Hurricane forecasts and warnings

[00:00:00]

RICK WEISS: Hi, everybody. I’m Rick Weiss. Welcome to SciLine’s media briefing on hurricane forecasts and warnings, where we’re going to span the spectrum from the complexities of mathematical computer forecasting to the complexities of human behavior in response to these kinds of warnings. I want to say, first, that I know it’s hardly news at this point that human-induced climate change is causing hurricanes to grow stronger and more destructive. We’ve all heard about the clear scientific evidence, for example, that, thanks in large part to climate change, hurricanes are producing heavier rain, their storm surges are riding atop higher sea levels, and in many cases, they’re lingering longer over land, causing increased flooding and infrastructure destruction.

We also know that more than 90% of the excess heat trapped in the climate system from human-caused global warming has gone into the oceans, providing the added energy that’s driving recent hurricanes’ extreme wind intensities and also contributing to the evaporation that, in turn, has been leading to record-breaking torrential rainfall. We know that, globally, the last few decades have seen a growing proportion of strong hurricanes and a corresponding shrinking proportion of weak ones.

And we know that, globally, hurricanes are reaching their maximum intensities further from the tropics, shifting toward temperate, heavily populated coastal regions that have not historically experienced them. They’ve shifted, actually, northward by about a hundred miles in just the past 30 years. So these and other fun facts – or not-so-fun facts can be found on the SciLine website, sciline.org, under Fact Sheets. I encourage you to look at our quick facts on hurricanes in particular to spend a few more minutes combing through these resources, which are all, by the way, vetted by outside experts and heavily footnoted, so you can trust where the scientific evidence is coming from, answering the question that we ask ourselves all the time which is, how do we know that? How do you know that?

But we’re here today to go beyond the story of climate change and hurricanes and look at a few hurricane science-related challenges that have been, I think, a little bit less closely covered but we think are deserving of attention given the likelihood that strong hurricanes are going to be a bigger part of our future. And specifically, we’re going to delve into three very interesting and important areas of research.

First, we’re going to hear from Dr. Joannes Westerink. He’s a professor of computational science and engineering at the University of Notre Dame. He’s going to tell us about some of the challenges and advances in forecasting hurricanes and storm surges, something we all pay attention to with almost addictive attention when storms are on the move, and it’s all nicely packaged when we see those forecasts. But it turns out there’s a whole hairball of uncertainties that forecasters are wrestling with in this process. So we’re going to hear about what’s going on there and how the work is going on to make this better.

Next we’re going to hear from Dr. Scott Weaver, the director of the National Windstorm Impact Reduction Program at the National Institute of Standards and Technology, or NIST, who’s going to talk about the surprising difficulty of getting accurate measurements of rain and wind at the level of our streets and around our buildings, where it really matters, and what all that means as we try to engineer more resilient communities.

And last we’re going to hear from Dr. Rebecca Morss, who’s a senior scientist and director of the Mesoscale and Microscale Meteorology Laboratory at NCAR, the National Center for Atmospheric Research in Boulder. She’s going to take us to the intersection of hurricanes and human behavior with a look at the science of risk communication and efforts to reduce vulnerability among people who are threatened by these storms and other natural hazards. Just before we start, I’m burying the lede here, but we are SciLine, a free, independent, philanthropically supported service for reporters covering science, health and the environment. We offer a wide variety of services for you reporters. Everything is free, and we encourage you, of course, to go to our website and see what else we do besides these media briefings. So with that, let’s get started right off the bat with Dr. Westerink, please. It’s all yours.

[00:04:15]

JOANNES WESTERINK: Thanks, Rick. Let me share here. All right, do you guys see my presentation?

RICK WEISS: Yeah.

JOANNES WESTERINK: OK. Well, thank you for the introduction. And first of all, give – I’m going to give a overview of where coastal storm surge and flood models are going, how they’re evolving. And, of course, this is in tandem with the enormous continued growth in computer power, enormous growth in the data that’s available to describe the geophysical systems, and a strong growth in the paradigms and algorithms that are used to turn processes into computer codes. So what I’d like to give an overview to kind of summarize this evolutionary process is two things that I’d like to focus on.

First of all is computational domains and meshes. And what you see to the left of you is this chunk of the Earth’s system, and it’s the western Atlantic. And what you see there in blue is the deep ocean. The orange is continental shelf. It turns out storm surges amplify in continental shelves. And then you see, in Louisiana here, a lot more detail. And so hurricanes are very large-scale events, and so we tend to be building models that are taking larger and larger chunks of the ocean. In fact, we’re now developing a whole new generation of models that are based on the global ocean, and we use those as our – really, our framework. We then start providing resolution in the areas that are important.

By the way, I should highlight that old models only had a very small part of the ocean because they simply didn’t have the compute power to do anything more. And then, over to your right is the mesh that supports all that data, the bathymetry, the topography, the roughness of the surface, etc. And these are what are called finite-element meshes. They’re a very ubiquitous engineering technology. And more and more computational codes that solve equations are going over to what are called unstructured meshes. And the resolution is very, very different as you go from different parts of the domain. All these things are little triangles that you see on your screen. And as you get onto the continental shelf where there’s more complexity, they get finer and finer and finer, and into areas like southern Louisiana, where they’re very fine.

So we’re going from about 20 kilometers to about 30 to 40 meters of resolution in this area where the pointer is. And again, it’s kind of like having a camera that has variable pixilation. So now to look at a little inset of what kind of detail this unified model can have, we’re just going to take a look at southeastern Louisiana. And this is southeastern Louisiana. Of course, it was on the news a lot post-Katrina. You can see the Mississippi River flowing down into its deltaic region. You can see Lake Bourne, the Carnarvon Marsh, Lake Pontchartrain – all connected with enormous detail in the channels, the barrier islands, the wetlands, all the channels within the wetlands, etc. So this is all part of this physical system that we have to describe to the computational code in order for it to resolve the processes and the physics that’s computed. If we now overlay what the computer code sees, it’s this over in here.

