Climate Change and Health Initiative: J. Jason West

October 08, 2019

"Connecting Climate Change, Air Pollution, Energy and Health"

Information

ID: 4532
To Cite: DCA Citation Guide

Download Transcript

00:00(students chattering)
01:55- Okay, let's get started.
02:04Let's get started.
02:05Can you hear me okay at the back?
02:07Yeah, okay, great.
02:09So I'm Robert (mumbles) I'm a professor of epidemiology
02:13in our Department of Environmental Health Sciences.
02:16And I'm also the Faculty Director
02:18of the Yellow Climate Change and Health Initiative.
02:20And we're very pleased today as our first speaker
02:24of this academic year to have Jason West,
02:27who's from the Department of Environmental Sciences
02:31and Engineering at the University of North Carolina School
02:35of Public Health.
02:36And we were just talking about how few
02:40public health departments have engineering in the name
02:43and how actually valuable it is to have engineers
02:47within schools of public health, as hopefully,
02:50I think you'll see when you see the work that Jason does.
02:55So Jason, has a great publication record
02:58he's published in the high impact journals like
03:00Major Climate Change, and Nature Geoscience
03:03and Environmental Health Perspectives.
03:06He's also had funding from a variety of sources,
03:09including the EPA has a the National Science Foundation,
03:15and the National Institute of Environmental Health Sciences.
03:19And so he's got, as you know, he's gonna talk
03:22to you today about connecting climate change,
03:25air pollution, energy and human health.
03:30(students applauding)
03:34- So I'm really happy to be here today.
03:36Thanks for the invitation.
03:38I spent yesterday, an exciting day for me in New York City
03:41for the climate week, and--
03:44- Sorry, (mumbles) to just the lights a little bit.
03:47- All right, yeah.
03:48I was just gonna say, I was having a hard time in my mind,
03:53justifying flying up here just to attend
03:55a climate change event in New York.
03:58I had thought instead of about maybe taking a sale
04:02(students laughing)
04:03but then I contacted Rob who had already invited me
04:06and asked him if we could combine my trips.
04:10And that worked out really nicely.
04:11So Rob, if nothing else, I should thank you
04:13for making me feel less guilty about flying.
04:17Okay, so I'm gonna talk to you today really,
04:21this is a talk not on one theme,
04:24but I'll be talking about a lot of the work
04:26that I in my lab is done over the past decade or so.
04:30I'll motivate that in a minute by talking about
04:34especially the human health angle,
04:36that the work we do is really pretty interdisciplinary.
04:39And I think you'll see that
04:40so I work on climate change in air pollution.
04:44My main entry point to climate change is
04:46through atmospheric science, which is kinda my background.
04:50But in particular, this interest in climate change
04:54has kinda taken off in connecting climate change
04:57with air pollution.
04:58So as climate changes,
04:59what will that mean for air pollution?
05:01Or as we take the necessary steps to address climate change,
05:04what would that mean for pollution and for health?
05:07So those are a couple of the themes that are explored here.
05:10I thought I'd start with this paper,
05:12which Michelle (mumbles) here also contributed to.
05:16That appeared a few years ago.
05:19I and colleagues this, if you look at the list
05:22of authors here, this is a purposeful combination of
05:26air pollution scientists
05:28and air pollution health effects scientists,
05:29we all got together in a room and talked about
05:32what were some of the big issues of our day
05:35trying to take stock of what's known about air pollution
05:37and health, and what are the big opportunities
05:40for the future.
05:41Some of our main conclusions I've pointed out here,
05:46one is how important air pollution is
05:48for global public health.
05:50And what's been really instrumental in coming
05:53to this understanding has been
05:54the Global Burden of Disease Assessment.
05:56So as I go along, through this presentation,
05:59I'll show you some results from the
06:00Global Burden of Disease Assessment
06:02and show you how my lab is doing some work to contribute
06:05to that assessment
06:06by mapping global surface ozone concentrations.
06:11Air pollution, it's health impacts our changing globally
06:14and will change in ways interrelated with climate change.
