Compound Hazards - Physical Science for Integrative Risks | 2026 ASP Colloquium

[music] welcome Colin to give us our first foundational lecture here on compound hazards. Colin is actually one of the speakers who I did not meet before today. Um so I'm very excited to have him here. But when as Mariana mentioned a year ago when we were writing the proposal for this colloquium, we obviously knew we wanted to send her um systems engineering and convergence thinking and we kept thinking like what's the hook? What's the big problem? What's the topic we want to focus on? And then I think it was right around a year ago that we said look I think it needs to be compound hazards and we were doing quite a bit of literature reviewing orienting oursel more deeply to compound hazards including looking at UN reports United Nations reports and other things and I came across a paper that Colin was a co-author on and a lot just kind of went from there once we got the proposal funded then when we were starting to think about who do we need I saw Colin's name in many different spaces as somebody who does have expertise and compound hazards. So, we are very excited to welcome you, Colin. Thank you so much for being here. I hope you'll also introduce yourself more fully to the students. Okay. And take it away. >> Thanks a lot. >> Uh the organizers have seen fit to allot me and also the other foundational speakers uh 90 minutes. So, I don't know if I'll use that all of that time. I I did put about 15 or 20 minutes of discussion prompts kind of in the middle to break things up a little bit. We can also take a short break or something if we feel like it. Um, I'm a research scientist at UCLA. Uh, for the moment I have a career change uh coming up later this year, but the last few years I've been at UCLA doing research on compound hazards and um the meteorological drivers of them, especially focused on extreme heat. And it's it's been an interesting uh journey, you know, living in LA. I'll talk at length um in the second half of the presentation about the LA wildfires. So I have been thinking about compounding for quite a while along with many others whose work will be cited here. Um but the LA fires really encapsulate, I think, a lot of the both physical and non-physical dynamics that drive things that we care about like yeah integrative risks. Uh maybe more more generally. Um this is my first time actually back at ENCAR. I was here as a prospective graduate student um over 10 years ago. So it doesn't seem that long really, but I guess I'm kind of an established scientist now and I certainly wasn't at the time. So I I also around the same time attended a summer colloquium um at the other uh Encar lab down in in the valley. So I yeah it doesn't feel that long ago that I was sitting kind of in your in your shoes. Um yeah so I guess that's probably enough by way of introduction. Um so the outline kind of the talk is that you'll see at the top and so I want to start off with a a broader kind of pitch about compound uh hazards. of course you're interested in them to a degree because you're you're here and there's a temptation to kind of say well this is a new thing that especially in climate science is is you know the frontier of how we're thinking about the earth system and humans and stuff but really if you look at like natural hazards or just kind of human experience more generally there's a lot more um background to this we just were using different terminologies at the So the top left is a a painting that was done in the 19th century shortly after this fire that occurred in Pesiggo, Wisconsin on the same day as the Chicago fire actually. And it um burned many uh thousands of acres. I think it killed more people than than the Chicago fire too, but it was in the backwoods of Wisconsin, so it doesn't get as much attention. Um, and the reason it's a compound hazard, of course, is not only was it super dry, there were high winds. Um, but those winds actually blew the fire all the way across the Bay of Green Bay, so like 100 kilometers or so, and of course, people and animals then like dove into the water to try to escape the raging flames. Um, another type of disaster that has always kind of been multi-hazard or compound in a way are hurricanes. This is um photo from 1925 what's called the great Miami hurricane obviously caused a lot of flooding. You can see the high winds um and yeah devastated South Florida and really affected actually the whole trajectory of the development of the region. So people of course knew about these things and cared about them but didn't you know have the tools or terminology to study them the way that we study them now. The kind of modern story of compound events at least from like my corner of climate science is usually uh thought to begin with this 2012 I IPCC international panel on climate change a special report on extremes which kind of summarized the understanding at the time of how compounding um effects across for example um the heatwave and drought dynamics ics uh could lead to events that were more impactful than either extreme would be by itself. So a somewhat you know narrow framing um and it was drawing from a lot of literature again kind of in the heat drought space a lot of kind of European expertise um but really kicked off this this concept in a lot of people's brains like the the sort of proverbial light bulb like wow this could really affect how we think about climate change going forward maybe we shouldn't just be looking at you know change in the 90 changes in the 95th percentile of extreme heat or something but what are the interactions of that with, you know, storms and droughts and things like that. So, it's been sort of an evolution in the past decade and a half and and and obviously a huge blossoming of work. Um, but this was a a conceptual landmark. Uh, another conceptual landmark I suppose has come from the the changes in the climate in the western US in the last 10 or 20 years. um especially all the droughts and wildfires and then their uh knock-on effects. Um this is from a nice summary paper by Amir Agakuchock at UC Irvine and his team which is a beautiful illustration of a really devastating event, the Monaceto mudslide. Um to people of my slightly more advanced age, this is still considered recent. You might not consider 2018 recent anymore, but as you can see, the the sequence was quite like stark, right? There's a drought, there's heavy rain, there's strong winds as well as, you know, dry a dry atmosphere associated with that. And then more heavy rain shortly thereafter. And because of the amount of vegetation there was to burn, the wind the fires are very intense um also driven by the the flames or the driven by the winds. And then because the soil is loose and there's no more vegetation when the heavy rain subsequently comes you get debris flows. And this was a a really uh devastating event for Monaceto. So it's about you know not only the extremes of each type but also the timing and the sequencing. If there hadn't happened to be a heavy precipitation event in that winter of 2018 maybe it had been pretty if things are pretty mild and it didn't um core until the next fall. all the vegetation would have had time to grow, debris flow might not have been as extreme. Okay, so kind of getting down to the level of more technicalities. Um the most there are a million definitions of compound hazard. It's one of these things people sort of love to argue about. I was part of this this paper um which maybe profered the most commonly one used one uh so far which is multiple drivers I I'll kind of explain what these mean um multiple drivers um andor hazards that in in the climate space usually that interact and create an impact potentially greater than the sum of the parts. Uh so what what does that mean? So a module so there's a modulator first but that sort of is behind all of this at the biggest level we can say you know climate change is changing [clears throat] impact of or the the characteristics of ENSO for example but then you have ENSO itself or or other things um different types of uh you know the NAO for example then you have you know the N so ENSO is affecting say drivers so the drivers might be like what we call you know weather kind of broadly right um Rosby waves. Um there might be a lot of rain in a particular season that causes saturated soil, something like that. Then you kind of drill down to hazards and try to represent hazards are some sometimes kind of um this there's a blurry uh line between them. But a hazard is maybe something that we isolate and really say this is the most direct threat to things we care about. So wild and things we have a kind of specific name for fire, heat wave, flood. Um these can be problematic of course and how to define them. There are, you know, how do you talk about extreme heat in the tropics? It's not really a heat wave, right? But generally speaking, you can kind of talk about specific episodes that lead to impacts. impacts are also hazily defined but something like consequences that we care about for people, assets or ecosystems. Okay, so compound hazard then can be driven by an extreme I guess it could be driven by an extreme actually modulator too like this upcoming extreme ENSA we're likely or Elino event we're likely to have um so we can have an extreme driver so we can have say a really severe cold front or something like that uh we