So this is the bathymetry – one of the many sets of variables that are held by that computational mesh. And you can see channels. You can see levee systems. You can see the different distributaries of the Mississippi, the barrier island systems, and the topography represented by the colors. So this is really the picture that we give the computer code in order to understand what this physical system is. And resolution is what it’s all about, and that resolution is held by the grid and – or the mesh, and that mesh resolution is variable. So this is the cutting-edge technology that’s becoming ubiquitous in the modeling world in order to better represent this. And again, all the channel systems have very high resolution – between 30 to 50 meters – and the ocean has much coarser resolution.

Now, the second part of the story is really the processes. And the processes are represented by very complex equations which are then turned into computer codes. And there’s, of course, global atmospheric models. There’s ADCIRC. That’s our model. That’s a tide and storm surge model, and then there’s a wind wave energy model. So if we were to look at that, we have the winds that are represented by models that NOAA runs – and a whole sequence of them. And those are used as input to our model. So these winds go over to the ATCIRT model over here, as well as to the wave model. And what we then do is we force those winds and they compute waves, and this represents significant wave height. And that’s on the order of about 8 meters here, and you can see it quickly attenuates.

And storm surge – you can see the evolving storm surge ready up at 2 meters against the levee system. And during the storm, that all evolves, and that is simulated. And you can see this complexity of the landfall process occurring where you have huge waves in excess of 10 meters here. You have storm surge in excess of 3 meters against the levee system. You have the storm surge propagating over the levees. And all this is made possible by the intercoupling of processes and codes – atmospheric codes with storm surge codes with wind wave codes – so those are the actual waves that you can see coming into the shore. And those two processes are talking to each other and translating information – sharing information so that they can both do a better job at computing the physics.

So that’s where state-of-the-art models are now, always improving resolution, always incorporating these processes. But now the question comes about, what are we doing about things like the global ocean? And Rick mentioned that it’s warming. It’s actually absorbing a lot of the heat energy, and therefore, it’s rising. But it’s also driving current systems, like the Gulf Stream here in the background picture. So in order to address that, what we’re doing now is taking big, global ocean circulation models, such as HYCOM, and – in particular, NOAA runs something called GRTAS, and we’re coupling that in with that whole sequence of models by borrowing its temperature and its salinity fields. When you do that, what you start seeing is you start getting gradient fields in the water surface and current system.

So on the top, you can see the gulf stream going here. That’s now coupled with all those other processes. And what it tends to be doing is lowering the coastal water levels. That blue means they’re suppressed – depressed, and red means they’re raised. So this is the kind of information we’re pulling in now into this modeling suite that couples and interfaces with all other models. This, by the way, is quite important in terms of what’s happening to coastal water levels in the southeastern United States, where there’s a lot of nuisance flooding.

Last but not least, we’re now also including a very important process that has been neglected for years and years, and it’s super important for things like Hurricane Harvey, where we really – the modeling didn’t consider the rainfall runoff simultaneous to the coastal surge. So we’re now incorporating whole linkages and interactivity between the coastal surge models and all its combined processes and the upland hydrology models or the rainfall runoff models. And this is important activity going on at NOAA, and we’re helping them with this. So the story, really, is ever-increasing resolution and ever-increasing complexity of the processes and their codes that represent those processes and interlinking and intercommunication.

RICK WEISS: Fantastic. Thank you. Dr. Weaver.

[00:12:49]

SCOTT WEAVER: Good afternoon, everyone. My name is Dr. Scott Weaver. I am here at the National Institute of Standards and Technology. It’s a pleasure to be here with you this afternoon. I wear two hats at NIST. The first one is primarily as a research meteorologist, and then I’m also director for the interagency National Windstorm Impact Reduction Program, and I’ll talk a little bit about that in a couple of slides. But in my first slide – if we could advance to the next slide – I want to highlight that there is a significant problem. So we have a problem. As you just heard, it’s about water and – as well as wind. And so we have this problem with increasing impacts from weather disasters. This is one of my favorite figures to borrow from our sister agency, NOAA. And basically, what you’re looking at here are the increase in the number of billion-dollar disasters that NOAA tracks. So if it hits – if the economic impact reaches a billion dollars, they count it as an event.

There are three takeaway messages from this slide. As I mentioned earlier, wind storms are not only about wind; it’s a problem of wind and water. The number of these events are increasing, as you can plainly see if you look at the bars, and the costs are also increasing, and that’s mostly driven, in the windstorm world, by hurricane impacts. Unpacking this slide a little bit, what you’re really looking at here – if you look at the bars, that’s just the number of events per year from 1980 to 2018. The yellow events are tropical cyclones and hurricanes. The green events is other kinds of severe weather, like tornadoes or other thunderstorm events. And you could plainly see that there is an increase.

If you dig a little deeper, you can actually separate this. There’s a nice separation in the year 2000, where there wasn’t even – there was zero billion-dollar events. If you compare the first half of the record to the second half of the record, the increase in hurricane events exceeding one billion dollars is 33%. So they are increasing, as well as other forms of weather and climate disasters. Interestingly, if you think about this from the economic perspective in terms of total costs, if you look at the gray line – I don’t have a pointer, but you can see there’s the gray line with the spikes in different years – you can actually step back through the record and track major hurricane events. You can track Hurricane Hugo in 1989. You can see Hurricane Andrew in 1992, Hurricane Katrina in 2005 with that big spike. You can see Sandy in 2012, and the triumvirate of Harvey, Irma and Maria in 2017 – so really remarkable.

