Lauren McKinley:
Hello everyone, and thank you for joining us today for our webinar series around supply chain optimization. We have a special event today. We are very excited that you’re able to join us. All attendee lines are on mute, and after the session we will send out a recorded copy. If you have questions for our presenters during a session, please drop it in the q and a chat box and we will get to as many questions as time allows. Today we’re excited to talk about crop science and applied meteorology innovations for the food and beverage sector. I’m excited to be joined by Everstream Analytics chief meteorologist John Davis, who leads our applied meteorology and climate team at Everstream to share more about these exciting new developments. With that, I’ll turn it over to John.
Jon Davis:
Great, thank you so much Lauren, and good day everybody. So yes, today we’re going to be talking about the intersection of weather and agriculture and how that intersection and things that we’ve done with that intersection over the last year plus then have helped the food and bev sector and some of the developments and some of the issues that we’ve had here and some of the challenges that we’ve solved within the space overall with this journey that we’ve had over the past, oh, about year and a half or so, I thought it would be good to do a little bit of a quick history as to how this came about, how the journey started and how things are today and what kind of product that we have going out to companies and clients of ours in the food and bev sector. So we’ll kind of take a look back and a year and a half ago or so, we began to think about some of the challenges that were occurring in the food and bev sector.
And when we think about those challenges, one of our clients, a big beer manufacturer, it was Molson Coors. They had a challenge and their challenge was that in their Eastern European source zone overall here, they had a very difficult time knowing the crops and whether the crops were good or bad before they came into the brewery as to making their beer across that area. It was a challenge they’ve always had, and if anything, it had gotten worse here over the year. So this was a year and a half ago or so. So last year during the 24 growing season for barley, of course barley is the key ingredient for beer and making great quality beer like Molson Coors does overall. So what we did is we looked at that challenge as to beginning to look at visibility in looking at how the crop is doing, good or bad in specific areas before it reaches the brewery, in other words, while it’s in the fields and overall while it’s developing during those periods here overall.
So within that challenge that we really worked together between client and everstream, we worked together for a goal. The goal was to get insight, early insight into the overall quality of a crop like barley during the growing season to reduce the amount of surprises that they would have after the growing season. Overall, there were just too many surprises out there and to have visibility week by week during the growing season so they can monitor this in the entire zone in each individual sourcing country overall. So this was how things develop and really what we did is we looked at making a model that would begin to solve some of these challenges that were occurring and in using overall data analytics, then making a model that would monitor crop conditions during the entire growing season from planting to overall the productive stages to harvest activity all before that crop shows up at the brewery to make the beer overall.
So from a modeling standpoint overall, not to get into great detail on this, so we built a model and we built a model that first uses overall crop information and we call this masked crop information. So we only pull the areas, in this case for Eastern Europe that grows barley. If it grows sugar beets or grows corn, it grows something else. It’s not included in the model, so it’s only looking at barley, and that’s the remote sensing overall. Then we look back historically and we begin to build a historical analysis of temperature precipitation, soil moisture, conditions of vegetative health index, and we build a model for the past 10 years or 20 years, and we look at each week of the growing season, how weather conditions impacted the overall crop. Did the weather conditions make it better, did it make it worse? And at different stages of the season, and that is done by week.
Of course, rainfall in the middle of a growing season when the barley is growing, that’s generally a good thing. Rainfall or harvest is a very bad thing. So that model changes every week to give the best representation of how the crops are doing in the large scale area and individual countries or even in sectors of individual countries out there. So that is the model that was developed here overall during the 24 growing season, in this case in Europe and for barley, that’s a spring early summer crop. That was kind of the test case of this overall, we began to tweak the model a little bit and begin to build it through AI and machine learning as to how the weather conditions impacted the crop quality, the yield if you will, and then refine that during the season and even after the season to make the model very robust out there.
So if we think about the predictive value of this and we think about, well, what does this model entail overall, it’s kind of interesting. So when you think about crop prediction from that standpoint, these are some of the items that are kind of in the overall crop index. And if we can go to the next slide here, Lauren. So in thinking of crop prediction, so we’re not going to again get into all the specific details of this, but we’re taking weather data, soil moisture data, and all of those yield data. We’re combining that to have a model tell us how the weather conditions are impacting overall crop development in this case specifically for barley. And it’s very unique and specific to individual crops here overall. So we build the history, we have weekly monitoring of that as we go through the entire season by country, by region, very important in this.
