Making Wisconsin Weather Forecasts Better

Tim Wagner

– So, welcome to the UW Space Place for our monthly guest speaker. Tonight’s speaker is Dr. Tim Wagner. He’s a researcher with the Space Science and Engineering Center here on the UW-Madison campus, and he’s going to be telling us how they are working to make Wisconsin weather forecasts better. So, I’ll give it to you, Tim.

– Great, thank you very much for coming out tonight, and I think we’re gonna have some fun. I’m Tim Wagner. I am a meteorologist by trade. That means that, yeah, I do look at weather stuff. It’s been my dream since I was a little kid to be meteorologist.

And I got to college, and I very quickly realized that I was a terrible weather forecaster, and I needed to find a new trajectory for my life quickly. And I found remote sensing, and it’s something that’s fascinated me ever since. And so I feel good that I’m contributing to society not by predicting the weather but by helping other people predict the weather better.

So we’re gonna talk about forecasting some. We’re gonna talk about how observations feed into those forecasts and how we at the University of Wisconsin are actually trying to make those forecasts better. And you may be thinking, yes, those forecasts, they’re not that great that great right now, and that’s not necessarily true.

We remember the bad forecasts, right? Nobody ever gathers around the water cooler at work and says, “Man, last night, Charlie Shortino “said it was going to be 86 and partly cloudy today, “and he was right.” You know, we don’t remember those. We only remember the cases in which things go bad, right?

But even when that happens, usually it’s a matter of the meteorologists were off by maybe 20 miles or something like that. True, honest-to-goodness, large-scale forecast busts are rare and getting rarer. Now what you’re looking at here on the screen is a part of what we call forecast skill, and it’s how we measure how good our forecasts are. And, surprisingly, accuracy is not the best way to determine how good a forecast is, and that’s because many times accuracy, well, it’s telling you something you already know.

Take for example. Here in Madison on average it snows about 38 days a year, so I can make a prediction that tomorrow it is not going to snow. Well, it’s an accurate forecast, right? But it’s August, it doesn’t snow. But even if I said every single day of the year that it’s not going to snow, I’m going to right almost 90% of the time. That’s accurate, but that doesn’t really even tell you anything all that useful.

Even in January, which is our snowiest month, it snows about 10 days during January, even if I say it’s not going to snow, I’m going to be right over two-thirds of the time. So, instead, what we look at is something called forecast skill, which is basically how good is a forecaster at telling you something you don’t know, predicting something beyond the average ’cause it’s easy to predict the average and on average the average is right, correct?

So the forecast skill is sort of getting at that above and beyond, telling you something you don’t know. And what you’re looking at here is a plot of forecast skill over time. The blue line represents 36-hour forecasts, so a day and a half, and the red line represents 72-hour forecasts, or three days. As you can see here, there’s this steady increase in forecast skill over time. The forecasts are getting better and better, and this basically goes from a scale of zero to 100. 100% would be absolute perfection, and we’re probably never gonna get there, but we’re getting close, and you can see that, basically, there’s this 15-year lag in between the day and a half and the three-day forecasts. What this means is that, basically, the forecasts that we have today going three days out are as accurate as the forecasts 15 years ago were for a day and a half, so we keep getting better and better. Forecasts are pretty good now, but, you know, we’re not satisfied with pretty good.

We want our forecasts to be great, and so what we’re doing at UW-Madison is working to find ways to make those forecasts better. Now, a quick discussion on what goes into a forecast ’cause it’s not an easy process. What a good forecaster does is they basically start by just looking around the country and seeing what’s going on.

What are the temperatures like in Texas right now? Where are the big wind patterns? All of those things, you look at them. You look at the radar and satellite imagery to see where the major cloud systems are, the major storm systems are. You look at what your computer models are telling you, and you don’t follow them blindly, but you look at what they’re telling you, and then you rely on your intuition and your experience, and that’s really where the humans bring value to the process. Forecasting is almost an art as much as it is a science. A good forecaster is able to look at all of these different data sources and weave them all together into a coherent story.

Sometimes they’re telling you, the different data sources are telling you completely different things, and the good forecasters are able to discern what the thread of what’s actually going on there is and give you a nice, skillful forecast because of it. Now, of course, one of the main ingredients that goes into anybody’s forecasts are these computer models.

And this is just an example of one output from two different computer models ’cause each of these models have different ways of assessing what’s going on and predicting what’s going to happen next. I grabbed these charts just about 48 hours ago, so two days ago, and you’re looking here at the accumulated precipitation over a three-hour period. That’s basically happening right now.

