Disclaimer: This is an automated transcript, we apologize for any errors. If you notice any problems, please email the show at teachmeaboutthegreatlakes@gmail.com. Thank you.

Stuart Carlton 0:00
teach me about the Great Lakes. Teach me about the Great Lakes. Welcome back to teach me about the Great Lakes a podcast in which I A Great Lakes novice get people who are smarter and harder working than I am to teach me all about the Great Lakes. And I'm enjoying this month this week. I'm joined this episode by my good friend Carolyn Foley. Carolyn, how are you?

Carolyn Foley 0:21
I am doing well. Thanks, Stuart. It's the weather's kind of wonky right now. So like I was doing laundry and there are super long sleeve shirts. And then there are sweatshirts, and then there are super short T shirts. Yes. Yeah. It's summer in the Midwest, I

Stuart Carlton 0:35
guess somewhere in the Midwest? I guess it's exactly right. I agree. The weather is wonky here last night, we had like a super long, it was like the longest thunder I had heard since. I don't know hurricane Harvey essentially. And I couldn't tell if it was thunder, or if I was just hungry ao anyway. Getting punchy already didn't sleep well. Last night. Carolyn is the main thing because of the aforementioned thunder, and because of the previously mentioned one and a half year old daughter. But so this week, I'm actually really excited. I think I say that every week. I'm always excited. I'm an excitable boy. We're gonna be talking about modeling, right. And so I've been thinking about modeling a lot, because it feels like modeling runs our life in some ways. And so I'm really excited to talk about modeling.

Carolyn Foley 1:24
But um, can we just clarify that we're not talking about like a marriage that America's Next Top Model modeling?

Stuart Carlton 1:29
Yeah. Why don't you clarify for that? Or clarify that for us?

Carolyn Foley 1:33
Well, no, that's really all I wanted to say is the modeling that we're talking about. If you're thinking that you're about to hear something about fashion shoots, you might not,

Stuart Carlton 1:39
that's such a solid point. And this shows you where I am in my life that it never even occurred to me that that type of modeling exists. But yes, that is also a type of modeling. Frankly, those types of models make more money than the people who make the types of models we're about to talk to. If any of those models are listening, and would like to sponsor the show, just reach out at Teach Great Lakes on Twitter. But anyway, this week, we have Dr. Madeline McGee, with us and she's going to talk to us about she does some, I'm gonna ask her, but I think it's like water modeling of some sort. And we're gonna use that to engage in a conversation hopefully, about modeling in general. But first, I have to decide which transitional music to do. Let's go with this one.

Dr. Madeline McGee is the office of great waters monitoring coordinator with the Wisconsin Department of Natural Resources. Madeline, how are you today?

Dr. Madeline Magee 2:34
I'm doing well. How are you?

Stuart Carlton 2:35
I'm doing great. Thank you so much for coming on the show. So let's start like really basic office of great waters monitoring coordinator. First of all, what's a great water? And what is the monitoring coordinator? Do I got the office part? Nailed?

Dr. Madeline Magee 2:49
Yeah. Great questions. So Wisconsin, DNR a couple of years ago went through a restructuring. And as part of that restructuring, they formed my office called the Office of great waters, we used to have the office of Great Lakes, which makes sense. And then we had a team that worked on the Mississippi River. So for those of you who aren't familiar with Wisconsin geography, we share a border with Lake Michigan, Lake Superior, and with the Mississippi River. The Mississippi River and the Great Lakes have many different issues, but they sort of overlap in the sense that we're working with other states and sometimes other countries. So they decided to combine both groups. And we have the opposite of great waters. So the great waters are the Great Lakes and the Mississippi River. Excellent. Well,

Stuart Carlton 3:39
I grew up right, right on the Mississippi River in New Orleans. And so I agree, it's a very great water, rock solid water among among the greatest. Okay, and so what do you do as a as a monitoring coordinator? are you monitoring like the water levels? Or what's the deal there?