We also have advisory sessions during the critical timeframes. In this case when barley is reaching maturity, when it’s just beginning to be harvested. So we have advisory sessions talking about the details, so then the client can make the best decisions at that key time here overall from a standpoint of where to source from, where not to source from, where is the quality highest, where is the quality of lowest, so on and so forth. Overall, and again, the test case of this was barley last year across a portion of Europe, but since then we developed this for lots of crops here out there in many areas of the world. But as long as you’re doing this from a crop specific standpoint, you can model any crop anywhere, build a history of how the weather impacts specific crops here all over the world and begin to evaluate that, quantify the overall crop conditions out there before the season is done when the crops are in the fields overall to make financial decisions, to make decisions on where or where not to source overall.
And that was the model that was developed here last year from that standpoint. So I guess with any model and with any kind of data analytics that you use, well one of the keys is going to be, well, how good is the model? This is just a representation as to looking at a season. In this case, the season would be during the spring and early summer. Of course, the individual weeks of the year are on the bottom end of the axis. Week one is the first week of January. You can see this is going to be spring and into early summer, of course early in the overall growing season, then there’s too much of the season ahead to have a high correlation of yield. But once you get to the middle of the season, that changes markedly. In other words, think of this as an R squared type of a situation.
So the coefficient of determination goes dramatically up in this case it would be during the month of May and even into early June. And that is the timeframe when the model overall predicts final yields and looks at the overall situation as to a very high correlation, near 80% correlation 0.8 if you’re thinking of this from an R squared standpoint of overall determining final yields. So we would call this the decision timeline for each individual country, for the region as a whole. There’s a specific timeframe where decisions can be made basis the model, because the model is predicting what the final yields will tend to be across these specific areas here overall. So as we go through the season, then the model is an extremely good predictor, extremely high R squared of looking at what the final yields will tend to be for any crops out there in any areas.
And we do this for each individual crop as we’re beginning to model the overall crop areas around the world, take of an example of the output. So every week we send out a weekly report, and in that weekly report then it’s going to look at how this year compares to other years. So we’ll now look an example, and in this case we’ll look an example of Eastern European barley. How good or bad are the conditions here prior to harvest activity? And what this does is look at the current week in this case where this was in the latter portion of May and how the overall yield in Eastern Europe compares to the last 10 years. This can be a 20 year look. It can be even more than that. It really depends upon the client and what kind of data that they have on their end in looking at the yields here overall.
So when we look at the situation here for barley, in this case in late May, it was actually the best conditions here that we had seen over the last 10 years. But it is a way to quantify the quality of the crop and compare that to all other years. In this case, 2025 is the best, 2022 is the worst. And again, this report is sent out weekly because much of the data comes in on a weekly basis overall. And all the data sets that we use that are integrated into the model overall, we also begin to look at forecast data and how this will change over time. Will the trend of the crops be getting better? Will they be getting worse based on forecast information? And again, a very key part of this is advisory services, getting on with the client and talking about the upcoming trends, what direction things are going overall to make early decisions and be able to use that information in an actionable way to make decisions here as they begin to get the commodities to make the products that consumers love to eat and drink and things like that.
So this is just one example of that strawberries in Morocco, it’s another example of that during their whole season, again week by week. And as you get further into the season, then you get a better determination of what yields will be like for example in Morocco for strawberries. And another example here overall would be a vanilla in Madagascar. And overall here, same basic thing. The growing season from start to finish, of course now it’s kind of harvest activity and Madagascar or Madagascar grows much of the world’s vanilla and where things are aligned right now compared to the past 10 years overall, that is the model. We started it with barley in Europe. It has expanded to many, many crops around the world over the last six months overall. And again, the potential here is because there are so many crops where the overall, what we call the mass data is very granular, very high resolution that we’re able to do this in many areas of the world here overall and begin to model overall crops in sourcing area.