And I realized that, for those of you in the room, the colors on this chart are a little hard to see, but, basically, you’re looking at the United States here. And you notice that there’s this nice band of rain basically following along the East Coast, and you’ve got some little pop-up showers here over the Four Corners region, and this other computer model is telling you the same basic idea.

If you drilled down into it, there’s some differences. You see that out over the eastern portion of North Carolina the precipitation is a little more intense, the center of circulation for this rain event is a little further north into, I guess that would be Quebec. There’s some subtle differences because, you know, no two systems are gonna forecast things identically.

And if you look at where the rain is right now, well, there’s some broad agreement with what’s going on. Remember, these images were created two days ago, and yet they’re doing a pretty decent job of assessing what’s happening. They’re not a perfect, and a good forecaster would be able to look at the trends in the different computer simulations of the atmosphere and pick out the one that’s more likely to be true.

I am not a good forecaster, and so I’m not really good at doing that at all, and I admire those who are. Let’s take a look, then, how these computer models actually work. Fundamentally, the atmosphere obeys the standard laws of physics just like everything else in this planet. So, basic laws. Hot air rises, cold air sinks, all right? Force is mass times acceleration. You know, if you hit something and then you hit it harder, well, you definitely feel more force because that fist is applying at a different acceleration once it hits your hand.

What goes up must come down. All these Newton’s laws of motion apply to the atmosphere just as they do to baseballs or anything else in the planet. The one thing is we live on a particularly difficult planet in that it rotates, in that the Earth is heated at different levels, right? The equator gets more energy than the poles do. The surface of the Earth isn’t uniform. There’s oceans over two-thirds of it. I mean, there’s all sorts of things.

So if you were to actually look at the equations of motion for our atmosphere, oh, boy, they’re ugly, and students recoil in fear whenever they see these things. Fundamentally, what these things are saying is they’re diagnosing the trend in the atmosphere, so if you know what the atmosphere is doing right now, these equations tell you what it’s going to do next. Notice that key thing.

For this to work, you have to know what the atmosphere is doing right now. Okay, that’s a lot of ugly math, so let’s get that off the screen. The first person to really figure out that you could do something with this was a brilliant scientist named Lewis Fry Richardson from the United Kingdom. He was a very interesting guy. He was an ardent pacifist, refused to participate in World War I, so, instead, he was an ambulance driver and shuttled the wounded away from the front lines towards the hospital. But he discovered all sorts of things involving the atmosphere, involving fluid flow.

He even was a bit of a poet and composed a poem about molecular friction, which I think is just totally awesome. But he had this idea that since the atmosphere obeys these basic laws of motion that you could actually calculate what the atmosphere is going to do next.

So he had this idea that he would look at some observations from Europe, and, basically, he would go six months in the past. He found a particular day that he thought was interesting, and he thought, “If I take these equations of motion “and I do all this ugly math “and I solve everything that’s supposed to happen, “then I can determine what the atmosphere “was going to do next.

“And since I’m looking at some day in the past, “I can just look up what the right answer was “to see if this technique works.” So what he did is he basically divvied up Europe into a bunch of boxes, and you see his forecast grid on there is basically looking over Germany. And he said, “Okay, I’m gonna take all these observations “from all these different locations, “I’m gonna put them into these boxes, “and if I solve these equations “inside each of these boxes, “then eventually they’re in the middle. “I’m going to figure out what the right answer was.”

And, in fact, he had this brilliant idea that he would build this giant spherical room with a globe painted on it, and there would be hundreds and thousands of people calculating what the atmosphere was supposed to do. They’d be located in these, basically, like theater boxes on this giant room. They’d be doing their calculations, representing the point on the globe where they were. Then they would hand those off to the box next to them ’cause, of course, wind blows everything around, right?

All of this information in the atmosphere is being transported all over the place. And so he had this idea that this is how we can actually predict what the atmosphere is going to do next.

So, going back to our buddy Richardson here, he did all these calculations for this particular point, and he was off by only 10,000%. He made some mistakes when it came to the process of taking the data and putting it into the boxes initially. It’s so key, getting the right data in the boxes, because if you don’t do that correctly, then everything goes awry.

And he recognized that’s what it was, but he knew that in the era in which he lived, basically, this dream of having numerical weather forecasts following the laws of physics was probably gonna be impossible. But he lived in an era without computers, and it wasn’t too long, maybe just 20 or 30 years after Richardson did his experiment, that computers came into play.

And one of the very first problems that the very first computer was used to work on was weather forecasting because Richardson’s basic idea was sound. You take the atmosphere, you divvy it up into a bunch of boxes, you solve the equations of motion inside each one of those boxes, and you transport the information from one box to the next, and you know what’s going on.