can have an extreme hazard that's the most common um that's the most common framing uh of compound kind of hazard is this box from extreme hazard driving an impact an extreme let me go back you know an extreme wildfire an extreme heat wave um whereas an extreme yeah driver would be something like the soils were extremely wet so just a small additional rain event caused a huge flood for example or there was a there was a there's a very deep snow pack so then there was a small rain event that caused an extreme flood. So that's the the less common then you can have that and then of course you can kind of focus on the extreme impact and work backwards. As climate scientists we don't tend to do this because we're interested in the extremes in the system we kind of are most trained to look at. But you can also start from okay there was a really terrible um harvest in this you know country in this year. What were the drivers of that? And maybe they weren't necessarily extreme but in combination they they they caused some problem. you know the timing of the rain or something like that. Um okay so a couple kind of notes then. So extreme impact can be defined well really all of these but let's say extreme impact can be defined by percentiles or thresholds um you know the the interactions across drivers hazards impacts can be modulated by the earth system the built environment social systems um and the modularities of interaction are really kind of what convergent science is all about. Um again as as a climate scientist I tend to my myself and my colleagues tend to focus on how the earth system modulates these these interactions but you can also focus on um you know agriculture um cities governance um yeah economics. All right. So, kind of more conceptually, I guess the way I would I would I would suggest these things are normally thought about is that you have on the y-axis risk. Many definitions of risk. I'll kind of say potential damage from a compound hazard is generally somewhat higher than single hazard although not necessarily so. And then kind of on top of that you have sort of bad bad luck or unlucky decisions. So in the LA fires again which I'll talk about at length there's a number of bad uh number of instances of bad luck or kind of poor decision-m that led to greater risk. So there's both kind of earth system, this is just another way of talking about I guess earth system and social system interactions that lead to greater risk um and yeah greater greater impact uh greater concerns. Okay. So the typo so the other thing about this paper that I was part of a few years ago is that it defined kind [snorts] of in a canonical in a canonical way different types of compound hazards. So without stressing this too much it's important I think to to lay this out at least you can can have it in mind. So there are four main ones that uh we we talked about with some somewhat subtle distinctions between a few of them. So the first one uh multivariat which is multiple drivers or or hazards which cause an impact at the same time in a certain location. So for example um heat heat and drought occurring simultaneously leading to crop damage. Um then you have um kind of compounding or interactions across time. So again kind of in a given location but over time uh multiple things happening. So for example I already talked about like wet soils followed by um you know rain heavy rain that causes wet soils followed by more rain that causes really severe flooding or potentially landslides. This is very much the Agakuch mudslide example. things distinct things happened in a distinct sequence that led to this outcome that was very severe. Then you can kind of imagine spa for spatially compounding it's it's uh multiple things happening at the same time or more or less the same time across different places. So this is created this is driven um in its impacts very much by which particular locations you're talking about. They can either be uh contiguous so or more or less contiguous. So the the images of the massive Canadian wildfires I think in 2023 um which you know Canadian wildfires although they're somewhat new kind of to those of us living in the US Canada's had boreal forest wildfires a pretty significant extent for you know decades centuries. Uh however the like when you get to a certain size uh in aggregate of those fires like you know 25% of a province burning then you have huge amounts the the smoke is so much that you can transport it very far downwind and create kind of a qualitatively more um you know serious impact uh for example in the eastern US or even they they even had uh smoke problems in uh Europe. up as a result of the the transport of the smoke. So if you'd had a few smaller fires, it wouldn't have, you know, been nearly the same or the more kind of classic examples like red basket failure. So there are multip you know the US Midwest and Argentina and Brazil or the US Midwest and Brazil and India or something all having uh you know like heat and drought or something around the same time which maybe really impacts the total global available amount of some crop like wheat or corn that's traded on global markets driving price increases and you know potentially famine in places that are uh unable to afford the higher price. So that would be yeah the most classic example of spatially compounding. Um but I wanted to pick something that was a little bit different than agriculture um for for the example there in the in the image. And then you have preconditioned which is kind of a a subset of temporally compounding I guess but it's more like there wouldn't really the kind of way to think about it when people use this term is the the first event wouldn't really cause an impact by itself. It's only by virtue of the second impact that anything really mattered. So for temporally compounding it's like maybe there was heavy rain and a minor flood but then there was more rain so there's a huge flood. Preconditioned is more like there was a bunch of snow and it would have melted gradually but then you had really heavy rain which obviously added more water and also melted the snow creating sort of something above and beyond either one individually. Um, and the example, I know it's kind of small, but there was a um rain on what's called rain on snow event that devastated um this village in the French Alps um a year or two ago. Uh so it's kind of harder to find good examples of that, but as a kind of category, it's useful to keep in mind. Okay. So I already I guess mentioned that the overlap, you know, when I say at the same time, that can be defined from either a physical or social perspective. If you have something like heat and drought, it really has to be at the same time to affect the crop in my understanding anyway. Whereas something like bread basket failure, if you have, you know, crop uh shortages in the US in September and then you have a drought in Brazil in December, that still affects the markets because crops are traded over many months. Um and then I was part of this um summary paper uh this past year and we kind of suggested that that there's maybe a need for this additional category and we're not the only ones to have suggested it but we we called it cross cutting. So basically things that intrinsically involve multiple of these at the same time. Um and the LA fires I think as I'll explain are a good example of that. Um okay so then other terms that people use that are useful to kind of map onto this one of them is co-occurring. So co-occurring is at the same time, right? So it's either it could that could be a multivariate event or it could be a spatially compounding event. Similarly, consecutive as I mentioned could be either temporally compounding per se of hazards or preconditioned some system state followed by kind of an extreme hazard. And then people also use cascading which is a little bit uh a little bit vague. Um, to me it primarily refers to like impacts. Um, I suppose it could refer to some of these others, but I guess when I personally use cascading, it kind of means um things happening within the impact space more so than in the hazard space. Okay. Well, enough of that. I suppose um the existing literature in kind of climate compound hazards was nicely sum summarized by this paper last year and they did some text analys an analysis and sort of broke up uh of the published papers that they found you know kind of using these terms compound hazard and uh compound event uh which of these topologies were they based on and multivaria is the huge uh lion in the room or uh Yeah, elephant in the room. Um, not sure why the cursor is there. Um, oh, that's just a from the screenshot. Uh, anyway, uh, so a lot of this is heat and drought. So, there's been a huge number of heat and drought papers, um, because it's relatively, you know, easy to understand, easy to quantify. Uh, some of these other ones you can see are kind of minor. Um so things are growing uh of course across all of these but Multivaria is yeah has by far the sort of largest um head of steam. >> Okay. I guess I'll ask if there's any questions at this point. Yes. Do you think that also reflects I think it's pretty much uh an example of um people following the trends. um the the 2012 IPCC report by Sonia Senatne and her colleague um I mean she's she's well known for her heat and drought work and was doing it 20 years ago. So I think that from that sort of