And it’s kind of unlimited the extent that you can do in this kind of situation. And this really comes down to overall data analytics and then looking at remote sensing data and beginning to model that basis of history and then begin to look at the forecast as to how weather patterns overall will tend to change the situation. So this was an example of kind of phase one of this product, right? Well, the second question that would always come up after the conditions this year that you will use the crops here to make products that people drink and people eat. What about longer term? What about a situation where if we’re thinking the next 20 or 30 years from a specific crop standpoint, what areas, how will areas fare either good or bad from a standpoint of climate change? And so that was the next item.
This was kind of phase two of this overall project. And so Europe is kind of an example that we’ve done a lot of work on from a crop standpoint overall. And not to get into details here, but certainly from a temperature standpoint, temperatures are warming and they’re warming to a little bit different levels in portions of Europe. Some areas a bit more than not. The interesting thing is precipitation the next slide. So when you look at Europe overall, and we look at how overall precipitation will change in general, Southern Europe gets quite a bit drier. Northern Europe gets quite a bit wetter. So what we can do is take that climate information and take the best climate models on the planet and we can do the same thing. We can take a look at what impacts barley or what impacts corn or what impacts soybeans or vanilla, and we can project that out into looking at yields overall.
In a lot of cases of the world when you’re dealing with the climate change impact on future yields, there are winners and losers in this situation. But this work can also be done crop specific over a period of 20, 30, 40 years looking at how changeable weather climate change, if you will, will begin to impact overall yield potential. So just kind of finally in Europe, interesting use case, lots of other areas that are interesting use cases. If we think about a crop, this example of one crop here. So in Spain, Spain is one of the areas that has quite a bit more dryness that will occur over the next decal timeframes 20, 30, 40 years. So it tends to get drier with time and also tends to get hotter with time. So for almost all crops across Spain, the yield potential is negative in that. And this is an example of one crop in looking at the projections of yields going forward through the middle of the century and looking at how yields are declining or decreasing over periods of time here because that is where the risk is the highest across the southern areas of Europe.
Let’s compare that to then yield projections further north across Europe in a place like Poland. So in Poland for the north, you don’t tend to have the increased dryness over 20, 30, 40 years in those areas. You do see some warmer temperatures in Poland, but not the extent that tends to stress crops. And in this particular example then the yields in Poland are basically flat. You don’t see a decline in yields over time going through the middle portion of the century. And this is true for crop areas, it’s true for regional areas across the globe. And the bottom line is that again, the same concept that was done for this season’s crop. You can do that for a crop and how the weather affects an individual crop over many, many decades. And again, the kind of line that we always use on this. And when you’re dealing with overall climate science and thinking about how yields change with time globally depending upon the crop, there are winners and losers in this situation here overall.
But that information is powerful and that information is crucial in future sourcing decisions, strategy, if you will, going out many, many decades and where you’re going to be sourcing your crops as you think about things toward the middle portion of this century. But it was just kind of another step, if you will, in the journey, in the process of thinking about getting overall specifics on crop conditions yields, but then also thinking about longer term issues as to how things will trend over time per crop, per area on a global basis here overall. So that’s kind of the journey that’s occurred here over the past year plus and been a fascinating journey and combining weather and agriculture and food and bev. There are other sectors such as pharma that they buy a lot of agricultural products in pharma, but that’s kind of been the journey that we’ve been on and it’s really been a fascinating journey here from a scientific perspective overall. So I guess with that, Lauren, I’ll hand it back to you.
Lauren McKinley:
Great, thank you, John. That was a great session and we hope everybody learns some great new information about whatever stream can do to help support crop sourcing and smart procurement related to climate and weather. So thank you John. A few questions here I see from the team, if anybody has any additional questions, you have a moment to please drop them in the chat Again, any questions we don’t have time to get to, we will make sure to follow up with after the session. Okay. First question for John, what crops are available to create these types of insights for?
Jon Davis:
Yeah, it’s a large database and that has really changed over the past five or eight years through remote sensing in the amount of data that is available and literally looking at crop growing areas, coffee or cocoa or soybeans and just pulling that data out. But the way things stands here right now, it’s nearly 250 crops that are available worldwide to pull out the data, the obvious ones with things like corn and soybeans and sugar beets and coffee and cocoa. But it gets into very specific crops too, like palm oil and individual peppers and places like India or places like Central America. So very, very defined. And so it’s almost 250 crops that we can look at. And we’ve also been doing some work, interestingly enough on even looking beyond crops. So dairy is something that we’re very interested right now in looking at how the weather impacts overall dairy cattle.