So this is ENIAC, which is considered to be the first real digital, programmable computer. It was developed by the military shortly after World War II, and one of the first problems that they looked at was weather prediction. I want you to notice something interesting about all the operators of this computer.

What do you see that’s the same about them?

– [Audience] They’re all women.

– [Voiceover] They’re all women, absolutely, because in the late 40s and early 50s computer programming was considered clerical work, stuff that the secretaries did. Basically, the bosses said, “Okay, we’re gonna do this weather forecasting problem, “and here’s kind of how we’re gonna do it. “Okay, secretaries, you go make it work.” These women were the first computer programmers.

They were basically inventing computer science as they went along, and they really did not get the recognition that they deserved.

Well, ENIAC did correctly predict the weather. They didn’t have a whole lot of computing power to work with. ENIAC did calculations, about 385 calculations per second.

By comparison, this little device from Apple that I carry in my pocket does billions of calculations a second. ENIAC was the size of this room that you’re sitting in right now, and rumor has it that every time they turned it on, lights nearby would dim because it was drawing so much electricity.

Well, the other thing, ENIAC did do it correctly, but they had one issue. They made a forecast for 24 hours into the future, and at 385 calculations a second it took them 23 hours to calculate that 24-hour forecast. So, congratulations, here’s your output, and here’s what the atmosphere’s gonna be doing in an hour. It wasn’t that useful, but they were excited because they were on the right track.

They knew that the science was sound. They were getting results that were accurate, and they thought, “In the future, “we’ll be able to get forecasts. “It will only take 12 hours to do it. “Won’t that be awesome?” Well, you keep going forward in time, and we basically, this basic idea is fundamentally underlying everything that we’re doing today with weather forecasting.

Now, when we look at these boxes, they aren’t just on the surface of the planet. Of course, you have to look at it in three dimensions as well because the atmosphere changes as you go higher and higher up. So here’s a little just cross section view, kinda what’s going on.

Inside those boxes, the weather condition inside a particular box, we say everywhere in that box is identical. So you get some sort of surface temperature, and you start putting temperatures in the boxes above it, and then you do the calculations.

You can’t solve every single point independently because the Earth is just too huge. We have a limited supply of computing power, and so what we do is we basically divvy up the atmosphere into boxes, and that’s why we do that.

Today’s computers attacking this problem are just amazing. This is the computer Yellowstone, which is operated by the National Center of Atmospheric Research. Remember, ENIAC did 385 calculations per second. Yellowstone does 1.5 quadrillion calculations a second. It’s just an astonishing machine.

Most of the major supercomputers in the world are devoted towards weather or climate calculations because it’s such a difficult problem. The basic science behind it is simple. You saw those equations.

But trying to solve that for so many points on the surface of the Earth, so many points through the depth of the atmosphere for so long into the future, assimilating thousands upon thousands upon thousands of observations, it becomes very computationally complex very quickly.

Now, go back to when I was just getting started in the field, and this was state-of-the-art. Each box on this map that you’re seeing here represents the size of one of our model grids in the past. These were 80 kilometers, or about 50 miles, across. And when it comes time to getting the model up and running, this was a pretty simple process.

I mean, basically, look at Madison here, where we are. There’s an airport here. It’s taking temperature and moisture and wind observations, so our computer basically just takes all that data and says, “Okay, everywhere in this box, “those are the conditions.”

And then over here to the east, well, we’ll look at Oconomowoc and Sullivan, whatever points are in there. And it’s pretty easy to fill up these boxes, but the thing is, when your boxes are so big, your errors start accumulating over time because you know that it’s rare that the temperature in Janesville and Madison and Columbus and everywhere else inside this box is gonna be identical, right?

There’s small variations. It’s those variations that accumulate over time that cause the forecast to go wrong, so one of the things that you can do to make your forecast more accurate is shrink the boxes.

Now, that greatly increases the number of calculations that you have to make, but computers keep getting more and more powerful, so over time it’s not gonna be that big of a deal to make our boxes smaller.

Now, you see there in the upper right-hand corner towards Milwaukee, the size of the old box. These are the size of our boxes today, 12 kilometers, about seven or eight miles, substantially smaller than they were in the past. Now we’ve got some issues because earlier we had really big boxes that we could easily average a few different data points that were located inside there and come up with a nice average starting point for temperature and wind and everything else to start our calculations from.

But, you know, now there’s plenty of boxes that don’t have any observations. I mean, does in between Monroe and New Glarus and Monticello, I mean, there’s not a lot of weather observing stations there. You go off into the driftless zone, again, not lots of observation stations.