06:18We looked also at air pollution science,
06:20which is making new possibilities through
06:23new ways of measuring air pollutants,
06:25measuring new chemical constituents that may be then
06:28we could put in epidemiological models to find out
06:31what component of air pollution is most
06:33important for health.
06:35We also have cheap sensors that can be widely deployed
06:38and are being widely deployed,
06:40providing a lot more information,
06:42even if the quality of those measurements is poor.
06:45We have satellites looking down at the world now
06:48giving us information every day about air pollution
06:51that's potentially useful for us.
06:53And computer models and that's what I do
06:56are becoming better for this kind of application too.
07:00One of the reasons why I wanted to start off with this,
07:02(mumbles) was we took some time in this article
07:05to talk about the need for the air pollution science
07:08community to work better and closer together
07:11with people that work in air pollution,
07:13health effects science.
07:14So when I think back to when I was a graduate student,
07:18I was firmly in the air pollution science world,
07:21I was not exposed at all really to help.
07:24And as I look out at even our air pollution science
07:27meetings, those are changing that I now see
07:29more presentations from health effects scientists
07:32or people that are making this bridge
07:35between air pollution science and health effects science.
07:37So that's a healthy change,
07:39but I think we have a long way to go still.
07:41Okay, in that regard, and maybe some
07:43of you will be interested to, in your career fill that void.
07:46Okay, my plan for today is to
07:50it's sort of the buckshot approach (mumbles)
07:52I'll talk about a lot of different themes,
07:54and we'll see if any of them stick with you.
07:57But first, I was gonna talk about global ozone
08:00and what drives global ozone changes?
08:04This is more atmospheric science.
08:05But the rest of the talk will be about,
08:09about air pollution and climate and health.
08:13So how many people die each year due
08:15to exposure to ambient air pollution?
08:18How can we best model global surfaces on distributions
08:21that's for the Global Burden of Disease?
08:23And I'll show you those results.
08:24How will climate change affect global air pollution
08:27and air pollution related deaths?
08:29So now turning our attention to climate a little bit.
08:33What are the trends in air pollution related deaths
08:35focusing on the United States?
08:37And the last question, if we slow down climate change,
08:40what are the benefits that we would see
08:42for air pollution and health, okay?
08:46Good.
08:48I'll talk a little bit about ozone.
08:50So I'm guessing many of the students in here (mumbles)
08:54from of a public health are studying public health
08:57and maybe don't know a lot about ozone
08:58so let me talk about that.
08:59So ozones forming the atmosphere by an interaction of
09:06non-methane volatile organic.
09:07So organics that come from motor vehicles
09:10from all kinds of different things, carbon monoxide as well.
09:15Trees emit volatile organics,
09:17those drive this cycle of radicals.
09:19The other important ingredient is nitrogen oxides,
09:23comes from motor vehicles and power plants
09:25and heavy industries, in the presence of sunlight
09:29gives us ozone.
09:29So the three important ingredients are in organic
09:34(mumbles) sunlight, and out of those chemical reactions,
09:37we get ozone.
09:39I'll be talking as well on the global scale.
09:42And when we look at the global scale, these fast reacting
09:46organics that are important in a place like Los Angeles
09:50for producing ozone very fast because these react
09:53on the order of hours are not very important than
09:57on the global scale.
09:58It's these more long live compounds.
10:00So carbon monoxide is really an important
10:02methane really important.
10:03Okay, so methane is admitted in large quantities,
10:07but it reacts so slowly contributes very little
10:10to urban air pollution.
10:11But on the global scale methane is one of the big drivers.
10:15Okay, so and by the way, methane and ozone
10:19are both greenhouse gases.
10:22So going back several years, I had a line of research
10:24looking at how emissions of these different precursors
10:28would affect both methane and ozone thinking about
10:31how do you control those, both from an air pollution point
10:34of view and from a climate point of view, okay.
10:37So as you motivate the first study here.
10:41We're interested in here in
10:44how global emissions are changing.
