarding this levels of alcohol that you are describing, these levels of alcohol are high in wines because we have warmer temperatures and the alcohol accumulates earlier than it used to do in the past. The questions that they want to solve is at, which they need to start harvesting, for example.

If they start harvesting the way they used to do it in the past, for example, the levels of alcohol would be too high. Then if they harvest earlier than usual, it’s true that they can control, they can reduce these levels of alcohol, but then the wine tastes differently. We are just asking about which is maybe the best period to do these harvesting in order to reach a compromise between these two factors. This would be one question but there are others.

They are also very interested in knowing the number of pest treatments that they should apply, because also precipitation is changing, and pests are very related to precipitation and temperature. Then this is something that climate predictions that we provide can advise you about. This is our usual information also for them to plan these pest treatments. There are other decisions as well, but those two I would say are the most important for the short term.

Toby: How do you get connected with grape growers and wine makers? How do they find out about the work you’re doing? I’m curious as to how you come in to contact with them?

Albert: Four or five years ago we were working in a project providing these information for the wind energy sector. In that case we were working with energy companies. Then one wine company that is here, but it’s an international company. It’s called Bodegas Torres…. They saw what we were doing for wind and temperature, and they said, “Okay can you do it for this specific questions in this region?” Then we started to work together. We saw that it was really interesting for us too. We defined it asa specific pretty small project. It was a way for us to develop a prototype, to have some material to show the things that we were doing, some presentations in congress things like that. Now, for instance, we are in two European projects leading this pack, there now it’s more easy, but the beginning it started like that.

Marta: The important thing to stress I think here is that in the past these users from the agricultural sector were just users that had nothing to do with us, but at that moment they are also partners in the project. That means that scientific institutions are participating in the project, but also users. We are committed to deliver in this project. It’s a task that they need to fulfill. We are developing this information together. It’s just not us showing them the predictions that we get, but it’s them asking to us what they want, and us trying to tailor these predictions to their needs.

Toby: Thank you. What do they tell you that they feel that they can do in the vineyard to mitigate against climate change impacts? Do they ask you for advice about this? I ask because in my wine blog, www.sustainablewineblog.com, I’ve interviewed maybe 50 wine makers. I always ask them about climate change. I always ask them about alcohol levels. A lot of them talk about things like better canopy management, and other techniques that they can use in the vineyard. Do they ask you about these things? Do you think there are limits to what can be done in that area?

Marta: Yes. They ask us about these things, and they can do different management actions. It would depend always on the vineyard, so you need really to sit with them and understand the characteristics of the vineyard they own, because not all of them are the same. Not all the vineyards around different places have the same problems.

For example, in one of the projects we are involved in which is called Med Gold, is an European project, the European h2020 project. We have developed what we call an info sheet, which is a piece of information, just two pages, where all these different management actions are described. The users in the wine sectors have told us for example, “I need to know when I need to prune my plants.” or, “I need to know where I need to start the pest treatment that I was mentioning before, or also when they need to start fertilizing or, when do I need to irrigate?”

These different things these are decisions that they come to us with and then they ask for advice. If you want we can share with you this info sheet. I think it’s very complete, because it has two different time scales. One is this medium-term time scales, so the decisions and the actions they need to apply in the next month. Then we also have some actions, some decisions they need to take in the long term. For example, when do I buy new areas for planting, or which varieties will I plant in the future? These are also questions that they ask to us.

Albert: In another project it has a similar approach, but in another project there is a specific question. The project focus on three different fields. One in Portugal, another one in Catalonia here, and another one in Italy. In this case the farmers want to have information about crop forsting. They apply crop forsting for this specific locations. They work with our data to see the difference between the field where they apply it and the other fields that they didn’t apply crop forsting.

Toby: Could you just explain that. You said crop forsting. I want to make sure I understand what that is.

Albert: In crop forsting it’s growing at the beginning of this season. Once the plant start to have leaves and everything, you are intense pruning. The plant has to start What they want to do with this is, at the critical season for the plant that is more less in July-August, it’s a typical season also when we have heat waves. With this extra pruning, you will make a delay on the plant. Then you move the critical season for the plant to September.

Toby: Great. Thank you that’s very helpful. One final question. You mentioned that Copernicus platform I believe, around consolidation of data and information about this, do you see a time coming soon when big data and maybe even artificial intelligence will be able to play a bigger role in helping agriculture in the Mediterranean and wine makers have a real time information flow to improve their farming and mitigate against climate change?

Albert: All these things are starting now. We apply artificial intelligence for specific things, but we use dynamical model. We use a model that resolves what’s happening specifically this year and then we use artificial intelligence to calculate some weather regimes of circulation types and see the impact of these circulation types on specific variables, but we don’t do a forecast, we based it on a statistical properties using directly artificial intelligence.

Of course, there is a lot of things to do here, and also European commission is putting a lot of efforts here, but I don’t know how it’s going to work now.

Interviewer: Okay. Thanks very much both of you for your time.

https://www.bsc.es