And so you’ve got plenty of places that don’t have any data to initialize, so what you have to do is you have to go through what we call data assimilation, bringing in the data and then smoothing it out so that each box has an initial point that you can start your calculations with.

Well, there’s plenty of locations that aren’t having any observations. And if we, in the past 20 years, basically cut our grid down by a factor of seven or so, in the next 20 years, we may be seeing two-kilometer boxes, a mile and a half or so. All right, well, there’s plenty of points there that aren’t going to have any observations in them. So what we need is we need to find ways to get better data, get more data to get these models up and running.

For decades, the limiting factor in forecast accuracy was the fact that our computers were just too slow, and we just had to run them at coarse resolutions so that they could run in time to get everything done. Well, our computers are getting faster and faster, but the amount of data that we have available to them from the ground, it’s stayed the same. We’re not building more airports with more surface observing stations, so we’ve got to find some other ways to get our data in there.

Now… When you, any time you’ve been past the airport and maybe you see in between taxiways or something like that, your typical surface observing station, this is what it looks like, and it’s got the standard observations that you associate with weather, temperature, humidity, wind speed, all those things. Now, one of the issues that we have is where those observations are coming from globally.

What I want you to do first is look at this map and just ignore the colors and just notice where the dots are. The dots represent observing stations, and you see right away that you basically don’t have any observations over the ocean. 70% of our planet we know nothing about. Antarctica, very few observations there. The Amazonian Basin, very few observations there. Sahara Desert, Outback, Tibet.

There’s plenty of places that don’t have lots of observations. Many of those places have very few people. Other times, there’s political strife or things like that associated with it. Now, what these colors represent represent how many observations these stations are actually producing.

This was a study done over a two-week period back in 2008, and all of these stations were supposed to be reporting at least four times a day. The blue stations produced at least 90% of the observations that they were supposed to. The green stations were 45 to 90%. The yellow stations were less than 45, and the red stations didn’t produce anything.

Well, now you see that, basically, the highest quality data, the stations that are producing the most data, coming from the United States, Western Europe, Russia, China, India, Australia, South Africa, and Argentina.

Nations that are relatively wealthy, right? Basically, the poorer your country is, the less likely you are to have weather observing stations to begin with, and even if you do have those stations, many of them aren’t working. I mean, look here in the middle of Africa. The Democratic Republic of the Congo has weather stations, but not a single one of them is producing any data.

Same with Iraq, or that’s probably Iran.

Same with Afghanistan.

Locations that are undergoing significant amounts of war aren’t producing any weather data. We are somehow supposed to be able to make accurate forecasts of what’s going on when only 30% of our planet can even support observations to begin with because they’ve got land, and only a small portion of that land is actually producing any good data. I mean, it makes sense.

Weather data, in a way, is a bit of a luxury, right? If you’ve got stability, prosperity, and peace, then you can start worrying about the weather forecast. But if you’ve got war, pestilence, or famine, well, having an up and running anemometer is probably towards the lower end of your needs right now. It’s something you might get to eventually. Well, this is all for surface data.

For the upper levels, things are even worse.

Our backbone of knowing what’s going on through the depth of the atmosphere remains the weather balloon. This device, you see it there, is called a radiosonde, and it’s, maybe, about the size of a box of tea. It’s got that sort of size and shape, so it’s only about six or eight inches across and kind of blocky like that, so pretty small, pretty lightweight, and it can measure through the depth of the atmosphere as high as that balloon goes. It measures the temperature and the humidity and the wind speed and direction and the pressure of where it is.

These observations are critically important because they’re telling us what’s going on through the depth of the atmosphere. If I stick a thermometer outside, that’s only telling me the temperature of the surface, but there’s 50 miles of atmosphere above me that I need to know something about, so that’s where weather balloons come into play.

Problem is weather balloons are expensive. Because you send this thing up on a balloon, it’s highly unlikely that you ever get it back, and so these aren’t cheap devices. Each one of them costs maybe on the order of 200 or 250 dollars You throw in the cost of the balloon itself, the cost of the helium to lift the balloon, helium is not cheap, but hydrogen has the unfortunate downside of being kind of explodey. You put those things together, these are not cheap.

And so you look at the locations where these observations are coming from. Our knowledge of the depth of the atmosphere is pretty much restricted to the United States, Europe, Russia, China, India. There are giant swaths of the world that we really don’t know anything about from direct measurements.

So what does that mean? Well, we’ve got a huge amount of atmosphere. Only a small portion of it is being observed directly, and so we need to find ways to fill in those gaps.