Dr. Madeline Magee 3:51
So I've been in this position about two years, and I'm gonna say, I think I will give you a different answer than the person who was in my position before me. But generally, I am I helped coordinate the monitoring of both the Great Lakes and the Mississippi River in terms of the status and the health of both of them. So things like water quality monitoring, our office deals with the areas of concern, which maybe you guys have talked about before on your podcast. We'll Oh, well, this is like a nice little intro then I guess.

Stuart Carlton 4:27
Yeah, I guess we're drinking for a few weeks.

Dr. Madeline Magee 4:29
jumping the gun on what you have planned. So we design I hope design studies that assess the condition of those areas of concern and determine whether we have kind of met our goals in terms of cleanups and improvement of beneficial use. On the Mississippi River side of things. We have the long term resource monitoring program. That's a multi state effort in coordination with USGS and Army Corps of Engineers to kind of us As the condition of the Mississippi River, and then one other part of my job is also beach monitoring. So I am the person that leads our beach monitoring for E. coli of the Great Lakes beaches to be protective of

Stuart Carlton 5:17
public health. Is that mainly that's in Wisconsin, only where you do that? Yes. So how big a deal I'm already getting this last and I apologize for that. But but so when I lived in Texas, like the beaches were the beaches had a lot of quality issues, right, because there's so much petrochemical production and transport in the area that the beach and like the beaches were really gross a lot of times. So is that like a big deal in Wisconsin talks are so many wonderful beaches in the beautiful lakes.

Dr. Madeline Magee 5:43
I mean, I, I maybe I'm probably biased. I would say we have really great beaches. Sure. Yeah. But we so what we monitor for is E. coli as a fecal indicator bacteria. I know big fancy word it

Stuart Carlton 5:59
means Believe me, I got people.

Dr. Madeline Magee 6:05
Life is fecal, I know you talk about it a lot don't use fecal. So as I'm sure you probably know, having small children, there's a lot of gross stuff in fecal material. And it's really hard to monitor for all the possible gross stuff that's in the fecal material. And a lot of that gross stuff can end up in our waters, and on the shores of our beaches. So we monitor for E. coli, instead of monitoring for the plethora of all the gross stuff that can be in there. As a way to say, you know, this beach has high E. coli that indicates there could be other things that may make you sick. So, you know, we have different levels. Yeah, as an indicator to say, maybe it's just stay out of the water today.

Stuart Carlton 6:53
But that's rare where that actually happens, because the beaches are so nice.

Dr. Madeline Magee 6:56
Yeah, I mean, our our beaches are pretty nice. Most Great Lakes beaches are pretty nice. There we

Stuart Carlton 7:01
go. There we go. I'm sorry, I'm focused on this because I wanted to go to the beach up in Northwest Indiana with my kids tomorrow, actually. But it seems like many of them are closed currently. So I'm just dreaming of beaches. So any beach talk is welcome, fecal or otherwise.

Carolyn Foley 7:16
So when you're monitoring for E. coli, do you? Do you actually go out and sample the sand? Or are there other things that you guys do? Yes. So

Dr. Madeline Magee 7:25
mostly, we sample the water at the beaches. I do coordinate with some folks on sampling the sand, but that's more on a kind of research basis, and that really to be protective of public health. So you go out, basically where small children are swimming, so like knee high. Take a sample of water. And depending on how big the beaches, we'll monitor either at one location to the beach, or multiple locations across the beach, as well.

Carolyn Foley 8:00
So I asked that question, for a reason. Stewart probably was like Carolyn, you're still taking us down the world away from modeling, but actually, um, so we have a couple of buoys at Illinois-Indiana Sea Grant. And when we were trying to get a new buoy, one of the things that the local USGS office who does some of this type of monitoring in Indiana beaches, actually told us that the they use the wind direction on the buoys to try to actually that they work it into a model that then can help them. So do you guys get that at all?