Next slide, please. So getting back to a little bit about what – you know, why is NIST interested in this problem? NIST is primarily a measurement science agency and, you know, we like to say here, if it could be measured and should be measured, there’s likely somebody working on it here. So it’s a cauldron of many different experts that are working on different aspects of science and technology. I found a really interesting quote from a director from over six decades ago that basically explained – you know, tries to get at the idea that, you know, measurements are just as elegant, just as important as a big scientific idea. And I think, in some ways, this foreshadows what we’re trying to do with our NWIRP program.

So NWIRP, which is short for the National Windstorm Impact Reduction Program – it’s a science-based program. It’s across four primary agencies within the federal government – NIST, FEMA, NOAA and NSF. And it has the mission to achieve major measurable reduction. So you see that synergy between measurement science and trying to measure the reductions. And it’s obviously a huge problem, as the first slide indicates. You know, one of the interesting things here is it’s – to actually do this would require a sustained interdisciplinary effort across many disciplines – across meteorology, climatology, engineering, social science. You could throw in epidemiology and economics in there. I mean, it really runs the gamut. In some way, all of the ideas that we’re talking about here today on this call are philosophically aligned with the goals and, you know, the ideas of NWIRP.

Next slide, please. OK, so let’s get a little bit into the weeds – just a little bit – on some measurement science issues. So the title of the slide says Hurricane Michael, but I’m going to step back and give you some general information. The map you’re looking at is the design-level wind speed standard from the American Society for Civil Engineers. So basically, what you do with this map is – if you want to look at a certain area, what this is telling you – we say Risk Category II. That means your common construction, like residential homes and businesses. And essentially, if you’re living in south Florida, this is telling you you should, for minimum protection, build your structure to withstand – depending where, but up to a hundred and seventy, and in the Keys, 180 miles per hour.

You’ll also notice further to the west, those wind speeds are higher, but if you’re looking closely, you’ll see that in the Florida panhandle, there’s a hole, in some sense. The wind – the recommended wind speed design standards are much less than other parts of the Gulf Coast. If you’re thinking about this in a climatological perspective, it’s interesting because other weather variables probably don’t, on average, vary that much along such a short distance. Why it’s important for Hurricane Michael is because Hurricane Michael, which – the track is indicated by that red arrow – actually made landfall right in that area where the wind – the design wind speeds are less than other areas of the Gulf Coast. The blue shading there roughly indicates the area where the wind speeds from Hurricane Michael were in excess of the minimum design-level standard. So we call that an above-design-level event, and, you know, that’s why you – part of the reason why you saw so much damage. I mean, obviously, the storm surge in Mexico City Beach was really remarkable as well, but from the wind perspective for the damage, that’s part of the reason why you see this damage. The other blue highlighted area is the what-if scenario.

So if you recall back when we had Hurricane Dorian approaching, you know, maybe after it skimmed Puerto Rico, some of the initial projections were for it to make landfall in central or maybe even northern Florida as a Category 5 hurricane. And that would have been another above-design-level event, most likely. It didn’t happen, but if it had, it would have been the second above-design-level event in less than a year. So the last point about this wind measurement issue is, to do these estimates, to find out what the surface-level wind field was, it requires measurements.

Oftentimes, as a hurricane is approaching or making landfall, we lose some of the basic measurements that are out there at the weather stations. And so that’s an issue that, you know, is important for being able to constrain the model that’s used to figure out if the wind speed was above design level, but even if that didn’t happen, the density of the measurements, the resolution of them, is still probably – is still inadequate to really get a good estimate. And so, you know, we sort of have two issues there that need attention to get better at these estimates that we could link them to the damage to better correlate the wind speed with the damage.

Last slide, please. As I mentioned earlier in my first slide, this is also a story about water. You heard our previous speaker talk about the upland rainfall runoff. So precipitation measurement is also important in disasters. Now, the precipitation measurement over most of the United States is gold standard. It’s very good. We have radar. We have several different ways to measure it, but there are instances where there are some issues. What I’m showing here on the right in the plots is the precipitation maps from different methodologies from Hurricane Maria, and if you just follow one of the red lines down, you can see that, in some places, you get very different estimates.

Now, this isn’t necessarily because of measurement station failures or other failures of measurement devices. There’s also methodological differences in computing these estimates, and there’s reasons why they do that. But in this particular case, the National Weather Service says there were measurement problems with the infrastructure, and the radar was completely destroyed, so that’s likely to have some impact on some of these estimates. And I’ll just finish that. All these measurements are a critical component when we try to reconstruct a disaster event, an extreme weather disaster, at the resolution that we’re going for to try to understand the impacts and the hazard levels that were experienced. So thank you.

RICK WEISS: Great. We’ll move to Dr. Morss.

[00:21:52]

REBECCA MORSS: OK. So thank you. And first, building on my colleagues, I’ll start with some context. So if you can, go to the next slide. Thank you. So while we’ve all been going about our everyday lives, advances in science and technology have led to dramatic improvements in weather forecasts over the last few decades. But this is what a paper in Nature called the quiet revolution of modern weather prediction.

So, for example, this plot here shows the improvement in hurricane track forecasts, so forecasts of where a hurricane will go and where it will make landfall. And if you look at this and compare the lines, you can see that today’s five-day forecast is about as good as a three-day forecast was 20 years ago. So that means that we can provide emergency managers, members of the public and others with about two days of additional warning about the regions and if people need to evacuate or take other protective actions. And, of course, there’ve been similar improvements for all kinds of weather forecasts and we’ve all experienced these over our lifetime.

So, for example, if you think about other kinds of weather, like winter weather and the kinds of proactive decisions that airports and airlines have made to change flights around in advance of weather – I suppose people are just getting stuck as much as they used to – there are big, huge changes. And this has transformed weather-related decision making with huge societal and economic benefits. Of course, some types of forecasts are more challenging than others, and we experience that. We’ve heard about some ways in which those can be improved, and there’s a lot of active research in that area to improve observations, computer modelling forecasts.