10:47This shows global emissions of nitrogen oxide
10:50one of those compounds that reacts to form ozone.
10:54Globally, in 1950, and I'm gonna flash forward
10:59by decade now, so in 1950, 1960, 1970 and 1980.
11:05So by the time we got to 1980,
11:06you see the emissions are dominated by the U.S and Europe.
11:10The spatial distribution, this is the latitude
11:12and they'll distribution on the right here,
11:14that hasn't really changed as emissions grew.
11:18But after that period, then this is 1990, 2000, 2010,
11:24we see emissions going down
11:27here in the U.S and Europe as
11:29we've implemented air pollution controls.
11:31And they've gone up pretty dramatically now
11:33in China and India.
11:34So the emission distribution is shifting southward.
11:39This is interesting, and perhaps troubling,
11:42because we understand from the point of view
11:44of atmospheric science, that a ton of emissions closer
11:47to the equator is expected to cause more ozone to be formed.
11:51And so we're asking the question here, basically,
11:55we'll focus on this period 1980 to 2010.
11:58So 1980 years before we had this change in the spatial
12:03distribution with emissions coming southward.
12:07We're gonna separate out the importance of the magnitude
12:10of the emission change versus the spatial distribution
12:14of the emission change.
12:15And the third ingredients here and the third factor
12:19that's really important is the global methane change.
12:22And we're gonna see how important each of those is
12:25for global troposphere ozone,
12:27that is the total amount of ozone in the lower level
12:30of the atmosphere, okay.
12:33So using a computer model, so I'm a computer modeler,
12:36and I work with models of the global atmosphere.
12:41We separated out these different influences.
12:43So according to our model, this is how the total ozone
12:47distribution has changed.
12:48Where it's increased the most is an indicator of
12:52where the biggest growth in emissions in the ozone
12:56has taken place, especially South and Southeast Asia.
13:01And then the contributions to this total.
13:04So 28 Teragrams of ozone contributions from the change
13:09in spatial distribution, the magnitude change
13:12and the methane change, these two on the bottom,
13:14though they contributed to the total amount
13:16of ozone present, have very little ability
13:19to explain this pattern of the total lows on growth.
13:24But if we look at the spatial distribution change,
13:27we have reductions in ozone, reductions in emissions,
13:32I should say, from the U.S and Europe,
13:33but pretty dramatic growth in South and Southeast Asia.
13:38And this gets us a lot further at explaining
13:42this total ozone growth.
13:43We were actually surprised by this that this is over half
13:47of the total, bigger than the effect of the magnitude
13:50and the effect of the methane change.
13:53This is another way of looking at this where
13:56this is the I should stay close to the mic,
13:58I'm told because we're recording.
14:00This is the equator, the North Pole, the South Pole,
14:03and then looking through the depth of the atmosphere here.
14:06This is the total change, the spatial distribution change,
14:09the magnitude change, and the changing global methane.
14:13In all of these cases, I should say in these two
14:16on the bottom, again, you don't explain the pattern
14:20that you see in the total ozone change.
14:22And this helps us to explain why this is so important.
14:25So as admissions have shifted, further southward,
14:29close to the equator now, those emissions are being lifted
14:33up by deep convection, we would say in a (mumbles)
14:37meteorological sense, reaching a higher level
14:40in the atmosphere than they do here.
14:42Once those emissions become part of the upper troposphere,
14:47they live longer, and they react to form ozone.
14:50That's what's driving this greater sensitivity
14:53of ozone to changes in our pollutant emissions
14:57near the equator.
14:58And you can see that really vividly here
15:00that these emissions that are from Southeast Asia in India
15:04are being distributed, lofted up very high,
15:08where they're reacting to form a lot of ozone.
15:11So our concern then was that as we shift
15:17and continue to to shift emissions toward the equator,
15:20that even if global emissions might decrease,
15:23if we're if the spatial pattern is changing,
15:25we might continue to increase global ozone.
15:29This was the work of Yuqiang Zhang who is my PhD student
15:32and that postdoc, he's continued that do a bunch more
15:37simulations where he's separating out then the influence
15:40of each we're looking again at the change from 1980 to 2010.