Dr. Madeline Magee 8:35
Yep, yep, we do. Not at all of our beaches. So we have almost 200 beaches on the Great Lakes coast in Wisconsin. And not unfortunately, not all of them are monitored. We basically have to prioritize based on what we know about water quality and what we know about usage. So obviously beaches where a lot of people are going to be at we monitor more frequently than beaches that have lower use, or you know, normally better water quality. And then to kind of supplement that is that monitoring is we have these Nowcast models. And so what they do is they predict the beach conditions, the water quality conditions at that beach for that day. So our beach managers, which are sometimes the public health departments, so in Wisconsin, I was maybe getting into the weeds a little bit but in Wisconsin authority to post beaches as open or closed actually lies with the Public Health Department, not with the DNR. I think that's different depending on the state that you're in. But in Wisconsin, it lies with the local public health department. So sometimes our beach managers are people at the Public Health Department and they run the models, although part of my job is to help them get the model set up and evaluate how well they are working if they need assistance with that sometimes those public health departments will contract to a consulting company or two At a university so UW Oshkosh does a lot of our beach monitoring and modeling here in Wisconsin for some of the beaches kind of like around our county, in Green Bay Area. Anyway, so yes, to use models. And what we do is we use the monitoring data and then environmental variables to develop a statistical model. That gives you essentially a prediction of whether you coli is high. That would be like a closure level, whether it would be an advisory level or whether it would be in the area that we call open, so that the beach manager, usually public health can make a decision about whether the beach should stay open or closed even without specific monitoring data. And the exact environmental variables depend on the beach. So when you're setting up the model, what you'll do is you'll start by looking at things like temperature, rainfall recently. So we do know that high E. coli levels are often correlated with high rainfall levels. You'll also look at things like wind speed and wind direction or waves, if you have wave information, all those can be correlated with beaches, and then

Stuart Carlton 11:12
develop your runoff from there. So you get all that information. So that's all based on measurements that are made by you know, probably oftentimes buoys maybe sometimes Weather Service. Doppler Doppler is I bet they're Dopplers involved. And, and so and so you take all that information, and you plug it into a statistical model. Now, I've had a fair amount of stats, but I'm also pretty dumb about stats. So so like a statistical model, when I hear that you think about like regression, is this like a regression type thing? Or what? What are you doing? Exactly? Or will not exactly what are you doing? Exactly? For somebody who's dumb about it? Like me?

Dr. Madeline Magee 11:47
Yep. Um, so basically, that is, what you're doing is you are making a series of regressions. And so for the beach modeling, we'll use multiple linear regression. So when you think about a regression, like in an Intro Stats class, it was maybe, you know, you have something on the x axis, like age here, I hurt my knee recently. So we'll say age of person. And then on the y axis, it's like, number of, I don't know, lifetime number of surgeries. Um, you know, and so you'll see that there's a correlation, I'm just assuming there's a correlation between those two variables. You know, you'll have a bunch of data points, and then you'll kind of fit a line to it. Yeah, when you use multiple linear regression, you kind of use that same idea, except instead of just having 1x. And one y, you have kind of multiple of your x variables.

Stuart Carlton 12:39
So So, so help me understand that. So is that the same as when you hear about like the idea of controlling for something? Is that sort of the same idea? Or is it you know, this is influenced we know, we were trying to predict outcome was based on these other variables that we know predicted. And so it's, instead of just saying, it'd be like, the effects of age on number of surgeries that also be the effects on age and I don't know, weight and activity and things like that, right?

Dr. Madeline Magee 13:05
Yeah. So that maybe I picked a bad example, because I have no idea how they control for age and health studies. But yeah, so you

Stuart Carlton 13:15
assume you do now trust me, hey, yeah.

Dr. Madeline Magee 13:17
So I, you know, depending on what, what your study design is, and how you set up your model, you can control by looking at the statistics, or you can control by basically splitting things into groups. So for maybe the example that I use, instead of just looking at age two surgeries, you would have one group of males and one group of females. Or then also, you know, splitting by demographics. So where they live can often be an indicator of, you know, their economic affluence. And, you know, maybe people who are more well off can afford surgery, so they have more surgeries. And you can kind of split them into groups doing doing it like that there are more advanced statistical techniques, as well, where you can kind of pull out some of those relationships. Without, you know, more simply just looking at one, one at a time.