nucleus things have ended up uh being like this. And also I would say that there's there's plenty of work on the other compound hazards that don't necessarily use those terms like natural hazards literature for example. Um, so yeah, it's kind of a phenomen. It's kind of a sociological phenomenon in my opinion that it's it looks like this. All right. Um, okay. So now I thought I would survey some of the approaches that people use um to study compound hazards as well as give a number of examples that maybe make this a little more tangible. So, and I tried to make this um a bit like qualitative. So, this is my kind of understanding of compound hazard um comp of of compound hazard science, I suppose, how it's practiced um and the mindsets and frameworks people use. So, I would say so there's going to be a few different examples within each of the categories. So the first category is kind of a physical analysis at the large scale. So one of those framings people use is they use the weather systems themselves, cyclones, ridges, fronts. This is a um an example of from a paper that was looking at um well using this meteorological approach. The image is from hurricane Sandy. So as you can see uh from the plot they are colllocating the areas of extreme waves, extreme wind and extreme precipitation. All of which of course played a role in the coastal flooding that occurred um in New York and New Jersey. And then the um like the the dark um the solid dark purple is where there's the the triple hazard. So all three of those were extreme, right? I think a percentile metric. Um and that I mean kind of the center of that is basically where the worst flooding occurred. Um so kind of you know straightforward mapping um but obviously driven by the storm the meteorological characteristics of the storm itself. Um you can also do this kind of thing for like um you know understanding where hot and dry conditions occur. based on ridges and the position of of atmospheric ridges and winds for example. Um another sort of approach is to use the seasonal interanual modes of climate variability for example ENSO uh you know the El Nino southern oscillation north atlantic oscillation um in the tropical regions monsoons um so in this uh paper that I was part of we plotted and the the what the x-axis is in days so basically the composite in the composite right around the time of the monsoon onset within a week or so after it is when the specific humidity rises and even though the air temperature falls a bit, the wet bulb temperature increases um pretty significantly and then kind of falls off as a specific community comes back down. Um so you know there's a very distinct sort of period when the meteorology of the monsoon drives the greatest risk of humid heat stress in northwest India. Um so you can also look at you know how um something like uh ENSO shapes the risk of you know global droughts or something like that. Then sort of somewhat distinct from that there are papers that really focus on the atmospheric dynamics that drive compound hazards. This is an approach that's most natural maybe for um for spatially compounding hazards although not necessarily the the most uh common um the most common uh framing I guess has been to look at uh the most common approach has been to look at rosby waves. This what this paper from Kaiicorn Huber and colleagues is about. So they you know his whole uh chain of research is around how there are different modes that the atmosphere favors. So like wave number five basically means there five troughs and ridges in a semi-stationary uh pattern which drive of course extreme uh heat drought risk as well as flood risk in certain regions per se and then also different regions at the same time. So like in this you know in this composite of wave number five um situations if you look at the bottom it's surface air temperature anomaly. So there's you know there's there's simultaneous heat risk when this pattern occurs in central North America as well as in Eastern Europe. Um and you know certain ones certain patterns like this have been growing in frequency of occurrence and duration and so on but that's it's it's fundamentally related to there's this type of large scale atmospheric dynamics and this is how it's changing and this is how it drives compound hazards. Um other things of this nature include like tropical easterly waves and their effect on tropical cyclones and heat and drought in the tropics. um sudden stratospheric warmings which drive again heat and and uh extreme precipitation risk. And then sort of similar to that large scale, relatively large scale dynamical view, but more focused on the ocean. um not as familiar with this literature, but people also look at you know how ocean waves create extreme conditions in the ocean which then of course feed into the atmosphere as well as of you know affect conditions in the ocean for for marine life um and uh storms and things like that. Uh people also, you know, so uh this example is kind of focused on upwelling and um variations in the um Hishio current. Um yeah, and then you know and this this is of course a dimension of things like Enzo with the the the Rosby and Kelvin waves that it sets off. Okay. Um yeah. And and then I guess the one of the points from the atmospheric point of view is that on the right side you have the 2 meter temperature and 2 meter humidity anomalies. So with this particular set of conditions, high SSTs driving high turbulent um heat flux on the left, you had both um really high surface um atmos atmospheric near surface temperature and specific humidity um which creates humid heat which is one of my focus areas. Um, okay. So, hopefully I'm not boring you too much with this whole list, but I thought it might be yeah, useful illustration of things. Um so another framing is around and and this is this can be done at the large scale but usually more local or regional scale is how the particular conditions often extremes of the land surface state drive um compound hazards usually through their interaction with some atmospheric phenomenon. So on the left you have the fire continuum quote unquote which is kind of similar to what was introduced in the aguchak um paper where you have basically say if you start maybe at 4 or 5:00 there you have maybe a fire you um which burns most of the vegetation then you have some um precipitation which recovers and the vegetation recovers then you have maybe more fire and so on um around and around and in California, this has been shown to happen on three to kind of 10 year time scales in the coastal areas. So, it's very like well established continuum of changes in the land surface state. And then if you look at the right, you can see that the um risk of flooding given um some spot on this land on this fire continuum is really strongly modulated. Um it in the paper they have different plots for different um different periods after a wildfire but um the other dimension is like at a certain time point the more land that's burned the greater the risk of really extreme runoff and therefore flooding because there's no vegetation to hold on to things um to hold on to the soil and so the erosion is very fast and so this is from a model um and in this particular um watershed the gray dots are the observed flood peaks. So in other words, if you plot if you if you don't consider this effect and this compound effect and you just plot the distribution of observed floods, most of which didn't occur within a year of a wildfire because wildfires are, you know, relatively rare and to have something within a year of a wildfires, an extreme flood within a year of a wildfire, relatively rare. Um so if you plot the observed distribution you kind of get okay the recurrence interval of 100 um cubic meters per second in the in the in the in the stream is something like n is something like 10 uh 10 or 11 years. But if you include the fact that you could have really um you could have this really unlucky situation of an extreme flood happen or extreme precipitation event happening within you of a wildfire that and and burning your entire watershed that adjusts the up the the peak flow that you would expect upward um quite significantly. All right. Hopefully I explained that somewhat reasonably. um is a bit complicated perhaps. Okay. So then kind of looking at the meteorology but at the at the smaller scale you can look at things like seab breezes um uh downslope winds, thunderstorms, many of the things that people mentioned were some of their favorite weather phenomena uh including myself. Uh so this is from a paper that was just showing like the humid heat risk in the Persian Gulf area is really tightly connected to the seabbze. You don't consider the seabbze, you're not going to get the combination particular combination of air temperature and specific humidity that drives the most extreme humid heat risk. Um yeah, and people also of course look at extreme inversions. Um downs slope winds are really critical for wildfire risk, that kind of thing. Um then there's hydrodnamics by which I mean things like river flow storm you know how storm surges and tides work um extreme tides interacting with storm surges in particular groundwater dynamics. Um this this is from a fascinating paper which um tracked uh in great detail this uh glacial lake outburst flood goff uh that happened in in Pakistan and yeah it's plotting like along the whole course