Ecosystem Forecasting with Cayelan Carey

(music)

Travis Williams

Have you ever checked a weather forecast? I'm going assume that that's a yes, so perhaps a better question is, why do you check weather forecasts? But for me, I check weather forecasts simply so I can be better prepared for what's to come. Should I wear pants? Should I wear shorts? Should I wear a hoodie? Or should I wear a hoodie with another jacket over top of it? Because regardless, I'm probably wearing a hoodie. All of these little decisions that I make each and every day to be prepared are aided by the help of a weather forecast. But what if we could also get forecast just as frequently about our ecosystems? Well, Virginia Tech's Cayelan Carey is one of the researchers that has developed an ecosystem forecasting system that does just this, and she was kind enough to answer all my questions about it. Cayelan is a Patricia Caldwell Faculty Fellow and a Professor of Biological Sciences, as well as the founding director of Virginia Tech's Center for Ecosystem Forecasting. Her research integrates freshwater ecosystem ecology and data science to examine how humans are affecting freshwater systems and in turn how changing water quality is altering human decision making.

Cayelan helped me better understand exactly what ecosystem forecasting entails and how it benefits people like water managers at different reservoirs and can drastically impact the drinking water that you and I have. She also shared that they are currently helping forecast for lakes all over the world and how having that diversity when it comes to the bodies of water and the locations has really helped them enhance their modeling and enhance their forecasting abilities and she also shared her favorite lakes and her favorite lake activities. I'm Travis Williams and this is Virginia Tech's Curious Conversations.

(music)

Travis

I know a little bit about ecosystem forecasting from reading a bunch of your stories and stuff, but I think I understand it kind of the way my mom understands football and that she knows it's a thing. She knows it's important. Could she break it down for me? No, she probably has no idea what linebackers do. So I'm curious, maybe just to start, what is ecosystem forecasting?

Cayelan

No, I think that's a great question because this is something that's actually quite new. But you probably have thought about weather forecasting. Most Americans, the majority of Americans actually look at a weather forecast every single day, and they use that forecast to make decisions. For example, like, what am I going to wear today? Am I going to go to that picnic? Should I go for that run outside? And our team at Virginia Tech is essentially making ecosystem forecasts to help managers and natural ecosystems make decisions as well.

So the idea is that we are using models that are being updated continuously by data from sensors that are deployed out in lakes and reservoirs, and that they are making forecasts of our water quality today, tomorrow, and for the next month into the future. And then if a manager of a drinking water reservoir receives that forecast, they can then see, wow, this is predicted that it's forecasted that a week from now we might have an algal bloom.

They can make decisions now to get ready for that bloom. That could be installing more filters in the water treatment plant, changing the chemicals that are being added into the treatment processes. It might mean staffing changes, having more folks in hand. It might mean even adding chemicals into the reservoir to prevent that bloom from happening. Ultimately, our goal is to be giving managers up-to-date, real-time information and future information that can help guide their decisions to make them give the best kind of information possible and make it most effective for their communities.

Travis

Okay, so it sounds at its core a planning tool, a tool for planning.

Cayelan

Rather than a forecast of the weather, it's a forecast of ecosystems that are increasingly variable. And I think these types of tools are needed more than ever before because we're seeing increasing variability in many of our ecosystems. And so in terms of thinking about drinking water, we now are having warmer temperatures of lakes and reservoirs. We're having changing oxygen conditions. We're seeing in many places an increased frequency of algal blooms that can be toxic or harmful to your pets if they get into the water.

And so if we can provide this information to managers, they can then be proactive and adapt and mitigate these changes the best that they can.

Travis

How did you get into doing this?

Cayelan

So I am, by training, a freshwater ecosystem scientist. So I love lakes, I love reservoirs. I grew up going on lakes and reservoirs with my dad, kind of fishing, hiking, kind of being in water, something that is kind of deeply embedded into who I am. And a big part of my kind of growing up was seeing the acid rain crisis in the 1980s in upstate New York of like many of the lakes that I loved were changing really rapidly.

And so I knew very early on that I wanted to be a freshwater scientist. And then I knew that I specifically wanted to do research that I could help people that was kind of directly able to help communities around kind of the lakes that I loved. And so when I became a researcher at Virginia Tech, I started a collaboration with the Western Virginia Water Authority. This is the drinking water authority that oversees the reservoirs that provide water for the communities in Roanoke and around the region. And through kind of almost a decade of collaboration, we realized that they needed these tools because they were experiencing water quality impairment, which was very challenging to know what was going to be happening in the future. They had a sense of historical baselines from their understanding of these systems, but it was impossible to know what tomorrow's water quality would be. And so we initially started off by building models together in which we could represent how the reservoirs were working. And then we could tie that with data streams that were coming in and with computers that were able to run models in real time so that we could then go from seeing what today's conditions were to starting to think about tomorrow's conditions and what the water quality would be into the future. And I get to work with a fabulous team here at Virginia Tech. I work with a really remarkable group of researchers that are bringing expertise in ecosystem modeling, cyber infrastructure, computer science, geoscience, decision science, all together. This is not work that did by myself. This was very much a collaborative effort. that's led by a lot of people together that I want to make sure they get full credit for.