Stuart Carlton 14:10
Let's let's pull it back to the ones you do know, we can we can speculate about agent surgery. Yeah, it's just, but now I want to get so you're saying oh, I don't want to get too in the weeds. Well, that's where we differ. I want to get too in the weeds. So you have this now cast, one of probably several models use and you collect all these data, and it spits out like a category. This speech is good. The speech should be closed. Is that right? Yep. Pretty much. Okay, based on some range, I assume of

Dr. Madeline Magee 14:38
Yeah. So in Wisconsin, we have thresholds if you were to take a measurement of E. coli. Based on that measurement, we have classes so like 235 envelopes open 1002 to 35 is an advisory approach per million. Yeah. Kind of counts per milliliter. Okay, okay. Yeah, and so those thresholds are based on risk. And so what you do in when you're developing kind of the nowcasts model, is there's two approaches to doing it. Some nowcasts models will try to correlate, you know, what the model predicts as the E. coli value directly. And then you're just kind of making a decision based on what the what the model says. That that, you know, modelled count is, what you can also do is you can kind of further classify it by say, well, here's our modeled count, based on all these regressions, where does it fit within our advisory levels, and then make a decision from there. But that way also takes a little bit into account, things like uncertainty. So there is some uncertainty in every model that you develop. And if you're kind of right in the middle of the zone, you can feel maybe more confident, if you're on the edge of the zone, you know, between that open and advisory level, maybe then you have to make a decision based on your models uncertainty as to whether you actually leave it open, or whether you would post it as an advisory.

Carolyn Foley 16:13
So you mentioned uncertainty what you what exactly does that mean, kind of why is there uncertainty? So

Dr. Madeline Magee 16:19
why, what is model uncertainty? And why is there uncertainty? If you I think if you were to ask a bunch of different people, what model uncertainty was, I think you would actually get different answers, they will be kind of in the same area. But there's not really like a, at least as far as I've seen, a clear and consistent definition of these things count as uncertainty, even though I think everyone kind of understands them similarly. But the way I like to think about uncertainty is model uncertainty is basically the stuff that you're not capturing in your model. So I'm gonna give an example of phosphorus and chlorophyll in a lake. So I do a lot of research on lakes, a lot of people know for, you know, inland lakes, is different for the ocean, we're going to talk about the ocean because I don't know much about the ocean. solidness, but all I know, and what my kids learn on wildcat, so I know a lot about whatever ocean creatures they talk about. But for inland lakes, we had a lot of really great research, you know, in the 60s 70s, that showed chlorophyll concentrations in lakes are related to phosphorus, right, I don't know if you guys have talked about this on your podcast, we'll add it to the list, it's an awesome, good thing to talk about. I think a lot of people kind of generally know that right? If you have more nutrients entering your lake, then you have more kind of algae production in your lake. And four in the lakes, the kind of limiting nutrient is phosphorus, mostly Spicers. So you can develop a model that relates the phosphorus concentration in your lakes to the chlorophyll concentration. So if you think about kind of your your regression plot graph, you would have phosphorus on the x axis and chlorophyll concentrations on the Y axis. So chlorophyll is how you are representing algae, basically. So you may get this relationship by looking at one lake through time. Or you could get this relationship by looking at, you know, every lake in Wisconsin that you have data for, and kind of plotting it all together. So that relationship though, that model, that statistical model that you've developed, isn't going to capture things like, Well, what is the impact of temperature on that day? To the algae? It's not going to capture things like, Well, what about other nutrients besides phosphorus? What about nitrogen? What about carbon in the system? It's not going to capture, you know, complicated food web dynamics, you know, if you have more fish that eat algae, you will probably have less algae in your lake than when you're comparing to a lake that doesn't have any fish in it. Right? Or, you know, something like that.