And one of the ways we fill in those gaps is with satellites. Weather satellites are critical to all of this. In fact, about 90% of the observations that go into one of these weather forecast models come from satellites these days. Now, we know satellites, of course, from the evening news and the pretty pictures that they produce.

This is the recent super typhoon. I’m not gonna try to pronounce its name because I will fail at it. But it recently went ashore at Taiwan. Latest death toll was 20, so it’s not great, but it could’ve been a whole lot worse.

Satellites help us keep eyes on these things, right? How else are we gonna see what’s going on over the Pacific Ocean when there’s nobody out there to see it? But satellites do more than just give us pretty pictures. We can get actual quantitative data out of them as well.

This, on the upper left-hand corner, is a plot of sea surface temperature anomaly, or departure from average. You notice, coming off of the coast of Peru there off of South America, nice, warm tongue of temperature.

This is, El Nino is back, and that’s gonna have all sorts of fun impacts on weather conditions around the globe. You can also get some temperature profiles from satellites. This is a recent example of a temperature profile from a satellite right over Madison.

And here’s an example of, remember, this is about five or six years ago now, that big, Icelandic volcano that erupted and stranded all sorts of people on either side of the Atlantic ’cause no airplane flights could make it through. Well, we can actually use satellites to diagnose where the ash plume of that is and track it around.

There’s all sorts of great things that we can get from satellites, including, like these sea surface temperatures, actual honest-to-goodness numbers that we can then put into our models, and the process by that works. It’s fairly simple. It’s all about measuring radiation. So consider your standard old-school incandescent light bulb. You screw that thing into a light socket, you turn it on, and light comes out.

Well, the reason light comes out is because you heated that thing up, that little filament in there, you put so much current in through there that it got so hot that it started glowing. And on the right there you see what an infrared camera sees when it looks at a light bulb. It’s producing tons of heat. A light bulb glows because it’s hot, and that’s why we’re trying to get rid of these things ’cause they’re extremely inefficient. We produce light from heat. So much of that energy is going into the heat, and relatively little of it is going into the light.

That’s why we’ve made the change to incandescent and LED light bulbs, which use different processes to produce their light and are much more efficient. Well, it turns out that, just as you could look at a light bulb and basically have it tell you what temperature it is, it can do that with the atmosphere as well.

And in the upper right there, you see just a plot of looking at the air using a satellite measuring radiation at a bunch of different wavelengths of the electromagnetic spectrum. And you see things like, oh, there’s some times where it’s nice and high and other regions, other wavelengths of energy, very specific wavelengths or frequencies of energy that you see it’s nice and low.

You know, this electromagnetic spectrum, it’s all electromagnetic energy. X-rays, gamma rays, visible light, infrared, ultraviolet, microwaves, radio and television, all are the same type of energy just different frequencies. So, if you tuned your radio down, your radio goes basically down to, like, 87.7 megahertz. Well, keep tuning it further and further and further down, you’re going to get into the infrared and into the visible and into the ultraviolet.

Now, nobody makes a radio that does that, of course, but it’s the same basic energy at play. Now, it turns out that as you look at specific frequencies, specific wavelengths of energy, they absorb different amounts depending on what’s in our atmosphere.

Here’s just some of the trace gases in our atmosphere and where they’re very good at absorbing. And if you actually compare these locations, you see, for example, there you have this ozone, big amount of absorption due to ozone. It’s located right there, we see this.

So what we can do is by basically looking at the energy that’s coming off of the Earth and seeing where it’s high and seeing where it’s low, we can come up with not only the temperature of the surface but the temperature of everything that’s in between the surface and the satellite.

It’s not a simple problem. One way to think of it is basically looking at some footprints in the snow some evening and figuring out what was the color of the dog that produced those footprints. You think, “That’s impossible.”

Well, it’s not quite impossible because you could see how deep the footprints sunk into the snow, so you know something about the weight of the dog, and you can see the shape of the actual toes, and you know that might correlate to breed, and so there’d be some ways to get at some of the information that you’re looking at. But it’s that kind of idea.

We’re looking at some of the basic signatures of something and trying to figure out how to turn that into actual useful information. It’s not simple, but it can be done. Now, we here at the University of Wisconsin basically invented the weather satellite.

Up here, this individual is Vern Suomi, who was a professor at the UW and basically came up with the idea of sticking a satellite way out away from the Earth’s surface so that it would be in geostationary orbit and basically follow the Earth as it spun and getting actual, quantitative information out of it. Nobody had done this before.

Now, it wasn’t always easy. Here’s a picture from the front page of the Wisconsin State Journal in 1959, and this was the second time that one of Professor Suomi’s satellites exploded shortly after launch. This time it was a bad navigation module, and so the army, there was not really a NASA yet, shut it down and just blew it up so it didn’t cause any damage.