The most quiet revolution has been so gradual and steady, you don’t always notice it until you look back. For example, Hurricane Camille, which affected the Gulf Coast and similar areas to Hurricane Katrina in 1969 – going back, there was some reporting after the fact that talked about how there were 15 hours of warning for the area the coastline most affected. So that’s when they started evacuating people in the areas that were most affected because the storm took an unexpected turn. And those forecasts were the capability of the state of meteorological science.

So if you think about today, when we get information five or more days in advance about, say, a Hurricane Sandy that’s likely to affect north New Jersey or an Irma that’s likely to affect Florida, the forecasts have gotten tremendously better, and that’s really enabled a lot of decisions. And because they’ve gotten so much better, we focus on the details and expect more detailed tracks and more detailed information about the hazards, like wind and storm surge and rainfall and so on. So along with this revolution in weather forecasting that’s enabled decision making, society has also experienced a revolution in information technology and communication. So there’s the Internet, cellphones and tablets, social media and so on. People can now communicate information much quicker with a much larger group of people through a much wider variety of mechanisms than they could even five or 10 years ago. And this, as we’ve all experienced, continues to change rapidly.

So we have these two revolutions. We’re able to forecast much better, and we’re getting better all the time. And we’re able to communicate information much better in much more complicated ways to a lot of different kinds of people. But those are only two parts of the equation. Forecasts are only beneficial if they’re received, understood and used. And so next, I’ll talk about how people communicate about extreme weather, especially as a threat is approaching and as the weather phenomenon itself and information about it is evolving. So if you go to the next slide, my colleagues and I and many others conduct research using social science and interdisciplinary methods to understand how people communicate, interpret and use different types of forecast information.

So on this slide, I have some examples from our research to understand people’s responses to information about a hurricane and their decision processes. So if you look at these slides, on the top two, these are examples of quotes from what things people have said using different kinds of methodologies. So the top two are from studies where we conducted interviews in focus groups after a hurricane makes landfall, and they showed the importance of communicating specific risks associated with a landfalling hurricane, as well as the challenges of how people perceive risk and how they understand something that they could never imagine.

So if you look at these top two, one person who said afterwards – this is after Hurricane Ike in Texas – I never dreamed of 7 feet of water. People prepared for 2 feet of water, maybe 3 feet, for winds, but they never dreamed that the whole first floor of their house – or sometimes their whole house would be underwater. And then on the right – top right is an example from a focus group after Hurricane Sandy, where someone said that we knew about the risk in our neighborhood, but the way the surge came in, this time, it was just mind boggling.

So that’s kind of related to how people interpret the risk even when there’s risk information. And then, on the bottom two – the bottom two are examples from some research we’ve done using Twitter data analysis and a survey, where people are talking about how they want to know the specifics of what will happen to them. So this is important going back to the last two presentations, that people don’t want to just know that a hurricane is coming, they want to know what might happen in their house so they can prepare. And that’s really important – or their neighborhood. And so despite all the improvements in meteorological science and information technology, those local details are still a challenge to predict and communicate.

From the bottom left, you hear – see an example where there’s someone who retweets a broadcast meteorologist talking about storm surge – maybe 8 feet with 10 to 12 – 20-foot waves on top. And this is for Hurricane Sandy in the Rockaways. And then there’s a tweet right back. That person retweets it, and they tweet right back at the broadcaster that they’re in this area – define major flooding, please. So the meteorologist thinks they’ve communicated – 8 feet with 10-to-20-foot waves on top, but the person wants to know, what’s going to happen to me? And then the bottom right, we have some examples from some surveys that we’ve done where people say they want things like more specific numbers about wind speed and height of storm surge and the information to be more local.

So the focus right now is on integrating expertise in social and physical sciences and engineering to learn how to improve weather forecasting and forecast communication in the ways that are most beneficial to society. So if you click forward to the next piece, a lot of the work that we and others are doing now is to try to learn about what hazardous conditions and impact people will experience – because that’s what people want to know – and connect that back to how we can predict this information and also communicate this information. So how we tie it all together to be able to tell people what they want to know and what they need to know to protect themselves given the achievable forecast scale at different lead times. So next I’m going to talk a little bit about the complexity behind the decision. So when people talk about decision-making for hurricanes or for other hazards, they sometimes think of it as simple.

So if you go to the next slide, we think of it as simple. People get a forecast or a warning from the weather service or from a broadcaster or a public official, and they decide what to do. And then we ask questions afterwards, like why they were in the area at risk – everyone knew that. Why didn’t they evacuate? But there’s a lot of complexity when people understand risk and make decisions, and we can think about that in our own lives if you think about how we make decisions – even things like when we’re approaching a traffic light that is turning yellow or we go to the doctor and they give us information. There’s a lot of complexity in how we interpret information.

So there are a lot of factors other than the actual risk – so the quantified risk that scientists can talk about or the perceived risk – the risk that people think about in their heads that influence people’s projected decisions. Of course, there’s finances. Do they have money? The transportation that they have available, whether there are constraints about their work that wants them to stay, concerns about traffic in terms of not being able give out quickly – there are a lot of constraints that people experience when they’re trying to decide about evacuation or other protective decisions, like boarding up their home. And there’s a lot of studies of that, and we know a lot about those kinds of things. And then it’s a lot more complicated than that because these things are all interacting.

So this diagram illustrates how, going back to the two revolutions I talked about earlier – the one about weather forecasting and how much that has improved over the last few decades, as well as how communication has really transformed in the last few decades – and what that means for hurricane forecasting, communication and decision making in today’s world, in what we call this modern information environment. So when a hurricane threatens, you see – look at this diagram. On the bottom, we have the kind of idealized forecast, and one response is for – often think of as one piece of information when someone decides what to do. But in reality, now, nonetheless, people get information about a hurricane starting five, seven, or even more days in advance. And, of course, at this time, there’s a large area that’s at risk, and there’s significant uncertainty about where the hurricane will affect people and what those specific causes will be – what wind and storm surge and flooding and rain people experience.