Stuart Carlton 19:17
So there are all these things out there that you know, you're not measuring that influence it right. These unknowns, I'm reminded now, as you're saying that of the idea of known known unknowns and unknown unknowns, right, which I first came aware of Thanks, Don Rumsfeld. But that's a different podcast. And so is that kind of what you're talking about there? And so those go into this uncertainty idea?

Dr. Madeline Magee 19:40
Yep. Yep. So all those things and so you can know things. You have your known unknowns, right that go into your uncertainty. There's also going to be things that you know, we don't know about yet that are going to lead to model uncertainty. When you have more complicated models, you know, outside of a statistical model. My background is a lot in hydrodynamic modeling and thermodynamic modeling of lakes. So we're taking a physical equation that we know, you know, to be true to be real and accurate. But then we're putting that into a computer. And so when the computer is doing all those calculations, you're gonna get some kind of error from those calculations. And so that is in the uncertainty, you know, that can be included in uncertainty as well.

Stuart Carlton 20:25
But so this is where people start to get hung up. And so I have questions about this, you're talking about putting all these things in your computer, like you take something, you know, in the real world, and you put it in computer with other variables, and it spits out a number and some uncertainty? Like, is there any way to know that the models are like any good? Is it all garbage? Like, do you ever is there a way to verify those? Do you verify them typically? Or what do you do as far as that goes?

Dr. Madeline Magee 20:47
Yep. So whenever you make a model, you do some kind of verification and the details of how kind of depends on the model that that you're building. So if you're looking at a is just a statistical model, one way, and you do the same thing, if you're thinking about a more mechanistic model, or fancy computer model, but one way of doing it is to take a set of data that you are using to build your model. So you may say, I'm gonna take data from half the lakes in Wisconsin, develop my phosphorus and chlorophyll relationship, you develop your model, try to get an estimate of uncertainty based on kind of the the statistics and the things that you know, and how well that, you know, simple model creates the data that you're using to build it. Then what you do so this is called calibrating the model, then what you do is you take the model, so whatever, either statistical equations that you've used, or, you know, computer simulation that you've used, and you plug it in with a different set of data. And so this is called model validation. And from that, you can say, Okay, I trained it on one set of data. How does it recreate a different set of data? Does it accurately recreate it? So

Stuart Carlton 22:12
if you have enough, in this case, legs, you can sort of split it, split it in half, and say, Alright, let's use this half to develop it. And then we'll test it in the real world and see how well it works using the other half of your sample, right? Yep. Interesting. Okay. And

Carolyn Foley 22:27
that's why things like long term monitoring or broad scale monitoring are actually really important. So it'll help you make a better model, right? Well, I mean, basically, if you you mentioned it, Stuart, that if you have enough lakes to do this, the the validation of the model, then you have more confidence that your model is good, right, Madeline is that?

Dr. Madeline Magee 22:52
Yep, yep. So if you have more data that you can use to create the model, and then also validate the model, you can be more confident in your model. The other thing that you need to consider is the, the the range or the conditions of your observation data that you're using to build the model. So I'm going to use a, you know, a lot of my PhD research was in kind of projecting climate change impacts on lakes. So we have, you know, great information from right now. But we don't really have a set of data that we can pull from that's going to say, okay, for lakes, specifically, we know what the current year, current air temperatures, our current climate conditions are. And so we're using those models. And then we are going to put in the climate projections to the models that we've built. But that's assuming that those you know, air temperature versus phosphorus concentration or whatever ice cover relationships that we develop using the data we have now are going to hold true when, when we're under a different condition for climatic conditions. Yep, yep. So if you have a set of data that has more variability into it, you know, a larger range. So think back to our phosphorus model. If you are only building your model on lakes where phosphorus is low, and then you try to use it to talk about lakes where phosphorus is high, it may not be entirely valid.