Travis

Yeah, so it sounds like you all have developed this modeling system. I guess I'm curious what a modeling system actually is, but also what is it that you all are doing that maybe nobody was doing prior to that?

Cayelan

Yeah, so in terms of like what a modeling system is first, when I think of a model of a lake, I'm essentially think of it as a video game of a lake. So we have in computer space a representation of water that's coming in from a lake. We have multiple layers from the surface to the bottom of a lake. We have representation using equations in which we can simulate what the water chemistry is and what the water temperatures are whether the lake is cold or warm, if fish can live in the water. And then what we can do is that within that model representation in our computers, we can then put in future weather that's going to predict what the weather is going to be over the next month. We can also pull in what the future water that's going to be flowing into the lake will be over the next month. And using those equations and that forecast that are essentially driving our model, we can then make forecasts of what the lake will be like over the next month.

And we can make forecasts of water temperature, the chemistry of the water, what organisms will be there. And then that will also tell us what the water levels will be. And all of these forecasts of all these different ways of thinking about water quality, whether it's water temperature, how much oxygen is present, whether we have water levels are dropping or rising, all of this is really valuable information that the managers of these different lakes and reservoirs where we're making forecasts can then use for guiding how they best handle today's decision making.

Travis

And was that something this, I love how you lay that out in like a video game scenario, because I can understand that. Is that something that no one was doing prior to this?

Cayelan

So to the best of our knowledge, our team here at Virginia Tech put together the first real-time automated water quality forecasting system that we're aware of in the world. And a key innovation that we have done here at Virginia Tech, led by my colleague, Quinn Thomas, is that we are not just making a prediction of the future conditions, but we're also quantifying the uncertainty associated with those predictions. So we're not just giving you a number of what tomorrow's algae levels are.

We're also telling you what our confidence is. So what's the likelihood associated with that number? And that uncertainty is really critical because we could give you a forecast of, let's say that there's going to be a kind of a certain level of an algae kind of in your water tomorrow, but maybe our uncertainty is really high. And if our uncertainty is really high, then maybe it wouldn't be the best to change your current water treatment processes. But if our uncertainty is really low, and we can give you this prediction that there's going to be a bloom, then managers would be more confident in terms of making a change to their current processes. So the uncertainty quantification that's associated with the forecast is one of the most, I think, important parts of what we do. And that is a really important kind of outcome of our work here at Virginia Tech.

Travis

That's awesome. And so I've also heard that you all have scaled this up to other bodies of water. I'm not sure if they're freshwater or if they're saltwater. I'm not sure exactly how many different types of bodies of water or reservoirs or whatever, but I'm curious, how did you go about scaling that up and getting other lakes involved?

Cayelan

Yes, so we currently every day are making forecasts for a range of different water quality variables for 15 lakes and reservoirs around the world. Three of them are here in the Blacksburg-Rownoak area. They're the drinking water reservoirs that are around Roanoke that we co-developed with the Western Virginia Water Authority. We are making forecasts for lakes across the US. We have forecasts for lakes in New Hampshire, in Florida, in Wisconsin, North Dakota and Alaska. these, some of those lakes are part of the National Ecological Observatory Network, which is a national monitoring network where there's buoys in those lakes where we can access the data in real time. And so those data are then available for us to feed into our models. And we have this done using a unique computer cyber infrastructure so that this is all happening in an automated, wireless secure way every single morning making forecasts that are then delivered to the local communities.

Travis

Wow, so they're sending all that data to you all and it's automatically being uploaded into your video game. And then you're able, you're doing that every single day for these folks.

Cayelan

That's happening every day for all these lakes and reservoirs for one to 35 days into the future. So you can see what tomorrow is going to be and then up to about a month and ahead. then every day that is in those, all of those forecasts are updated as we get new data from all the sensors in those lakes. And then, so we're constantly improving and entering on the forecast into the future so that they're getting better as we get closer to kind of projected impairment or into the future.

Yeah, well, that's another thing that I was very curious about. When it comes to working with so many different bodies of water and having this, how does that benefit you all and what you're trying to do? it, I mean, I'm assuming that it's like anything else. The more you do it, the better you get at it. But I'm just wondering if getting a diversity of body, bodies of water, does that enhance it?

I think fundamentally from my perspective as a scientist that the more we can learn about the predictability of lakes and reservoirs, the more we're able to improve our models of how lakes function. So what we're learning, for example, is that lakes of different depths have different predictability in terms of how accurate the forecasts are. We know about lakes that are clearer, that have kind of, can see to the bottom more kind of deeper tend to have more accurate forecasts than lakes that have more turbid conditions. It's kind of when you look at the lake, it's more cloudy on the surface. Information like this is helpful because we need a wide range of lakes for kind of testing these hypotheses. And then we can improve our models so that ultimately we can have a kind of a generalized framework that could be applied to lakes, other lakes and reservoirs around the world.

Travis

What's the benefit of working with folks internationally on a project like this?