of the river um all these flood damages took place so there was a really extreme um yeah there was a really extreme flood driven by really extreme um uh upstream precipitation which then created kind of these extreme conditions all the way down the river and landslides and other things. um and a number of people I think maybe several hundred people died uh okay so then I think this is the final category now so then there's you know stat statistical and and numerical kinds of approaches which I'm not going to belabor but of course there are many of these uh probably the single most common one and this is maybe the reason why heat and drought are so uh well represented in the literature is that there are many Python packages and things for using copulas to model heat and drought. So copulas are a way of a a numerical technique or I guess a statistical technique that um take into that takes into account both the individual distributions of two variables for example heat and drought as well as their interaction. So the probability of of say um a season or a month being in this corner up here is a function not only of what is the risk you know what is the 95th percentile of of heat and the fifth percentile of drought. uh but the fact that the uh say the fifth percentile of drought influences being in the fifth percentile of drought influences the 95th percentile of heat you know and amplifies it um in some places like Europe quite quite strongly um and then you know the what's what's plotted here is the um the crop loss of wheat in the Iberian peninsula but the basic point is yeah it's a bariate distribution that hope you know that um is kind of well established and there's different types of copulas of course. Um another framing marov and proson models um proson processes are things that happen uh kind of randomly in a time series. So like this kind of illustration is what what is the risk of event type A happening and and event type B happening within a certain time period of each other. This of course matters a lot if you're concerned about two particular things that have different probabilities happening around the same time and amplifying each other. Um, marov models are a way of measuring stoasticity basically. Uh, so it's uh when um the it's it's it's a class of models where the the current state depends only on the prior state and not the many states before that. So it's it's just a way of kind of doing a random walk through earth system um possible earth system states. Um another thing that is useful to know about generalized extreme value model fits. Again there's many ways to do this um to estimate like really extreme the point is to estimate really extreme return periods from relatively short observational or model simulations. So like you know we have data for all the stuff that's in in solid colors and then you know what is the rest what is the thousand year 10,000 year event um in terms of the thing we care about in this case minimum temperature. So this isn't really this isn't compound specific or anything like that. It's just useful for um it's useful for using in combination with with other tools. Um again because we're often focused on extremes when we're thinking about compound hazards. Um something that is a little more compound specific this whole class of things this paper um and I have all the literature um at the end I guess you'll have the slides so you can look these things up. This is a really good review paper on this um and other approaches. Uh essentially causal influence causal inference algorithms are ways that to use observational data first to build a network of how different um different elements of a system interact with each other and then to uh incorporate interventional data like we change one thing what happens to the to the other things in the system and then eventually to get to the point of okay if we have this whole complicated system and we change a few things at a time. How does the whole system and its impacts change? So this is very useful for compound hazards where we don't have maybe a clear priority idea of of how different elements interact. Um especially when you consider variations in space and time. Um so I'm not going to pretend that I know everything about this, but it's a useful approach uh for sure. Okay. Um I think this is the last category but I can't promise. Um so another I would say set of things is high resolution system models and usually kind of focused on relatively small scale interactions. So as opposed to you know global models which have a lot of assumptions and um average over wide relatively wide space and time scales you can really drill down to what is happening on the scale of maybe some kilometers or less for different types of compound uh drivers or hazards and this can reveal really interesting interactions. This paper from about 10 years ago was looking it was one of the best papers that um made the convincing uh argument and and quantified how droughts affect extreme heat. Um, and they were able to produce these plots, you know, that at a certain level of relative soil moisture, a certain amount of heat adection because this parcel that they're manipulating in the model heats up because the the energy input is going more toward um warming the atmosphere than is to evaporating the moisture which is no longer there. So you can kind of build these more conceptual pictures of what happens you know how compound hazards are really created. Uh okay so that then there's regional and global models which you can use to obviously aggregate things on larger scales. This is from a paper looking at uh the co the coincidence in consecutive years of drought followed by rain. So like basically wildfire risk to to some order of approximation with global climate change the risk of that increases in Europe. So it's just uh you know regional global models have have their uses as well because you can kind of average things and efficiently calculate risks at at large scales. Uh and then okay this is this is the last one now. So I've been quite interested recently in story lines and scenarios. I there are other cross-disciplinary approaches too but I sort of picked this one because I think they are really flexible and useful in this space. I'm sure they'll come up many times in the course of the next two weeks. Essentially they there's a blend of them ranging from quite quantitative ones like okay we had this flood and from a certain storm and what if we increase the storm precipitation by 20% because we expect that it's going to happen in 2050 and then what would happen you know as that propagates all the way through the the drainage system of the city or whatever and then there's there's quite uh quite qualitative ones um I was or a co-organizer of a workshop that happened last year. We developed these pretty handwavy storylines where a whole bunch of things for example happened in the Great Salt Lake and there was a toxic dust storm and you know people didn't prepare for it and it affects everything down to the level of like ski tourism and things like that. But if you do these things with one of the maybe secrets of like science is if you do something that seems kind of crazy but with experts then you can actually have credibility. So we actually had you know agency managers and stuff who contributed to these storylines and they were like yeah this is you know if such and such happened this is probably how we how we would respond and how the government would respond um and how businesses would respond and people you know are prepared for drought but they're not prepared for drought followed by whatever. So it's a useful way of kind of integrating across expertise and you can have multiple of them and all this kind of thing and then you can use them of course I guess coming back to like climate science um which is my specialty you can use those things to then inform what you care about right instead of just saying well droughts in Utah will become 30% more severe or something you can say well the risk of this particular scenario is going to become 30% more likely and is is more meaningful at least to a certain group of people and has more obvious um takeaways right to um for for the meaning um of the science. Okay. So moving on maybe I'll ask if there's any questions if I go to the third section. Yeah. Um, how much overlap is there between hazards and like the attribution worlds? Because I imagine probably a lot of the things that people want to attribute are like actually compound hazards, but I feel like a lot of attribution papers I've read haven't really like used that language or kind of that framing. >> Yeah. Um, this was a hot topic of conversation at a workshop I was at a few weeks ago. Um, some of the same people are in both worlds, but the most fair thing to say is there's not a whole lot. the attribution literature is a bit formulaic, I might say, and they haven't really developed good formulas for doing compound attribution. Um, and part of that comes down to it's it is hard to define what a compound hazard truly is. It's it's much easier to say there's a certain heat threshold and we have these simulations from these models and we trust them. So I think it's about model it's about developing the the frameworks which we can do and then it's also about trusting the models and the more elements you kind of add into the mix the harder it is to say here