Stuart Carlton 24:30
So that's interesting. So So in developing these models, what I'm hearing is it just requires a lot of domain expertise, I think, right? And, and so it's, I think about like, right now, there are three models that I think about all the time. Not literally all of the time, but figuratively all of the time, and that some

Carolyn Foley 24:51
portion of my brain is always thinking, wow, I would argue

Stuart Carlton 24:55
that some person or a brain might not always be thinking Carolyn, but I'll stick with it. And and those are right now. It's COVID-19 models of which their tongues you know, I go to the 538 homepage and I look at the COVID models all the time, other climate models, which you know, my research, I'm not a climatologist, but a lot of my research has to do with climate change beliefs and attitudes as a social scientist, and then like election models are going to suddenly start cropping up. And there's all sorts of ways to do that. But so developing those models takes like a ton of domain expertise. And I think we saw that with COVID. Were people who are not experts in immunology, or epidemiology, maybe we're developing some models that, at least at first, kind of stuck. So when you're looking at models like that, how, how can you evaluate them? You know, what can people do when they see those to evaluate the quality of the models? I guess?

Dr. Madeline Magee 25:47
Yeah, that's a great question. And also something that I'm gonna say I get personally irritated with sometimes when they definitely cute don't know about this, stop talking about it. So one thing to do, is as kind of a lay person who sees maybe you see something about a model on the news, or whatever. So first of all, you have to understand that if you are not at the primary source, so at that peer reviewed journal article, you should go look for it, you should go look for you know, the 530 blog, where they actually talk about how they came up with the model, or if you know, a COVID model is mentioned on the news, you should try to catch the name of the author or where the study was done and try to look it up so that you can actually evaluate it properly. Then when you look at a model, some important things to consider are, who actually is publishing this model? Do they have expertise about it? You know, I have a lot of expertise on Lake modeling. I do not have any expertise on ant colony modeling. So if I were to publish something about an ant colony, you should be like, I'm not entirely sure that this is the right person to be talking about this. And it may be that, you know, one person is has expertise in computational modeling, but they have subject matter experts as their co authors, you know, that would be one thing to look for, who is actually putting the model out? Is it? Do they have the expertise to be to be doing it in the first place, then, if you are able to dig into the weeds, one thing to look at is the assumptions of the model. That's always really, really important. I'm going to use COVID as an example. I am not an expertise in epidemiological modeling. But you've seen a lot of work come out talking about the importance of masks. If one model simulates where nobody wears masks, and another model simulates where everybody's wearing masks, they're going to have two different results. Right? So you're saying, Well, how do I compare these two models? Look at the assumptions, one assumes no one wore masks, one assumed everyone wore masks? Probably the answer is going to be in the middle of those two extremes, right. So look at the assumptions that are made in the models. And even if you are maybe not a subject matter expert, I think you can still assess whether those assumptions are correct or figure out whether how those assumptions may have affected the results of the model. For example, the next thing that you want to do, I think, is look at how the model itself is presented, what what are the drivers of the model? Then you want to look at how the authors present the uncertainty and the results of the model. So when you actually get to this, you know, primary literature, a lot of modelers will do a good job of explaining uncertainty, say we have uncertainty because of this, our result can change because of this. So digging into those weeds a little bit will help you interpret the model in the way that you need to to make whatever decision you're trying to make.

Stuart Carlton 29:29
Well, this is really interesting. And so I think that modeling is going to become more not less prevalent in our life. And so I think this is a really good start. We're going to talk about this kind of stuff again and again and again. Because I think it's a really great example of how to use the context if somebody's really, you know, first of all really interesting work that you do in terms of how do you predict how much poop is on a beach? And no, that's

Carolyn Foley 29:49
not what she's predicting.

Stuart Carlton 29:55
More to the point, the predictors of poop, which was an My band in high school, and, and so, so that anyway, but using that time to talk about this part of thing and this has been really, really wonderful and pretty interesting. But that's actually not why we invited you on this show. The reason that we invited you on teach me about the Great Lakes is to ask these two questions. And the first one is this. If you could choose to have a great donut for breakfast, or a great sandwich for lunch, which would you choose?