15:45Looking at the influence of each world region change
15:49on the total ozone change,
15:50and here's the methane change as well.
15:52So this is the total effect.
15:54And we see here that East Asia is important, that's China.
15:57That's not surprising, they led the world in
16:01manufacturing with huge emissions associated with it.
16:05What is surprising here,
16:06is right next to it is Southeast Asia
16:08as important for globalism.
16:11And if we look at the emissions,
16:13the emissions from Southeast Asia are much smaller
16:16than the emission growth that's taken place over
16:19these three decades from East Asia.
16:22So we're really highlighting here how important
16:25emissions are, that are near the equator,
16:29and in particular, from Southeast Asia, suggesting
16:32that there really are sort of emission hotspots
16:35where each ton of emissions has a much greater influence,
16:39on global air quality than emissions
16:42from further north, okay.
16:46So that's your bit of atmospheric science today.
16:49I'll turn our attention to health.
16:51And our first question will be,
16:53how many people die each year due to exposure
16:56to ambient air pollution?
16:57I'm gonna take a minute and get into that.
16:59So, Rob introduced me as an engineer
17:02and my background is engineering.
17:04I had no schooling and public health
17:06had no idea what public health was about,
17:09really until I did this study,
17:13I had been for a few years,
17:15I had a fellowship to work in UPA headquarters in DC.
17:19So there's a fellowship program for PhD scientists
17:22to go into government offices.
17:23And I thought at the time that I'd be leaving academics
17:26for good to pursue a career in policy.
17:29And I learned a lot about how people
17:33formulate policy questions in a place like DC.
17:37And that changed how I approached problems.
17:40So I became interested in health.
17:43Health is an interesting topic, but my main motivation
17:46actually was to think about it from a cost benefit
17:49of policy analysis point of view.
17:51The health was, to me the benefit
17:53of the cost benefit analysis.
17:55That's why I wanted to study it.
17:57So my first study, there was I became aware as I just sort
18:01of explained to you that methane affects
18:04the global background of ozone.
18:08We had been thinking about methane,
18:11obviously as a greenhouse gas.
18:12And there's good reasons to reduce methane
18:15as a greenhouse gas.
18:16I thought I look at it in different contexts.
18:18And I asked the question, could we justify
18:21reducing methane emissions, because of it's reductions
18:25in ozone, and the health benefits
18:28that would come about from that?
18:30So this was published in 2006.
18:33I called up Michelle Bell, who had the number one paper
18:36at the time on ozone related deaths
18:40and I talked through with her.
18:41How do I use that information in
18:44what I'll call now a risk assessment?
18:46So using epidemiological information to assess health.
18:50So what I did here was I use my global atmospheric model,
18:54put in a simulated a 20% reduction
18:57of global methane emissions,
18:59overlaid that on the world's population,
19:03and found that the reduction who knows on that came about
19:06from reducing methane avoided about 30,000 deaths in 2030.
19:12When I put dollar sign associated with those deaths,
19:15and compared it against the cost of reducing methane,
19:19and I could look up from the climate literature,
19:22the ways that we could think about reducing methane
19:24and how much it costs, I found actually that the benefits
19:28to health outweigh the cost.
19:30So that was kind of cool.
19:32And it's suggested that we could be thinking about methane
19:34controls from an air pollution management point of view,
19:38as well as from climate change management point of view.
19:42Okay, but one of the things that I was only vaguely aware
19:45of at the time, this was actually the first time
19:48or certainly one of the first times that anybody had used
19:52global atmospheric model
19:54to drive a health impact assessment.
19:56And what I wasn't anticipating at the time was,
19:59that would be that the major direction of my research
20:02ever since that, okay.
20:04So what I'll talk to you through now
20:05or some more applications,
20:07where I'm using my global atmospheric model,
20:10or using models that are used in the community
20:12that I came from, and now using them
20:15for Health Impact Assessments.