But Suomi said he was gonna pick up the pieces and try again, and the third time was the charm. He got his satellite up there and made the first measurements of energy coming away from the Earth’s surface.

Nobody had done that before, and it’s just so fundamental to our understanding of how weather and climate works, and so we’re proud, basically, to say that we are the first institution to look at weather satellites, and today Wisconsin remains a world leader in not just satellites themselves but other instruments in visualization, in outreach, in observations around the world.

Some of the finest Antarctic meteorology studies are going on here on our campus. So we’ve got so much going on in the world of satellites. So we’re really, really proud of what satellites can do.

But they’re not the solution to everything.

Satellites have issues. One, they’re difficult to calibrate. So once you stick a satellite up there, there’s not much that you can do about it, right? You’re not gonna go send an astronaut 20,000 miles away from Earth’s surface to go fix it, so if something goes bad on it, either its detection ability or just some mechanical system, well, you’re done, and you just need to spend another billion dollars to get another one up there.

Because satellites are so far away from the Earth’s surface, they don’t see the surface all that well, and the surface is where all the interesting stuff happens. Satellites are really good at looking at the top of the atmosphere, but as you go further and further into the atmosphere, they’re not as good as figuring out what’s going on. And because they’re so far away from the Earth’s surface, they can’t pinpoint in very tiny locations. Their footprints, as we say, the spaces that they view, are going to be on the order of miles across instead of nice little pinpoint observations that we could use to stick into our really tiny boxes.

And, fundamentally, they’re expensive, right? To get a satellite up there, to pay NASA or some private contractor to send it up there, to hope that everything works and that it’s not gonna break because there’s nothing you can do about it if it does.

Overall, that’s going to cost you a ton of money, so we can’t just use satellites. We need stuff down at the surface as well. Now, some of the things that we’re doing at SSEC are some novel ways of coming up with additional weather data. Getting this depth of the atmosphere stuff is so critically important that one thing that we could do is start using observations that people would be taking normally.

You see here an airplane, this is a Southwest 737. And through work that we’re doing with the National Oceanic and Atmospheric Administration, these airplanes, some of them, have been equipped with temperature and moisture sensors. The temperature sensor, that’s nothing new.

Airplanes have been taking temperature observations for decades now, but the moisture, it’s a trickier thing to measure, and we’ve been doing some work on how to actually measure that well. Well, think of it. A Southwest airplane goes up and down maybe five times a day.

It flies a lot of short, hot flights around the country. Up there in the right-hand side of the screen, you see just a map of these planes as they’re going along. The brighter colors mean that they’re closer to the ground. Well, this airplane, it takes off and it lands, and it takes off and it lands.

Well, if it’s measuring temperature and moisture and altitude as it does that, we’ve got the same information that a radiosonde, a weather balloon is telling you, but it’s giving it to you much more frequently at a much lower cost, so one of the things that I’m working on right now is basically showing how useful these things are.

We’re gonna take this data, we’re gonna plug it in to one of out computer models and just see how much more skill our computer model has because of it. I think it’s really fascianting work, and I think it’s gonna make a big difference with very little expense to the taxpayers. We’re also developing other machines that look up at the ground. These machines are called AERIS, atmospheric emitted radiance interferometers. We’ll see another one in a little bit.

But this was a machine that was invented here at Wisconsin that is able to measure the profile of the temperature and the moisture from the surface, and it does it every 30 seconds.

Weather balloons, because they’re so expensive, we only send them up once every 12 hours. So think of how many 100s of observations a day we’re gonna get from this machine, and it’s nice and accurate. We’ve shown great agreement with our radiosonde balloons, and we’re working right now on demonstrating to the weather service how useful a network of these would be.

Now, we really are interested, fundamentally, in finding new ways to observe our atmosphere and not just finding those new ways but making sure that they’re useful, getting that information out to the general public, out to the rest of the weather community so that they can utilize it and make forecasts that make everyone’s lives better. And so, to that end, we developed a couple of new sensing systems we’re just sort of putting together.

First is we’re gonna create a really robust observation site on the roof of our building. For those of you who’ve hung around Madison, you’re probably familiar with our building. The Atmospheric, Oceanic and Space Science building is the building that has all of those satellite dishes and antennas and everything. It’s about two blocks east of Camp Randall Stadium.

And we are proud to say that, discounting the capitol, it is, I believe, the tallest building in the city. Now, Van Hise Hall on the UW campus has a taller roof, but it’s sitting on top of Bascom Hill, and that’s cheating. We’re starting off, like, 150 feet lower than they are, so the fact that they’re taller doesn’t really count. But you see the view from our building is fantastic. We like to go up there and grill some brats every now and then.