Dr. Madeline Magee 30:25
I'm going to respond with a clarifying question. Am I alone? Or are my kids with me?

Stuart Carlton 30:33
That is a good question. I want the answer for you, not your kids, because anybody who chooses to have a doughnut with their kids is insane. And so we are going to pretend for a glorious moment that you're at a conference, you're winning an award in your hometown for best modeler. And, and your reward is you get to pick a great donut for breakfast or a great sandwich for lunch all by yourself. No kids, if you have a partner or spouse, they can be there or not. That's really up to you.

Dr. Madeline Magee 31:04
So I'm gonna go with a donut because my kids really like donuts. And I usually don't actually get to eat a hold on it ever. They're less enthusiastic about sandwiches so I can usually get a sandwich.

Stuart Carlton 31:20
So you get a whole doughnut, not just the doughnut hole. Good. So what what kind of what kind of doughnut so you're in I can't remember you're in the middle of nowhere Wisconsin outside of Madison. And and so where where I'm visiting the middle of nowhere, Wisconsin, where should I go to get a really great download.

Dr. Madeline Magee 31:36
So in my middle of nowhere, Wisconsin, we I mean, I don't actually technically live like in a town. 13 acres, but the closest actual town only has I think they have 7000 people. So I usually get my doughnuts either at actually the grocery store has a really good bakery with good doughnuts. Or there's a local little restaurant that is known for their breakfast. So they have some donuts their

Stuart Carlton 32:07
doughnut restaurant. Never thought about that. That's excellent. I'm in on that. Okay, and then

Carolyn Foley 32:12
the restaurant. Oh, yeah, it's called.

Dr. Madeline Magee 32:13
It's called Schubert's

Stuart Carlton 32:15
Schubert's. I'm gonna take a guess s ch, u b, e r t, apostrophe s. Yep. I will look it up. And I'll put a link in the show notes if they have a website, although in a town of 7000. I don't know. We'll see. And if so, I will be sure to add Schubert's we're gonna have a tour one day we got to get funding, it's gonna be hard to convince the government to fund this, I will be honest. But we're gonna get funding to do a doughnut and sandwich tour of the Midwest. If we call it a trail. Maybe they'll fund it if we call it a trail, local food heritage lunch tree, you

Carolyn Foley 32:45
know, the models that we talked to you at the beginning of the podcast, they can support this Yes, particular. Yes.

Stuart Carlton 32:51
So models call me. All right, great. So the next question is what is one piece of life advice that you have for our listeners, we'd like for them to take something home. In addition, all this really great knowledge about modeling, you know, what's a little something that you want to share with them in terms of how to live the good life? It can be big or small, serious or silly?

Dr. Madeline Magee 33:13
Yeah, this is a great question. So I think the best advice that I've ever been given that I didn't always follow, but I definitely think you should follow is when you are trying to make a decision about something, you should go with your gut. So I think sometimes, especially like I have a science background. We'd like to you know, do a pros and cons list and kind of overthink things. My gut has never been wrong. So I think you should you know, if Should I take this job? Well, are you excited about it? You get that like happy fuzzy feeling, then yes, you should probably take the job. Do you kind of have like that? Like, I don't know if I should take it and really feel it kind of like feels weird in my belly, then you probably shouldn't take it. And I think that's so far worked out in my life.

Stuart Carlton 34:05
As somebody who used to be a natural scientist and as trans slowly evolved into a social scientist who does more and more qualitative research. That is music to my ears Madalyn. Great, wonderful piece of advice. So where can people go to find out more about you or the work? Is there like a social media thing or website? Where should we send them? Yeah.

Dr. Madeline Magee 34:23
Great question. I have a website. Hey, I gotta find out what it is because I don't actually haven't memorized. It's Madeline mcgee.weebly.com. I will link that in the show notes as well. Yep. And I also have a Twitter and my Twitter handle is at Madeline army, gi Madeline

Stuart Carlton 34:41
our McKee are for modeler. Excellent. Well, Dr. Madeline McGee, the Office of Great Lakes, or great waters, excuse me, Mississippi, offensive great waters monitoring coordinator with the Wisconsin Department of Natural Resources. Thank you so much for coming on to teach us all about the Great Lakes. Yeah, thanks

Dr. Madeline Magee 34:59
for having me.