is like the answer and the then and you know here's the clear baseline that we had pre-industrially or 50 years ago and here is the the situation now. And that's why story lines and scenarios are so useful because it's difficult to quantitatively account for all the possibilities that go into especially the most damaging compound hazards. Yeah. But I think there's a lot of potential for attribution. I think I'm sure we'll see compound attribution become a lot more prominent in the next five years or something. >> Yeah. >> Um I have two questions. First, could you talk about the attribution literature because I'm not familiar with it. Um just Yeah. I I don't know what that is um as much. And then also I guess I'm curious about um how do you define what T0 is? How do you define what the first compound or the first hazard is especially when it's sequencing events? Because I imagine it must be hard because the landscape itself is shaped by meteorology. Um so yeah, I imagine that there's debate around that. So I'm curious if you could speak to that. >> Yeah. Um those are both excellent questions. Um for the second one um it's yeah it's really tough and it depends I guess the answer that you get depends on who you ask. So if you talk to heat people they'll say okay you know there's a heat wave and it started when the temperature or the temperature humidity metric went above the 90th percentile and ended when the metric went below the 90th percentile or you know some cumulative version of that. Um, if you talk to like sociologists, let's take the European heat wave, you know, that just happened, they'll say, well, there was another heat wave in May and then we just had this one the beginning of July and people haven't fully recovered from that, right? I mean, both from a sort of maybe physiological standpoint, from a maybe government resource standpoint. Uh, so those things are intimately connected. Um, but if you take the meteorological view, they're not really connected. different, you know, different uh meteorological drivers. Of course, they are connected in a sense of like will the warmer climate and that kind of thing. Um and maybe soil moisture didn't fully recover or something. Um and especially if you take the more systems view, I don't want to yeah take the systems like uh like eat this the social system perspective like lunch before it happens. But um if you take that kind of view that people will say in my kind of experience well you know we have the same like governance right and those kinds of things are a through line through everything that that's happening you know and and all the different tropical cyclones and um droughts and fires that impact like southeast US all kind of have a cumulative effect and the governance you know and the trust in government government is low and you know all this kind of Um yeah. So I think from the meteorological point of view, most compound events have something of a starting point where you can at least make the argument that a system is kind of near some resting state. There's also people use like the term poly crisis, you know, that everything is just becoming more extreme and there is no normal. And I'm not sure I completely agree with that. But yeah, I guess you know one the world is so complex one has to sort of start somewhere and usually that somewhere is something was not having an extreme impact and now it is having an extreme impact or could have had an extreme impact and hazy. Um what was the other question again? Sorry. Oh about attribution. >> Yeah. So the brief the brief answer is um people have uh groups have model simulations of various hazards and so they have a distribution of those hazards under a pre-industrial climate usually um is how it's defined. So you know we for a certain region in a certain month you know we have the distribution of like the temperature possibilities um then we've put the current like uh greenhouse gas and aerosol um loads and maybe the current land surface state into those models and run them again and say okay here is the distribution of this particular um metric over this particular scale in the new climate and what is the difference between those and you can measure that either by like the temperature uh you know the the the difference between the means of those distributions is whatever 2 degrees or the risk of exceeding a certain temperature threshold 35 degrees is increased by 50 times or something like that. So it's quite um sensitive to the model and the definition of the event. Like if you take you know recent European heatwave and you take a box over Paris right you might find the risk is such and such but if you take a slightly larger box the risk is different. So it's um on one on the one hand very well established and like hard to critique but on the other hand there are a lot of assumptions that go into it. Um like in the most basic version of things you know anthropogenic climate change has absolutely increased the risk of heat waves in Europe by but by how much >> and maybe people will comment more on this later. >> I actually don't think >> listening for me. Thank you Colin for explaining what glove was. Um but just if anybody's interested this coming Thursday actually the nationalmies is releasing a report on attributions of extreme weather events in the context of climate change. So there are a lot of resources that we could certainly point you to as well. Just wanted to add that to Colin's definition. >> Yeah. So when we're talking about defining the original event, do you think that given like how crossar just kind of would it be limiting or would it be helpful to have a standardization for defining the initial event? >> Yeah. Um, so when you're defining the original event, would it be helpful or harmful to have a standard system for defining the event given that it's so cross-disiplinary? Would that be is what I'm asking? >> I don't think it would be too limiting limiting, but I think it would be very challenging to get people to agree. >> Yeah. >> Um, but I think it's worth attempting also. Yeah, I was just going to add to your response that um setting baselines for extreme events like that and going towards a crossd disciplinary perspective it's um an ongoing discussion in the extreme event science world now because when you define an event someone might say what's the criteria for defining the event that way and why didn't you consider another criteria so that is where stuff like um significance testing and you sensitivity testing comes in. So if you're writing like a paper or doing a research in that light, it will be very very beneficial to have like significance testing and sensitivity testing your analysis. So it helps. All right, I realize I've been talking quite a lot longer than I kind of intended to. Um so I'll try to focus on what's most important here. And I also had some discussion questions. Maybe we can one or two of them. um in the interest of not being here forever. So um this next section I was going to get into I will get into disaster ingredients and by disaster I mean kind of compound hazard in together with its impacts. So most broadly you know any weather climate event whether it's extreme or not has contributions from the meteorology variability climate change across these different spheres of the earth as well as human um action and then to make a compound hazard basically some of these have to kind of be not in your favor. So perhaps based on the questions it's kind of useful to talk about these. So one of them I would say is hurricane Helen from the southeast US almost two years ago. So the the main story from this one is your cold front that led to heavy precipitation saturated the soils. Few days later hurricane Helen comes up. There's way more rain. Um it was I guess in inches. there's like 15 to 20 in some places 30 in that led of course to floods and landslides. Then you had various societal factors like infrastructure um socioeconomics, insurance and and lack thereof. And so there was a huge loss, a lot of deaths. And some of that loss and some of those loss losses and deaths could have been, you know, uh reduced by addressing the societal factors um generally before the event had happened. Um but maybe responding to it as well. Another one you might not have heard of or perhaps did if it was a number of years um before was big locust outbreak lasted three or four years across I think it actually went all the way to India but the main impacts were in East Africa and Arabia. So you had a heavy monsoon in 2019 I think that saturated the soils but then you had warm and dry conditions and apparently this is the combination that locusts uh that that makes grasshoppers become locusts. Uh of course there's a lot of poverty in that region there were the early warnings for this were not very good. So you had many impacts across um health, livelihood um and ecosystem. Another kind of more minor one, but this is a good like preconditioning example. So I threw it in uh last spring in uh Turkey, they had a warm March that led the fruit trees and the flowers to bloom early and then they had a real strong cold snap in April which um also had a lot of hail associated with it. can see this guy kind of mindfully looking at his I think tulips. Um so you know this train trade networks and so on that affected the outcomes which included mainly the large increase in food prices that lasted