People start to communicate and process that information. So that’s represented by the different colored symbols of different kinds of people interacting over the land. And that involves complex individual psychological processes, as well as social and cultural practices, since we’re complex, social human beings. So those are the ways that people process the information and make decisions. And then now, new forecasts come out all the time. There’s official forecasts that come out maybe every six or 12 hours or more frequently as the hurricane approaches. But there’s all kinds of other people that are producing new kinds of forecast information based on that, probably every second or even more frequently. So new information is coming out all the time. And as the hurricane gets closer, the area of risk often narrows, and there’s – more specific about how the hazards. And people are processing all that information and deciding what to do in this really evolving, complex situation.

So if you think about our everyday lives, people are really immersed in a vast sea of information in general, and the same goes – is true for a hurricane or another kind of weather hazard. People are immersed in this vast sea of evolving risk information, and they access and interpret that information through really complex interactions in their own heads, as well as across the physical world, where they’re talking to real people in real life, as the digital world, where they’re integrate – interacting with a lot of people who they may know or people that are – may – they may not know that they’ve just become connected with over the Internet. And all of this comes together in a complex and evolving way to contribute to people’s evolving risk – interpretations of the risk and their – the way they make decisions.

So it’s really important, putting this all together, to communicate about risk with meeting – by meeting people where they are in the context of what they’re doing and thinking and the information they’re getting – the other kinds of information, how they’re processing that, and how it connects to their everyday lives. So related to that, I’ll close by talking briefly about the importance of thinking about forecasting and communicating risk not only to the people who are easiest to reach or those who have the loudest voices that we often hear on social media or otherwise, but to all people, including populations that may face challenges in accessing or understanding information or may have more constraints on what they can do to protect themselves or may have other kinds of vulnerabilities.

And so we approach this by asking about how communication of risk, especially in this modern information environment where a lot of people are connected on the Internet or have cellphones or on social media, but not everyone is – how does that alleviate or exacerbate social vulnerability to hurricanes in general and particularly the most vulnerable members of the public? We really want to make sure that people who are going to suffer the most and are going to have the most difficulty recovering afterwards really get information to protect themselves and do the best – make the best decisions they can.

And one way of thinking about this is that what many people experience – the stress of a hurricane, making decisions and all of the challenges associated with that – is really exacerbated for some people based on their characteristics or other kinds of factors, and it’s important to think about that. So when we think about the overall goal of risk communication, it’s really to help people understand their risk in their own context and make the best informed decision possible. So we sometimes think of this as a top-down situation, where someone can tell them what’s going to happen and what to do. And then they should go and do that, and definitely authoritative information can be very important for some populations. But for some people, as we know, authoritative information is actually not important, and they actually sometimes want to do actually what the authority says they shouldn’t.

So it’s really also important to think about risk communication in terms of helping people understand what their risk is in a language that makes sense to them in their own context, what can they do, and meet people where they are. So one way to do that, as we talked about in the earlier presentation, is to extend the predictions of weather to predictions of weather hazards and impacts. And so not only to say, here’s a hurricane, this is where it’s going to track, but to say, these are the kinds of winds you might experience. This is the storm surge you might experience. This is the flooding you might experience – to provide the information related to the forecast that people make – need to make decisions.

So it’s really translating the meteorology into something that’s closer to what people – is going to affect people and what they need to make decisions about. And then, also, to forecast and communicate in ways that connect with people’s capacities – people have different capacities and different kinds of groups have different ways they make decisions and different approaches even to life and what they want to do. And so to really think about that when we’re communicating and forecasting and make sure that we’re communicating in a way that connects with what people’s capacities are and what kinds of decisions they can and want to make, and not just what decisions we would make, as communicators or weather forecasters. And then also to communicate in a way that is collaborative – so not focus on communicating to people in communities at risk, but really on communicating with them. And this happens outside the context of a hurricane. That’s when a lot of the relationships get built that enable the communication during the threat, when people trust each other and they know what kind of information different kinds of people want and so on.

So I wanted to close with that by emphasizing the importance of communicating to all kinds of groups of people and to thinking about it in that context.

[00:34:20]

RICK WEISS: Great. Thank you very much, Dr. Morss. Want to remind the reporters now, if you look towards the bottom of your screens, you’ll see something that says Q and A. That’s the place where you can submit your questions. You can designate to a particular speaker or more generally. And I’m going to start right off the bat with a question from Michael Tutton at Canadian Press, who says – Canadians, this may be for you, Dr. Westerink – but Canadians are wondering, particularly on the eastern coast, with what degree of confidence can we expect the Atlantic and eastern North Pacific hurricane rainfall and intensity to increase, and to what extent is Nova Scotia and Newfoundland and Prince Edward Island, in particular, exposed? I know these were areas that were in Dorian’s path for a while. It certainly surprised me, at the time, that the damage was heading that far north. Can you talk to that issue?

[00:35:22]

JOANNES WESTERINK: Certainly to the – I guess, the meteorology is not my area of expertise, but hurricanes are, of course, becoming stronger as they’re propagating further north. And there are fairly wide continental shelves, so there’s risk for exposure in some of the regions. So it very much is an evolving picture. But in terms of the actual hurricane risk, that would be outside of my domain.

SCOTT WEAVER: I’ll take a crack at a comment on that, if that’s all right. It’s a little speculative, of course, but, you know, up there in that region is usually where hurricanes are transitioning to what we call extra-tropical systems. So the interaction of a tropical system and your typical, say, you know, fall-or-winter-type nor’easter coming in – or frontal system coming in, you know, is highly uncertain at this time. But as it was noted in the beginning of the presentation when the – when Rick was going over the climate change facts, you know, the tropics are expanding.