But what we’re doing right now is we’re introducing a whole new network of systems that are going to help us ascertain what the atmosphere is doing with a variety of different things. We’ve already got some systems on the roof capable of measuring winds, capable of measuring what the current temperature and humidity and everything is. This device measures how high the clouds are, but we’re also gonna put a temperature and moisture profiler on there. We’re gonna put radiometers that tell us basically how much gunk is in the air, how clean the air is on a particular day. We’re gonna install a camera up there that allows us to see 360-degree view, whole hemisphere of what’s going on with the sky.

We can actually do some cool things with that. That’s just not pretty pictures because we can analyze that picture and detect where the clouds are and where the sky is clear and from that come up with a good measurement of how cloudy the sky is. And, in fact, right now I’m working with a colleague of mine out in Oregon to detect specific cloud types.

It’s not just one thing to see that it’s cloudy or not but another thing to see is that a cumulus cloud, is that a stratus cloud, is that a cirrus cloud. It’s an immensely difficult problem, but, boy, is it fun. I really enjoy it.

So we are going to be putting these instruments all together over the next year or so, and we are going to then have the data from these available online so that you can go see what the air is like over Madison, not just the temperature and moisture at the surface but through the depth of the atmosphere, how the characteristics of the atmosphere are changing from day to day. It’s gonna be really cool. We’re really excited about that. Something else that we’re really excited about is our new mobile laboratory, what we’re calling SPARC, the SSEC Portable Atmospheric Research Center.

Because it’s one thing to have all these instruments on the roof in Madison, but we get certain weather types here, but our instruments that are sitting on our roof are never gonna tell us anything about a hurricane. They’re never gonna tell us anything about wind flow through mountains. They’re never gonna tell us anything about the tropics.

So what we need to do is we need to come up with a way to go learn more things about these phenomenon, so we need to find a way to take our instruments on the road, and that’s what SPARC is. You see, our observations, they’re not just useful for forecasting. We need to get more observations just so we understand how the atmosphere works. Those big, nasty equations that I put up at the beginning of the talk, well, those work to a point for certain phenomena, but we really need more data just so we can understand physical processes.

Why does something behave this way and something else behave another way? There’s so much about the atmosphere that we don’t know, and our existing data network isn’t enough to tell us that, so what we need to do is we need to go out into the field and take our instruments with us and see what’s going on. So SPARC has been built basically from the ground up as a new tool for us to go and actually investigate what’s going on.

This is a picture, on the left, from a field project that we participated in this summer. We’ll talk more about it in just a second. On the inside, we have all these instruments plus we’ve got nice working space for scientists to monitor the instruments, keep track of what’s going on.

SPARC has everything that we need to see what’s going on at the surface. This little doohickey is a prop vane, which measures wind speed and direction. You see, it basically looks like an airplane without wings. Well, that propeller there, as the wind blows through it, the faster the wind is, the faster it spins. You detect how fast it spins, and you know what the wind speed is, and that big tail causes it to always be pointed into the wind.

You’ve got temperature and moisture observations, so you’ve got a nice, complete, pressure’s in there too, so you’ve got a nice, complete view of what’s going on down at the surface.

But we’re not just interested in the surface, so we launch weather balloons as well. In fact, we’ve got a chamber built into our trailer there just to hold the helium tanks so that we can fill up the balloons and let them go. And on the right there, you see what a typical plot of what a weather balloon looks like. It’s really, at first glance, it might be hard to read, but, basically, this red line is temperature, this blue line is dew point, which is a measure of humidity, and you’re seeing this from the surface all the way up to 100 millibars.

This is probably 40 or 50,000 feet above the surface, so a nice big depth of the atmosphere. As this balloon rises, it’s being blown around by the wind, right? There’s a GPS tracker in the balloon. And so if we know how far away that balloon is being pushed, well, we know something about the wind speeds. So, on the right here, you see wind speed and direction. These little bars point to the direction that the wind is coming from, and the more flags are on it, the faster the wind speed is.

So with this weather balloon, we’ve got a nice view of what’s going on. Of course, we don’t just want to launch weather balloons ’cause even for us, one, they’re expensive. Two, they take a long time. It takes an hour and a half for them to get all the way to the top of the atmosphere.

Well, a lot can change over that time. So we’ve got some good remote sensing instruments as well. We have what are known as lidars. You’re probably familiar with a radar. You see it on the TV news all the time. Well, the word radar stands for Radio Detection and Ranging.