most of the following year. Um and of course the farmers in the local economies were strongly affected too. So um maybe we'll just do this one for discussion. So maybe we all can reflect on compound hazard like that you personally have experienced. What was it? What would you say was well understood or well managed about it and what maybe was less understood? So maybe like one kind of good and one bad aspect. So just think about that for a couple minutes, chat with your neighbor and uh we'll skip the other discussion questions for the sake of time. anything. >> Yeah, maybe just chat with your neighbor and then and then we can share out a couple examples. >> Uh just like two or three minutes. heard of the New York. >> No, >> it doesn't exist. Relax. Oh, no. [laughter] >> All right. Any volunteers to share what they've discussed? Any volunteers? I could also just stare at somebody. I can start. Um, I thought of a noraster that hit New York when I was a kid and it was on I think New Year's Eve and we got really heavy precipitation and wind and flooding and then it was followed by a deep freeze and so then everything uh all the streets froze. Um, I'm not really sure about um what was well managed about it because I was young and and wasn't really aware, but it was definitely quite damaging and and hard to travel afterwards. >> Yeah, it's a classic combination. And I'm from the Northeast, so I I'm familiar with that. Like heavy precipitation, high winds, and then extreme cold. When I was a kid, I guess one of the things that got me into meteorology is we had one of those and there was a there's an ice storm and then it was extremely cold. We lost power and it was very dramatic. Anyone else? >> Maybe one other example and then we can see where you are. Um, I was talking about an experience that I've had with fires, smoke, and heat where uh sort of the focus was on evacuation and getting ready to go, but sort of less attention or we knew less well what to do about, okay, we're feeling bad because the air quality is really sucky. Our power's out. Um, you know, do we open the windows because it's cooler outside than inside? Uh, but then we can't clean the air. So, how do we decide to do that when we're not at the point of like getting out of town yet? but sort of the individual level decision-m and the relative priority across these events happening at the same time that maybe have different feelings of urgency. >> Yeah, absolutely. >> Do you want to do one? >> Yeah, we can do one more for sure. >> Um I didn't give a super big chance. We ran out of time. Um but um so I went through Hurricane Helen. Um, and one of the things that I don't think was actually mentioned and isn't very well known about is that uh, Western Carolina and into Tennessee as well had a really big issue um, with bacteria in water outbreaks across Wells Springs and even municipal like city water. Um, and that was just one of several instances of like health outbreak issues. um especially that we have a really large elderly population in the area and so like I've met women who were having um urinary tract infections and it turned out to be ecoli in their water and there was no education there was no knowledge of that um and so that was a really interesting experience um and a lot of like underlying things are still happening in the region that just aren't really well known because it's very niche I should say >> yeah and these things are also rare like we may hopefully experience experienced something like that only once in our lifetime. So it's it's not the way that people usually learn, right, from experience when you're faced with something novel. And then also, yeah, a comment is frequently something that's frequently commented on, but not still really addressed about compound hazards is the long tale of recovery when the media attention and the emergency funding have gone away, but there's still like an aftermath often in the health space or something like that or infrastructure. Okay. So, I'll skip these other scintillating questions. Um, all right. So, I want to talk through and I don't have too much time, but I want to talk through a little bit about the LA fire and how it kind of encapsulates LA fires and how they encapsulate some of these compound elements. So, it's a huge disaster, 200 billion or something and losses, 440 deaths, the really really strong hurricane strength wind gusts. The compound nature of it included that there had been essentially no precipitation leading up to this. This is early January. Um then you had really strong winds, very dry air, and you had then multiple fires at the same time in LA County. So it burned all this area in dark uh in dark brown. This is satellite image. Um the blue line is about what what burned. and the same slightly different scale, but the black line on the bottom is that same area that was burned. And you can see that the Calire threat map didn't even include uh a lot of the burned developed area. So Cal Fire, in their opinion, there was essentially no risk of these neighborhoods burning that then burned under these extreme conditions. And the updated map isn't much better, by the way. So then this kind of fits into this general framing of quote unquote surprises. Something I've been thinking about quite a bit. Many elements to a surprise. Headlines from across the world about how different things were supposedly surprising or maybe actually surprising but shouldn't have been as the one indicates. So when I mean surprise, I kind of mean the severity of each hazard by itself. the compound or multi-hazard nature of those things as well as other systems like human systems interacting with those compound hazards to produce yeah surprising and and disastrous effects. The the actual potential for extreme wildfire behavior was very well forecasted. If you follow NSF uh sorry uh National Weather Service NWS um bulletins, this is like as dramatic as they could possibly be. Uh this is from the day before. Um and then a few things that happened that really I think encapsulate the challenges of compound hazards um and the interaction again with you know human systems um emergency response planning emergency response that kind of thing. So I boxed um the most relevant parts. So one of them is that they as kind of the icons at the bottom right indicate that it turns out and I had no idea living in LA that this was true, but it turns out that they basically were completely counting on being able to drop water and flame retardant on the on the fires to slow them down so the firefighters could put them out just as a general method of firefighting. Well, the winds were so strong that for for firefighters safety, they couldn't fly either the helicopters or the planes. So then the fire, you know, not only did they lose the ability to fight the fires, they also lost the ability to monitor in real time where the fire was going. So this was, yeah, extremely devastating. And almost all the destruction occurred while the winds were so strong that they couldn't fly. As soon as they could fly, they immediately kind of knocked the the fire stopped moving so quickly and they were able to contain it. Another thing was the multiple fires happened at the same time. So the one fire, the Palisades fire started around midday and the the Ian fire um further east started in the evening. So they had dedicated LA County's most of LA Countyy's copious fire resources to the one fire. Then the other fire broke out and then there were a few other smaller fires too. And the LA fire chief literally said, "We prepared for one or two but not four." So they just didn't have enough people. And it was really surreal if you saw the coverage of this like media people wandering streets where all the houses were burning and there would not there wasn't a single firefighter around. There were just so many so many firefronts that even in LA County with its 20,000 firefighters or something they still couldn't cover them all. Um and then another one um was failures of the emergency communication system which we supposedly have and supposedly test. Many people in particular neighborhoods didn't get any alerts. There were number of false alerts. The fact that the power was off and people's devices might have died was maybe another factor. Um there's just so many things that went wrong basically and contributed to the devastation and the loss of life. Um so kind of to summarize then you can you have this set of extreme hazardous physical drivers like fire weather. So like yeah so mo soil moisture and vegetation moisture is very low. Wind speeds are very high. Then you have other things um like no one apparently thought about houses and like paint and gas burning. You know uh a lot of the houses were old so there was lead in them. Um, yeah, this reservoir was didn't have any water in it because it was under maintenance. Anyway, so there's all these other factors in addition to the ones I mentioned before that can implode a system that's already primed for disaster. We have this phrase disaster waiting to happen, which maybe is really appropo. Um, so the ignition point was at the top right. The winds were from the northeast and many I mean 9,000 houses were burned in that fire. entire neighborhoods were basically leveled. Um yeah, so there's a common a commonly used schematic in the disaster kind of risk community including at the uh United Nations level which