So while I don’t have definitive information for you, if you have stronger hurricanes moving northward interacting with the other systems – like Hurricane Sandy or – I can’t remember the storm, but it was in the fall of 2000 – it was in October of 2000. They made a movie about it, “The Perfect Storm.” If you have those situations developing, then it could increase the rainfall risk markedly.

[00:36:55]

RICK WEISS: Great. I wonder if – here’s a question from Tom Frank at E&E news who’s asking if someone would like to evaluate the forecasting for Dorian. Dozens of counties, he says, in Florida and Georgia were under evacuation orders, perhaps unnecessarily. Did Dorian show the need for better forecasting?

[00:37:17]

REBECCA MORSS: So I can comment on that. Dorian was definitely a challenging situation, and it was one of those storms that we saw with Hurricane Matthew a couple of years ago as well – where one of the particular challenges was that the storm was kind of – it wasn’t exactly clear where it was going to be curved – this was actually the same situation with Irma as well. So this one was going west, and then exactly where it starts to turn north and how quickly makes it a challenging forecast because there are so many places that it could hit along the coastline. And so it’s really the details of the forecast that are a challenge.

I am a meteorologist. I work at a place with lots and lots of meteorologists. So I was hearing from people about the rapid intensification of Dorian. That was pretty well-forecasted, actually. This was a challenging situation before it hit the Bahamas. So the fact that people were able to know that it was likely to intensify from a tropical storm up to a strong hurricane was a great success that probably wouldn’t have happened five or 10 years ago. But it showed, really, the challenges of predicting the details of where the storm’s going to go in a way that could help people make decisions. So there’s definitely still a lot of room for improvement and there’s a lot of investment in that right now in terms of trying to improve those details for the forecast track.

[00:38:30]

RICK WEISS: Great. Question here from Neela Banerjee at InsideClimate News. What are the ways that climate change may have fueled Hurricane Michael? I’m particularly interested in how quickly it intensified before landfall, realizing that research into rapid intensification is still nascent within attribution science more broadly.

REBECCA MORSS: I can’t comment on the climate change…

SCOTT WEAVER: I can take a shot.

REBECCA MORSS: OK, go ahead, yeah.

RICK WEISS: Go ahead, Scott.

[00:39:10]

SCOTT WEAVER: So one of the things that was noted before – we took a look at this before Hurricane Michael made landfall. There’s actually, really – I thought about putting the slide in the presentation, but there’s a – you could look at the Gulf of Mexico ocean temperatures – the sea surface temperature, and they were about 2 degrees above average for that time of year. The issue is that you have naturally varying cycles – things that vary on decadal timescales and year-to-year fluctuations that can account for some of that. And then you also have basic – globally averaged and global footprint of increasing sea surface temperatures.

So given the other conditions being in a perfect, you know, situation, you know, Michael was able to essentially extract the maximum amount of energy from the ocean surface, and that’s a hurricane’s job – to do that, to turn it into wind and rainfall, and deposit that on the coast. And so it was a very efficient situation from that standpoint. So I think, you know, untangling the intersection of the natural varying cycles in ocean temperatures combined with the increasing trend – you can – you could begin to get, you know, some sense for it for the potential impact.

[00:42:23]

REBECCA MORSS: Yeah, and I’ll add to that that rapid intensification is a risk that hurricane forecasters and meteorologists worry about – have worried about for many years. So it’s something that happens, and that’s a big concern. But as Scott mentioned, it’s whether there are other factors related to climate change that can really increase the risk of that happening, especially so close to land.

And in addition to what he mentioned, it’s not just the surface of the ocean. It’s the layer of what – how deep the layer of warmer water is beneath that. So the hurricane will track over the ocean and it will pull up heat from the ocean, and that will mix the top layer of the ocean. And then if there’s more warm water underneath, it will continue to extract more energy.

[00:41:02]

RICK WEISS: Great. A question now from Mark Schleifstein. He’s at the Times-Picayune New Orleans Advocate. And this is directed to you, Joannes. In south Louisiana, he says, we have rapid changes occurring in the shape of the coast – erosion, coastal restoration projects, new levees being built. What’s being done to assure that surge and wave models can take those changes into account fairly quickly? Ditto for Mississippi and Alabama in light of the BP projects that are going on there as well.

[00:41:39]

JOANNES WESTERINK: Great question. So first of all, the data is getting much, much better that is ingested into the models. And so in a lot of – since Katrina, there’s been huge amounts of water-penetrating LIDAR that has become available. So really the near shore the bathymetric and topographic profiles are getting better and better. And the second, really, thing that’s happened in tandem with that is that a lot of the meshing has become much more automated.

So whereas it used to take a man year to put one of these finite element meshes together, now there’s meshing technologies that we’ve been developing that can mesh a very complicated region in a matter of days and really take and absorb that data much better. So the precision and turnaround time has been vastly improving. So that really allows it to be a reactive process, where, as the coastline changes, these kinds of updates can be quickly implemented. The other thing I should add is that the level of resolution and the level of detail in the really dendritic connectivity of coastlines like Louisiana’s are much better represented with these much better gridding technologies. So things that were left out and channels into – penetrating into the wetland, only the major channels were considered in the operational models running today.

But in the new generation of models, streams on the order of, let’s say, 10 meters or so and that whole connectivity, which is very instrumental in early flooding, is much better represented. So I would say the data and meshing technologies are – in addition to the processes that I discussed, the process of inclusion – that those are going to be much more reactive to changes in the coast.