And we know that heat stress is very detrimental to dairy production. The cows produce less milk basically, so that also can be modeled here overall from that standpoint. So it’s mainly crops about 250 of them here overall kind of almost unlimited. Not every crop in the world, but a lot of crops here. And it’s also tends to be unlimited from a standpoint of the areas you can look at and model. It can be by country, by area. We have one client looking at soybeans in the northern hemisphere and in the southern hemisphere, very large scale here overall. Or it can be very specific to sourcing areas that a company is looking at overall. So it’s kind of unlimited from that standpoint. And again, this database is growing all the time, which makes us excited overall as to the things that we can do in the future.
Lauren McKinley:
Great, thank you. Another question, what crop areas of the world will be impacted by climate change the most?
Jon Davis:
That’s a real number one, a really good question, and really getting into some of the specifics overall. So when we begin to model how climate change will affect crops, a lot of it comes down to, well, how does the weather affects crops? Some crops are more temperature sensitive, like a corn crop is much more temperature sensitive than let’s say soybeans overall. So for every crop it’s a bit different as to how projected climate change will tend to impact yields overall. So it’s got to be done on a crop by crop basis overall and within areas of the world, we’ve showed a little bit of the work we did on Europe, but in South America and Southeast Asia and Australia, very interesting things and it tends to be very crop specific and it tends to be very regional and it sometimes tend to affect crops totally differently because some crops tend to be more sensitive to heat or excess moisture or lack of moisture, whatever that is.
So again, kind of use the line before in the overall presentation, is that crop wise via climate change, there are winners and losers, and it comes down to individual crops in specific areas. I think in general, if you look at overall agriculture, the tropics tend to be a bit more higher risk than areas in the far northerly latitudes, places like Canada, places like far Northern Europe, things like that. That also is crop specific here as well across those areas. But yet it really comes into the details. The details are really important as to what areas will tend to be hurt the most and what areas will tend to not be hurt in some areas of the world. There isn’t actually an improvement in overall yields out there.
Lauren McKinley:
Great, thank you. John. Looks like we have time for one more question here. So a question around climate impact. You mentioned climate impact related to crops, crop projection yields and predictions. With this model, does everstream also help support facility risk related to climate change?
Jon Davis:
Interesting. Yeah, we do. And so whether it’s like facility risk or thinking about movement of crops here overall, so much of the time once crops are harvested, then one of the prime transportation modes is river transportation is an example of that in the facilities along those rivers, the Rhine in Europe, the Mississippi in the us, the Parana in South America. And so what we’ve been doing lately is doing some work on movement of commodities and the facilities along those rivers. For example, Panama Canal would be another example of that here overall. And in looking at the situation from a standpoint of what does climate change, how does that impact the distribution when you tend to have difficulty in transportation, rivers, canals, so on and so forth. Almost a subset of what we talked about from a crop standpoint here overall. And so that’s work we’ve begun to work on here right now in looking at places like the Ryan or the Mississippi will risk of transportation, movement of a commodity, will that tend to be getting more risky with time, less risky with time? Does the overall distribution tend to change here over time? Possibly more flooding conditions across areas? So yeah, it’s kind of work that we can also do as a subset of the work that we’re doing with climate. It then includes facilities along those river systems here over overall and the canal systems. And then what tends to happen from a standpoint of movement of these commodities as they’re being moved so companies can produce things to drink and to eat.
Lauren McKinley:
Great. Thank you, John. I appreciate it. If anybody’s interested in learning more about the information that John shared today or the crop prediction solution, please reach out to us here at everstream. You can write to [email protected]. Again, we’ll send a copy of this recording to all registrants at the conclusion of this session. And please join us for our next spring webinar series session, which will take place on June 25th about launching a disruption risk management framework. Some more great insights to come. Again, thank you for your attendance today. Thank you to our presenter, John, and with that, we will conclude the session. Have a great day.