the different versions of but one of them can kind of be represented by this. You have hazard at the bottom, exposure and vulnerability, the responses to the hazard obviously including the exposure and vulnerability and then you have the impacts and you have a it's a whole cycle and these things kind of like obviously a lot of the time you go from hazard to impact via the other things but they can also feed back on each other. So you can enter the cycle in different places and it can spin around and then that maybe is situated within the pre preconditions across different systems and then as I was kind of mentioning about the long tail you have the post- disaster outcomes again across different systems which can modify this risk cycle over time plus other climate non-climate drivers like you know climate change for example Um yeah so like with story lines for example the question we kind of pose is what interactions might push systems past their breaking point. The usual way of approaching you know changes in climate hazards is what is changes in compound events? What are the changes in climate hazards individually and collectively and how does that go through the cycle but you also might just start with the cycle as a whole. This is I guess getting at the what is the appropriate scale and approach question and a bit of the systems gesturing at systems thinking. Um so you can both of course when you ask this question about pushing systems past the breaking point you can of course generate new risk knowledge that way. Um you can also better operationalize existing knowledge that's siloed within different comm um communities of research and practice. Um so uh one of the most common kind of phenomena within this cycle is reservoir and levy effects. So when you build kind of illustrated at the bottom there when you it's well documented that you build a levy people then kind of like you know the phenomenon that people like football players are more aggressive when they have helmets compared to rugby players right when people have a levy then they start build they say oh I'm safe then they build things closer to the water that they wouldn't have built otherwise and then things are maybe fine for a while but if you have a really extreme event like say hurricane Katrina then you have much more devastation than you would have had if you hadn't had that levy. So, it's obviously kind of a fra ethical um and economic and whatever problem. Um but suffice it to say that there are these things called levy effects or maybe reservoir effects on the on the drought side and those um are modulated or those modulate exposure and vulnerability primarily um I guess the hazard maybe as well. Then you have sort of psychological effects like recency bias for example which can contribute to certain uh problems in the Orville dam crisis in 2017. It was documented that the dam operators were trying to conserve water because California had just been in a huge drought, but it was a flood and the flood was it turned out quite severe and so they were kind of pushing the operation of the dam toward conserving water when they really should have been trying to get rid of water and the dam um was damaged and yeah, there was a spillway. There's a spillway that was damaged and people worried about the dam collapsing because of how the dam was operated. But that was a understandable kind of psychological outcome of the dam of the the recent drought that California had just been through. So that affects the responses and then you have um yeah kind of things that involve the hazard, the impact, the vulnerability kind of all interrelating um to cause devastating impacts. And just some examples of those like just take the bottom one I guess meanwhile some communities in response to the water scarcity had migrated. This is kind of similar to the drought one to the dam one just a minute ago. The the communities have kind of been scarred by a drought, so they moved closer to water, but then because they were closer to water than they were usually when the floods came, they were more affected. Um various other things like at the top example about vegetation being damaged by the drought, the soil was compacted leading to more runoff increasing the severity of the flood. So you kind of have to look kind of quite closely just to understand some of these important dynamics of compound hazards. You're not going to get these things from running your global model. Um but it also makes it a bit hard not impossible but hard to generalize um because the same I mean you don't have transient communities um in you know I don't know Southeast Asia the same way you do in East Africa. So you're not going to have like a migration effect on exposure for example you have other things. Um just kind of skip this. Um maybe I'll do it actually. I guess I'm almost at the end. Um so another example of compound dust storms. Just another kind of example about dams and drainage and agriculture. Then you throw in the climate a drought because you have irrigation dependent land that's not being irrigated and a drought then you have a lot of loose soil and that can be lofted by strong winds to create severe dust storms. So it's a whole set of interactions that lead to that are necessary really and sufficient for creating uh severe dust storms. Okay, in in the case of Iraq anyway. So this is a modification of a figure that I made um a few years back where we kind of were this is another example of the kind of hazard impact response feedback loops. Um the x-axx is just upon further reflection I I feel like exposure and vulnerability don't really directly create hazards but they do create impacts and they also are affected um by the responses to things and on the bottom. So comp combinations of physical drivers basically lead to impacts uh hazards impacts responses. Then you have sort of both um uh yeah you have exposure vulnerability of physical drivers. Then you have environmental decisions, human decisions um that affect the hazard and then of course the response is a human decision that flows essentially from the impacts. Anyway, um so maybe that's not useful, but um yeah, I think basically the the point I'm trying to make, I guess, is that a lot of time compound hazards are often thought about only considering a few of these boxes, as I said at the start. And it really is important to try to at least in very rough terms approximate what the chain what each of these terms in the chain kind of approximately are before you start your analysis. Um yeah and these things are also time varying. Um uh okay so I guess I'll kind of wrap up with the the questions and the kind of direction of compound hazards. The most commonly framed question and the most proaic one is you know how can we improve research and practice around you know hazards and impacts and responses and environmental decisions and all this kind of stuff. And then that ranges through then you know how do we like learn from events? How do we keep the memory of the lessons of events alive all the way to you know how do we change our risk attitudes so that we're valuing things appropriately. So the example I'll use to maybe capture your imagination a little bit is this is from an event 2004 called Hurricane Pam which caused this whole range of events kind of multi-hazard event I guess you can say compound event um but it was actually fictional it was just a s a scenario exercise put on by this institute called the institute of emer emergency management I think which simulated a strong I think category 5 hurricane hitting New Orleans. Then a year later an actual category 5 hurricane hit New Orleans and the event and the impacts are actually remarkably similar. So question this begs is well well if they had the knowledge all this stuff was going to happen why didn't anyone do anything? The answer is a few things were were changed like people in the aftermath of hurricane Katrina said yeah that actually participating in hurricane Pam exercise which included like the city and state emergency managers and the state government and all the all these people like it did help us like know that we should preposition you know medics and that kind of thing but the the larger at the larger scale I would argue this didn't really do that much because the levies still broke you know uh and there still a huge um problem with the recovery of neighborhoods even decades later. So um maybe if it had been 10 years before Hurricane Katrina's impacts if Pam had been 10 years earlier before Hurricane Katrina, maybe things would have been better, but I'm not encouraged. So um I think it speaks to how our decision-m processes are not really set up to manage the most extreme kind of compound impacts. Um because something like levy maintenance is not directly connected or motivated by the fact that uh you know we could have 40 or 80 billion dollars in in damages. There's a million examples of this kind of disconnect between people who have the power to do something and the people who are kind of faced with the aftermath. And this is true for every kind of you know climate hazard but especially I think for compound ones because of their inherent complexity. Um, and this is uh one of the closing um images here of Sunset Boulevard and um during the well the photo was taken just after the Palisades fire where they had warnings the fire was coming but it's moving so