[00:45:46]

RICK WEISS: So I’m going to take the moderator’s prerogative here to add a follow-on question to that because I’m curious who’s doing this work. If you’ve got new data and you need to mesh it and a reporter wanted to know who’s doing this and how – what is really involved, maybe they even want to visit and see how these new data are being added – are these computer scientists? Are they mathematicians? Are they meteorologists? Or is it just a machine that sucks it up and spits it out?

[00:44:12]

JOANNES WESTERINK: No, it’s code development in support of the main physics codes. So – and in fact we have several contracts with NOAA that’s doing this. And so NOAA is using the technologies that we’re developing, but we’re developing them right here in our group.

[0:44:31]

RICK WEISS: Coding technologies?

JOANNES WESTERINK: Well, not only the coding technologies, but also the meshing technologies, which are the drivers and support system – right? – for the actual codes. So there’s the codes – the physics codes, right? – that represent those processes, but they’re supported by the meshes. And the meshes contain the critical geometric, bathymetric, topographic, physical system information that makes the computation accurate. So only when you describe that system to the code will it happen. And that meshing is also becoming an automated system that’s very much able to adapt – right now still statically, so one shot at a time. But eventually, it’ll be able to adapt dynamically in time as these technologies evolve.

But the idea is very rapid turnaround evolution of the meshing and the detail that it contains as, for example, Mark mentioned, that the coastline is very much changing with all these sediment outlets that are going to be distributing sediment or hopefully distributing sediment over the wetlands and rebuilding them. So that needs to change in the code. And so you would mesh very differently with a stream going through there that would have presumably been formed by the outlet and the sediment that’s being deposited on the adjacent wetland. So that’s all morphing, and that needs to be changed in order to get a good picture of – or to really understand what a storm surge might be doing there.

[00:46:10]

RICK WEISS: Got it. Question here from Joe Mario Pedersen from the Orlando Sentinel. We all pay attention to these CAT-level wind speed categorizations, but he’s asking or stating that NOAA has stated that tropical systems’ deadliest weapon is rainfall and storm surge. Was Dorian an example of this over the Bahamas?

REBECCA MORSS: Yeah, I’m not sure about Dorian specifically. I think people have done statistics of the deaths from hurricanes and tropical storms over the last 40 or more years, and definitely storm surge and flooding is a large killer than the other kinds of hazards. But it does depend on how you look at it. So for storm surge – for example, Hurricane Katrina, there was a lot of deaths that were difficult to attribute, but there are some storms that are major killers from storm surge and flooding. And then there are other storms where the deaths from wind-related issues are kind of more spread out.

So there’s things like trees falling on people’s houses and electrocution and other kinds of things. So a lot of it depends on how you look at it. So I would say that a lot of it depends on the storm, and it’s important to know that there’s multiple kinds of risks associated with a storm. And then depending on where you live, there might be a combination of those risks or there might be one or more that’s worse. And so I think that’s really important to think about.

So – especially how to communicate that. So with Hurricane Harvey, before it made landfall, it was pretty well-known among meteorologists that there was a huge risk of rainfall-induced flooding. But a lot of people didn’t understand that as well as they could have or didn’t know that their areas were at risk for various reasons. And so there are still gaps in communicating sometimes what scientists know and how they can get that information out to people. I think it’s challenging with a hurricane because you’re often communicating to a broad population. And so we want to communicate to people who are the highest risk right along the coast, but then the people who might experience something else – the message can get a little difficult to understand if it’s you or if it’s someone else that has to worry about a certain thing.

[00:48:24]

RICK WEISS: And Dr. Morss, I think this may be a follow-up for you. This is coming back from Tom Frank at E&E news. He’s asking, isn’t the real issue with communication how to communicate long-term risk? I don’t mean – you know, he’s not – he doesn’t mean three or five or seven days in advance, but longer-term than that. So few people have flood insurance, he notes, and end up with enormous losses. What can be done to get more people to buy flood insurance, which is supposedly mandatory for many property owners?

[00:48:52]

REBECCA MORSS: Right. So that’s a big challenge. People have been working on that for many, many years since there was flood insurance. In fact, my parents say that when I was born, they lived in the area that was prone to flooding. And they had a choice of buying flood insurance or buying my diapers and they chose diapers. So that was before the National Flood Insurance Program. So that’s an example. My parents are scientists, so – I mean, I remember flooding when I was a kid.

RICK WEISS: (Laughter).

[00:49:05]

REBECCA MORSS: So that’s the example of the kind of choices some people are making. And the National Flood Insurance Program was supposed to help that, as well as to incentivize flood insurance. And so there are improvements, but definitely still challenges. So I would – the way I think about it is that communicating risk across timescales is important.

People knowing their risk on longer timescales is important so they can do the long-term preparations they need to do, as well as so they can be ready – they can have it in their mind that something could happen so that when they get the specific information that a storm is coming and might produce flooding that they have that in their mind. So I mentioned earlier that I did – conducted interviews after Hurricane Ike in Texas where we talked to people afterwards who didn’t really realize that they were at strong risk from flooding. They thought that there had been mitigation done, that if there was flooding, it wouldn’t be that bad. And so even though there were really small things they could have done to prepare, they didn’t do that because it didn’t occur to them there would be 7 feet of water.

And so part of that is because people didn’t have the knowledge in their mind beforehand to be able to conceptualize that so that when they got the forecast, it made sense. So we’ve done lots of work in this area, and there’s current research going on now with things like virtual reality and augmented reality to try to see if those kinds of techniques can help people realize that the place that they envision as this dry place that might even be far – far-ish from the ocean – they can’t see the ocean – really could be under that much water. And those are important for helping people make long-term preparations like buying flood insurance or doing other kinds of mitigation as well as for responding when a storm approaches.

[00:50:51]

RICK WEISS: That’s very interesting. Seeing one’s own house underwater with an AR system would, I would think, be a nudge of some kind. We have a question here from Randy Loftis at Texas Climate News. Are there parts of the Gulf or Atlantic coasts where understanding is better or worse than in other places – that is certain spots where forecasts are likely to be more or less accurate in the same way that some of those building codes may be varied around the coast?