Cayelan

goodness. feel like, it's incredibly rewarding, first off, but I think we've also learned so much from working in these very different systems. And an example that I'll give you is a lake in southern Australia called Lake Alexandrina. It's a really large lake. It's shallow, but has a very large surface area. But it also has kind of saltwater intrusion that comes in from the southern ocean into the lake. And so as a result of developing forecasts for the South Australian Department of Environment and Water, which manages this lake. It provides drinking water from the River Murray for the city of Adelaide, which has over a million people. We've learned a lot about how we've needed to improve our models for representing sea-line conditions. So we don't have as many salty lakes around here. And so we've had to think a lot about improving our model structure and framework and kind of the equations within our models so that we can make accurate forecasts of not just water salinity, but a range of other conditions in these kind of more saline systems.

Travis

That's fascinating. don't know a lot about Australia outside of Bluey, but that sounds great. And it sounds like that that's super, it's gotta be super helpful to have that insight from other places. just, when I hear about Australia, I do know other things about Australia. I know that they have a lot of creepy creatures there and just a very different diversity. I think their ecosystem is different than especially Southwest Virginia. So I imagine that is pretty helpful.

Cayelan

I feel like we've learned so much from our Australian colleagues. They're really innovating ecosystem models on the freshwater front. They're facing increasing pressures in terms of climate change on freshwater ecosystems. And we're able to leverage their tools for developing our forecasting system and innovating. And this is really done in very strong collaboration with those communities. Our team has been to Australia to run some workshops over the past year. We've had a National Science Foundation grant where we've been running collaborative workshops with Australian water managers and both the US and scientists back in Australia. And it's been remarkable in terms of just the knowledge transfer across many of same issues we're facing and also slightly different.

Travis

Well, I you mentioned that you've all launched the Center for Ecosystem Forecasting here at Virginia Tech. I think in 2024 it launched officially. It hasn't been very long since that launch, but I'm curious, what's the vision for that? And we talked a lot about water, but is there the potential to do some ecosystem forecasting in other areas?

Cayelan

Yeah, so our aim of our research center is to make actionable forecasts that lead to insights. And that can be insights into both understanding how ecosystems function, as I've described before, but also for insights for managers and other decision makers and policymakers. Our team has a very strong focus on water, but we're also making forecasts for a suite of different systems. For example, Quinn Thomas is leading a national, actually international forecasting challenge, which is collecting forecasts for forest carbon fluxes. We're also leading a forecasting challenge that's happening for ticks and for beetle communities too, in addition to water quality for lakes and reservoirs. And in sum, that forecasting challenge has collected almost 6 million forecasts to date from communities that are coming in and developing and sharing models from across the US and beyond. So it's possible that you all might be able to forecast like how many ticks we'll have or the potential for really back ticks in my backyard next year.

We have certain sites that are located using observational data, again, from NEON, the National Ecological Observatory Network. And they are sharing data that we're leveraging. And one of the monitoring sites is actually up at Mountain Lake Biological Station. And so that's pretty close to us here in Blacksburg. And that's a great place where we can access data and where we're making forecasts.

Travis

Well, I got through all the questions that I had wrote down. So I have two questions that I'm curious about. One is what's your favorite lake or body of water? And the second one is what's your favorite thing to do at a lake? Because we've talked a lot about, I mean, I know you're focused on a lot of things happening in the lake, but what do you like to do with the lake?

Cayelan

Great question. I think this is an important thing for every freshwater ecologist to know. I have two favorite lakes. My first is Lake Sunopee, New Hampshire, which is a really beautiful, clear, glacially formed lake. It's a very deep, very high water quality system. It's a lake where I've been working for almost two decades in partnership with a local lake association there. And my second favorite lake, if I'm allowed to have two, is Falling Creek Reservoir, which is a small drinking water reservoir located outside of Roanoke and it's where a lot of my lab does work here. And it's just a place, it's located in this beautiful mountain valley. And when I kind of drive to the site and I get there and it's completely in this beautiful forest, I just have this sense of calm. There's been a lot of research done on how when you stare at lakes or other bodies of water, it just decreases your stress levels. And I feel that instantly when I'm at both of these places. It's just kind of my home, if you will.

And your second question was, what do I like to do in water? I like to collect data, but I also like to get in water and I like to swim. There's something really special about being underwater and to see a lake from underneath. think there's a lot that we can learn from being submerged completely in our ecosystem.

Travis

Well, that's cool. And those two things kind of go hand in hand. You collect the data and then you know you're safer to be submerged in the water. Awesome. Well, thank you so much for talking to me about this.

Cayelan

No, thank you for asking these questions. And I think for folks that are interested in our work, we love to talk with folks. And I think ultimately, one last point I want to make is I think of Virginia Tech's mission as a land grant school. This is work where we're really uniquely aligned with that and that we're doing this kind of cutting edge scientific innovation. But we're also doing work that's directly relevant and useful, hopefully, for the communities that are around these bodies of water. And it's something that I feel really proud of being a part of here.

(music)

Travis

Thanks to Caleyan for helping us better understand ecosystem forecasting. someone you know would make for a great curious conversation, email me at traviskw at vt.edu. I'm Travis Williams and this has been Virginia Tech's Curious Conversations.