quickly uh that people all had to evacuate on mass and because people are evacuating on mass and it's LA there's a huge traffic jam. Was it really anticipated there's going to be a huge traffic jam? really. And people actually had to abandon their cars because they were stuck in this huge traffic jam and the flames are coming their way. Um, and sadly, you know, there's this is kind of been LA as a as a prototype or um as a as a place on the front lines of climate hazards has been known for a long time. There's this really incredible um LA Fire Department video from the early 60s that I recommend uh watching. It's very 60s in a way called design for disaster and it's all about how like the land use and emergency systems are not set up for big disasters um affecting tens or hundreds of thousands of people at the same time. Uh and yet 60 years later we're kind of making the same mistakes. Um okay so in my last couple minutes here uh you can read some of this on your own time but um what what should I highlight? Um yeah, nonlinearities in compound hazards I think are a real frontier. We have a lot of regions we don't know that much about that. You know, part of this sort of sociological element of this is that we're lacking quantitative and qualitative data that we need to even know that events are happening much less to sort of understand them. A lot of dynamics. I kind of won't get into that, I guess. um assessing I I think we need better ways to assess what kinds of compound events matter. So we have physics based models that we usually use. Expert judgment is something that is brought in with like story lines or scenarios. But I think as the Pam hurricane Pam example illustrates is under underused especially from domains that are not directly connected to um physical science physical science or meteorology. And I think that's why this you know workshop is so great bringing in some of you who have really more divergent expertise from my own. Um obviously community traditional knowledge so many times you talk a handful of times I've gotten to talk in a structured way to people who have traditional knowledge or you know live in communities that have been affected by disasters. They say oh yeah it was obvious that such and such you know would play out that way but you know how many times is that in a peer-reviewed article right? Um and then yeah I think as I already uh referred to yeah these areas of exchange are across different disciplines of research in practice and ways to incentivize them give people credit for them addressing key bottlenecks are really really critical. Okay so um this is the final slide now I promise. So um I'll I'll talk a bit about like some of these things in the practices panel tomorrow but you know funding knowledge silos are both really critical issues. Um yeah decision making integration um communications and trust really for the most rare and complex events are especially challenging. Um and also yeah people's lack of um personal experience is maybe part of that you know you don't kind of you kind of bewildered and I was um in the evacuation zone for one of these fires so I yeah I got to experience what that was like very bewildering um and then getting down to the bottom two points there I mean compound events compound hazards are increasing really dramatically in many ways and that I think there's a fundamental time mismatch between those events both kind of the quantitative change as well as the attention that we're paying to them compared with the plotting pace of most assessments in the scientific and technical space and yeah the intrinsic time scales you might say I'm not sure yeah there's no answer to any of these things but just issues that come up you know when trying to understand compound events and impacts better. And then and then yeah also like these things are uncertain that affects the communications and trust that affects um how we how much we can say with confidence and that yeah it tends to encourage us to retreat I think a little bit to our corners of expertise but it's important not to retreat and that's why this workshop is so important to give you all the skills and that's I'm sure it'll help me too for the days I'm here to have the confidence to um the confidence in the tools to to address some of these things to the extent that we can. So there's the literature to look at later and thanks a lot >> before you go. So >> Colin didn't think he was going to use his 90 minutes, but he had a lot more to say, which I think we're all very grateful for, Colin. So thank you for that. And I love this take-home point. It's important not to retreat. when is something outside of our own um deep expertise. Mariana, may I steal two or three minutes from our morning reflections? We'll sort of segue into that. Colin, are you willing to take a few kind of questions more reflecting on the entire talk from anyone? Yes, >> here. >> Thank you so much for the the presentation. uh at least we know how to read uh scientific literature right now. Yes. Uh my question is um I like the question you raised that what interaction might push the system back from your presentation. I noticed that there is a pattern when it comes to this um area of research is always from trying to know the hazard behavior know the impet and the response. So my question is it's always like a top bottom interaction and I know that when you talk about cross disciplinary and all that I said oh this is it this might bring the interaction from the woman's perspective angle but I kind of see that it's still like at the emerging level in order to know how we can bring in this um system is it possible going with the trend of research is possible to change this from the bottom to top. Not looking at the expert view alone, looking at the users because they the one that have this major impact. Is it possible going forward? Thank you. >> Yeah, this is the kind of thing that is more considered within the domain of natural hazards I suppose. So that really more integrates domains of expertise like human geography, sociology, um governance and planning. Um you know my like I guess I can only speak to my own motivation and interest which is to use the insights from that kind of work to inform physical to to inform physical science to sort of optimize the questions we're looking at and the tools we're using. But um in the end when you have a hammer everything looks like a nail. You know we have these physical models people want to use them to to understand the propagation of risk and there are there's so much space one of the motivations you know for me is there's so much space to better design our studies to take into account those kind of bottomup interactions like you're talking about. But I I would say that I'm not an expert in that and that's more what natural hazards as a science is kind of about. Um I identify more as a climate scientist. So I want to use that and be motivated by that. Yeah. Hopefully this is a good kind of grounding from the physical science perspective, but it's certainly yeah not the only perspective or the most valuable one or anything like that. >> And hopefully to think about partnering with people who have that different kind of expertise as you're thinking about your modeling, right? like closely and as part of teams. That's sort of like where the future of this is going to kind of co-inform each other and blend those ideas. >> Absolutely. >> Yeah. Maybe one more question. >> Yes. >> Thank you. My question is on the scenario example of hurricane Pam and Katrina because I also work on scenario based research and I was curious. So you mentioned like the time element that it was only a year before. Were there any other sort of lessons I guess like was there anything that was improved based on the scenario exercise or was it just considered to have been ineffective? I hadn't heard of this example before. So just curious to hear more. >> Yeah. Yeah. And there's a really nice afteraction or after exercise report that came out um last year like yeah on the 20th anniversary of Hurricane Katrina I think um I think I think they the emergency managers who wrote this report said that they thought the evacuation was better organized that hospitals had beds prepared. Uh that kind of thing but it was pretty it was pretty marginal. Yeah. And while it's possible that more time could have led to better results for Hurricane Katrina, it's also possible that that was kind of all that was going to be done realistically cuz it's a perennial problem, right? To there are many things that could happen and people aren't generally motivated politicians, utility companies. Another good example is undergrounding of utilities. There's been a lot of examples of power lines starting fires in California. They're still reticent to do so even though the the payoff economically is like 100 to one or more between the cost of the damage from the wildfire and the cost of underground and the utility. But unless someone is sort of forcing them to make that that investment for the benefit of everyone, they say, "Well, why should I spend 50 million to do this? It's not going to benefit me. It's right." So there's a sort of [clears throat] there's a sort of regulation and cultural norm aspect that's very strongly ingrained um behind that that underlies a lot of these compound hazards and their responses. >> Let's thank Colin again everyone. [music]