Well, a lidar is a radar, but instead of using radio waves, it uses light waves. And so maybe sometimes, if you’re driving through downtown Madison, you’ll occasionally see a big green beam coming out of the top of our building.

Well, that’s one of our lidars, and here’s what one of our lidars looks like up close. Basically, it sends out beams of light, and those light beams reflect off things and come back a little bit differently than the way they went out, and it’s that change in that light pattern as it returns that tells us something about how the atmosphere is behaving.

So, on the top here, this is the lidar that was producing that nice big green beam. This tells us all sorts of characteristics about the atmosphere and about the clouds that are above it. I think this image is really cool because, look at that, look at that rain that’s coming off of this cloud. This cloud is four kilometers above the surface, so that’s about two and a half miles or so. And just look at how pretty that rain is. I just think that’s gorgeous.

Down at the bottom here, we have a wind-profiling lidar. As the light beams come out of our lidar and run into just gusts and eddies and currents of wind, they get bounced around, and the light comes back a little bit differently. That tells us something about the wind speed and direction. We also have one of these AERIs on it, which gives us a profile of temperature and wind every 30 seconds. So we’ve got all these great observing tools, and we’re already putting them to good use.

This summer, SPARC participated in a field experiment called PECAN, or Plains Elevated Convection at Night, and I got to spend three weeks in Kansas chasing storms, basically, around the countryside. Over the six-week period that this project was going on, we left our GPS on just to track everywhere that we went, and you see that our system basically went everywhere between Oklahoma and South Dakota and all the way east to Waterloo, Iowa.

There was some debate, like, well, shouldn’t we just go home for the night rather than going back to Kansas? Notice how close we were then. But for this project, SPARC, along with many other instruments, there were over 100 meteorologists out in the Great Plains this summer chasing after nighttime storms.

This is a good example of how we need to go find the data to learn more about the processes that create the weather we don’t understand very well. Because we understand daytime storms very easily. Sun comes up, sun heats the ground, ground heats the air, hot air rises, takes moisture with it, gets to the upper atmosphere, forms clouds, storms, rain. Easy.

But all of this began with the sun. Once the sun goes down, we see these processes. We see clouds and storms forming, but we don’t have a good handle as to why that’s happening. So the federal government decided this was an important enough problem that it needed to operate a field project and send all of these people, all these machines. We had mobile radars, we had systems similiar to ours, three different airplanes buzzing around, cars with instruments on the roof driving around chasing after all of these storms at night this summer so that we could understand the processes more. Because now that we’ve got all this data, we can start to analyze it.

We can figure out why it is that these processes behave the way they do. We can use that information to feed it back into our computer models, and we can use that to make better forecasts. And it’s not just better forecasts for the sake of, you know, is it nice to have a picnic today or tomorrow or anything like that.

But bad forecasts have a real economic cost because every time an airliner gets grounded or a train gets delayed or shipping companies, trucks, have to reallocate their resources around weather, these have real, economic costs.

And better forecasts allow us to make sure that we’re not spending money unnecessarily, so we are committed to making sure that the observations that we’re taking with our new systems feed into the cycle of producing better forecasts. They’re gonna have a better outcome for everyone.

So to sort of wrap up what we’ve been talking about. Right now what we’re doing, we’re developing new instruments. We’re finding new ways to take data that we’ve already collected and teasing out new data sources from it, observing our atmosphere in ways that we haven’t before, putting that information into new computer models, and then sharing our results with the National Weather Service and just the general public because it doesn’t do us any good just do research if we don’t share the results because we’re not gonna be great at getting that information, that new knowledge, out and causing things to be better for others. But where we’re going in the future is probably gonna be even more exciting because we’re gonna have this new suite of rooftop observations.

We’ve got this new satellite called GOES-R. It’s the new generation of geostationary satellites. It’s finally gonna be launched in 2016. It’s been pushed back and pushed back and pushed back, but we really think it’s gonna happen now. This is going to be revolutionary. It’s gonna be four times better than our existing series of satellites, and UW has been at the forefront of developing ways to utilize that data, new algorithms, new processing, new techniques. So we’re gonna be really excited once this thing gets up in the air.

We’re actually building instruments that are physically gonna go in space and other satellites. We’re already drawing up our plans to put SPARC into more field campaigns. And so we think that putting all these things together, you look at this plot that I started off with, the forecast skill.

We think with all the work that we’re doing we’re gonna see these things, these trends, continue. We hope that with all the work that we’re doing, ultimately it’s gonna result in more knowledge of the atmosphere and better forecasts for our end users.

So with that, thank you for coming out tonight, and I would be happy to take any questions that you have.

(audience applause)