Stuart Carlton 35:05
What a great opportunity to learn about modeling and really get into some nitty gritty detail. It's such a fascinating topic, I think.

Carolyn Foley 35:13
Yeah, I think that was really, really great. Yeah. But I think all of the episodes that I'm on wind up being a little bit nerdy, but that's great.

Stuart Carlton 35:21
What an amazing coincidence. It is. Right? Yes. incidents? Yes. Carolyn was something you learned about the Great Lakes today.

Carolyn Foley 35:29
It's not necessarily something that I learned. But I want to say that, yes, Wisconsin does have a lot of really pretty beaches. Yeah. And it's great that they are being monitored.

Stuart Carlton 35:40
I feel like many is really great. They're being monitored. I feel like they're there. There are two types of Great Lakes states, those with Wisconsin envy, and those that are Wisconsin, although I don't think Michigan fits in there very well. But yeah, they do have a lot of really great looking beaches. And so we can

Carolyn Foley 35:54
do another podcast on the Upper Peninsula and where it truly belongs, because there are some people who do not believe that the Upper Peninsula should be part of Michigan,

Stuart Carlton 36:02
oh really wants to hug it all for themselves. Classic Wisconsin move. Yeah, that's excellent. So, you know, what I learned today was really thinking about, you know, when I've done some modeling and statistical modeling, and stuff like that, and my research, but I haven't really thought through how they're used to make real world decisions in this ways, and how those things because usually, I'm trying to explain a phenomenon not predict something. And so to talk to somebody who was predicting things day to day through a Nowcast type thing, I think was really interesting. And a lot of good tips, which we will summarize a lot of good tips, I think on how to interpret models that you might see the real world. Well, thank you so much for listening. Everybody you can go to check us out on social media at Teach Great Lakes is the Twitter thing. And that's all we have. You can see our website www dot teach me about the great lakes.com Carolyn, where else can they find the work that we do at Illinois-Indiana Sea Grant which I realized I have not mentioned until right now.

Carolyn Foley 36:57
We are on Twitter, Facebook and Instagram. If you search for i l i n s e a gra en ti, you will find it also we have a website. I see grant.org

Stuart Carlton 37:12
Go to the website, go to the social media do the thing. And we will see you in a couple of weeks or so if you remember we're not releasing on the first Monday of every month and the third Mondays of most months. Maybe some months. We'll see. Either way, we'll see you and in the meantime, keep great in those lakes at De DPDT Ah, it just occurred to me so donuts Tim Hortons There we go. We went the whole episode with you on and I I did not make a stupid hackneyed joke about Canada. So I just found out when we were a kid this was going to get out of out I'm sure the there was a brand called President's choice. Do you know President's choice? It's like a store brand of food. Yes, we ate President's Choice decadent cookies when I was a kid because the superstore which is this sort of local grocery store where they roller skated in the aisles. to stock the stock boys and stock girls are gonna stock men and women would roller skate that was a big deal superstore. And we had President's Choice cookies. And on the back that all this stuff in French, and my uncle came over and was like, Well, that seems I won't use the exact word to use because it involves a bleep. And you told me I'm not allowed to bleep. And so they they had all this frontrunner my uncle said, that seems like a big deal over just a cookie with a bleep in there somewhere. And I always thought you're right. That is kind of weird, but it occurred to me just yesterday that that's because it's a Canadian brand. So yeah, thumbs up.

Carolyn Foley 38:36
Really good. Rock Solid.

Stuart Carlton 38:38
They also have this oatmeal cookie anyway. Yeah,

Carolyn Foley 38:40
we're going way, way off.

Stuart Carlton 38:41
Now. I know. I'm starting to get it and we're starting to lean into what the podcasts really be about, which is cookies.