[00:51:20]

SCOTT WEAVER: So I can’t necessarily speak to whether one area is more accurate or the other. But at the end of the question, you’d mentioned the building code issue and why – you know, why there’s the hole there. You know, those maps are essentially built off of historical data. And so if there weren’t many strong hurricane landfalls in a particular area, that’s going to be reflected in the map.

Now, there’s some statistical processing. You know, a hurricane landfall is a relatively rare event when compared to, say, just having rain every few days, right? You have many measurements of just garden-variety rain, but a hurricane landfall is – it’s relatively rare despite what it may seem like has been occurring. And so there are some statistical methods that are used to, you know, take a rare event and kind of build up the sample of cases so that you can get an estimate of the probability of its occurrence. And – but it’s still very anchored to what happened in the historical database.

So in the case of the Florida panhandle, for instance, there just haven’t been many or any – I’m not really sure on that, but it’s either very little or none – hurricane – category 5 hurricane landfalls in that area. I will say that the calculation of those maps – the data goes up to 2012, so it doesn’t include things like Hurricane Michael. In the next edition of the standard, they will add in the intervening years’ data, and that should be reflected in the maps. So we’ll probably see difference – a difference in those maps.

[00:52:51]

JOANNES WESTERINK: I can add a little bit to that also, if I may – that the shape of the continental shelf and its width also play into that risk in terms of when you get a specific type of storm in one region versus another, as well as the shape of the inland bay and water system – what that looks like.

So that can dramatically increase or decrease risk. In general, narrow continental shelves – the storm surge is going to be much smaller than broader continental shelves. There’s simply a lot more area to build up to storm surging. And then the complexity and bathymetry of the inland water system also can very significantly amplify storm surge.

So those are big factors in terms of risk from one spot to the other. Also. the projection of systems like a deltaic system – for example, the Mississippi River Delta – tend to capture a storm surge and it gets amplified there. So the risk from one spot to the next is very, very different. So location is a major player in terms of the storm-surge-risk side of it.

[00:54:01]

RICK WEISS: Great. I’m going to try to squeeze in two last questions that we have here and end on time. One is really more of a comment, but I wonder if anyone would want to address it. This is from Bryan Norcross at WPLG-TV in Miami who notes that this statistic about water being more deadly or harmful than wind holds up only if you don’t count Hurricane Maria, where perhaps 3,000 people, he says, died because of wind damage. Does anyone want to address that outlier?

[00:54:31]

REBECCA MORSS: I’ll make a comment. First, hi, Bryan, how are you? Also, right – as I mentioned, Hurricane Katrina is one of those outliers and Hurricane Maria is another one. And so a lot of it depends on how you attribute those deaths. People do that differently sometimes, and a lot of those deaths are due to a combination of not the immediate risk of the storm, or the immediate impact versus other things. And so for some storms, wind is really important.

For others, surge or freshwater flooding or a combination is really important. And so it’s important to think about all of them. And if you put – you know, if you put one storm in or take one storm out, sometimes it can change the results.

[00:55:15]

SCOTT WEAVER: Can I add a quick comment on that, too? So obviously, yeah. It’s an extreme outlier. In that situation, it was a multi-hazard event as well. So, you know, the rainfall led to thousands of landslides across the area. And so trying to get at, as Rebecca had mentioned, you know, direct versus indirect deaths is sort of, you know, not a standardized process at this point.

And, you know, you had significantly – a very unique event with the landslides, the rainfall, the wind and the storm surge all conspiring around different damage patterns. And, you know, some people were trapped and maybe couldn’t get their medicine or couldn’t get to a hospital. And so trying to attribute that is fairly difficult and the methods vary.

[00:56:03]

RICK WEISS: So a good kind of standard warning to reporters when presenting statistics like what is the biggest killer in a hurricane – to always be aware of what you’re leaving in and what you’re leaving out – good general rule of thumb. A real quick last question and we’ll close, and this is a follow-up from Mark Schleifstein at the Times-Picayune New Orleans Advocate. Are there any efforts going on to move the AdCirc coastal circulation model inside the New Orleans levees for both rainfall and surge events?

[00:56:35]

JOANNES WESTERINK: The answer is yes. For the sTOF – NOAA’s sTOF forecasting model that’s run by NCEP – we are actually coupling to WRF-Hydro, so that is the hydrologic coastal surge coupling. And we also have a project under a program from National Science Foundation called PREEVENTS that is integrating hydrologic models in general in a very automated way with AdCirc.

So they’re going to be – it’s going to be dynamic physics, and the physics will shift as the hydrology and hydraulic strain. So as the storm surge propagates inland further, the standard hydraulics physics will take over. If there’s either wetland or downhill hydrology, rainfall runoff, different kinds of hydrologic physics will take over. So the idea is – that’s evolving in the community in general is to have very dynamic types of physics that change depending on what the needs are.

[00:57:37]

RICK WEISS: Well, I have to say it’s gratifying to hear about all the progress going on in measurement and prediction, in trying to get people’s behavior in line with the actual risks. Some excellent stories to be written in these areas. I want to thank you all for participating in today’s hearing or briefing and – luckily you’re not before Congress today.

But I do want to remind all the reporters who are online today that when you shut down this briefing, you will get a prompt to take a short survey. It’s three questions. It’s really short. It really helps us do a better job for these briefings. So I hope you’ll take the minute or so it takes to respond to those questions. We’re very grateful for that. Other than that, you can expect to see a full video and transcript of this briefing up a little bit later this week.

We encourage all of you to follow us on Twitter, @RealSciLine, and to visit our website, sciline.org. Thanks to our three guests very much. And reporters, we’ll see you at our next briefing. Good afternoon.