20:18So the question that I asked just go back a couple slides
20:22how many people died prematurely due to exposure
20:24to outdoor air pollution every year?
20:28If we look back several Global Burden of Disease Assessments
20:31ago, the first answers to those questions only looked
20:35at cities because it was in cities
20:37that we had observations
20:38we didn't have observations elsewhere.
20:41And so they were only estimating in the
20:44Global Burden of Disease, the effect of air pollution
20:47on health for the fraction of the world's population
20:50that lived in the city, ignoring everybody else,
20:53but we know where pollution is going up in a lot of places,
20:56even rural places.
20:58So our first attempt at doing that was
21:00that you use a computer model, the computer model
21:03has an advantage because it's got
21:05complete quote global coverage.
21:07It's got disadvantages, of (mumbles) grid cells
21:10that don't really tell you what people are breathing
21:12in an urban setting.
21:14And it's got biases, okay.
21:17But nonetheless, we used it and that gave us the first
21:20estimate of global air pollution related deaths
21:25as a global total.
21:27Here was the next study in that line.
21:29This is Raquel Silva, who is my PhD.
21:32I use a bunch of chemistry and climate models.
21:34These are simulations that were run for climate research,
21:38but they also output ground level concentrations of ozone,
21:42and PM2.5 and one of the neat things is they simulated
21:46today, which in this study was year 2000.
21:50And they also simulated the year 1850 as being
21:53before the Industrial Revolution.
21:54So we took the difference between air pollution in 1850
21:59and 2000 and called that human caused air pollution.
22:03And then assess what that meant for global human health.
22:06So these are a bunch of different models
22:08that all ran the same experiment.
22:11This for ozone, you see, there's a spread
22:13of different results, using the different models.
22:17When we looked at this, this is the average of those.
22:19But the error bars here reflect both the uncertainty
22:22and the concentration response function,
22:24and the spread that we get from the different models.
22:27And it turns out that the uncertainty that comes
22:30from the spread of the different models,
22:32outweighs the uncertainty contributes more
22:35to this overall uncertainty, then does the uncertainty
22:38and the concentration response function.
22:40So that was kind of interesting as well.
22:42But globally, half a million or so, deaths related to ozone,
22:48related to PM2.5, about 2 million deaths.
22:51In a minute, I'll put those numbers into more context
22:54for you, you know, how do we think about that
22:56and how do we compare what that number means?
22:59I'll just finish talking about this study.
23:01This is the average of the many different models we use.
23:05This is for ozone, with most of the world's deaths occurring
23:09in India and East Asia, obviously huge populations exposed
23:13to highly polluted air.
23:16Here, we've looked at it deaths per million people
23:18in these different regions, it's certainly higher there.
23:20But even North America stands out is pretty high
23:23as well there, even though air pollution is has gotten
23:28less severe through time, okay.
23:31And in East Asia, I mean, (mumbles) PM2.5
23:36half the global total is in East Asia or so, okay.
23:40So that's an example of the type of work that we can do,
23:45addressing this question.
23:47Will come back to that question when we look at the
23:49Global Burden of Disease Assessment.
23:51This was our
23:54contribution to the Global Burden of Disease Assessments,
23:58where my lab is now looking at the statistical methods
24:01for how we can best model global
24:04surface ozones concentration.
24:06So we wanna understand all around the world
24:08what people are breathing at ground level.
24:11The challenges that we've got a lot of measurements
24:14of ozone air pollution in the United States and Europe
24:17and much less elsewhere.
24:19And I'll show you later we have huge voids where
24:24of Africa for example, where there's very few observations.
24:27So going beyond where we started,
24:30which was let's just use a model to estimate
24:32what people are breathing.
24:34Now we're going to fuse together in a statistical way
24:39the global surface ozone concentrations,
24:42I'm sorry, the global ozone observations
24:45and an ensemble of global models, okay.
24:47So we have a big team working on this
24:50we're working with Owen Cooper and Kai-Lan Chang.
24:53Owen is the chair of what's known as the tropospheric goes
24:56on Assessment Report.
24:57They've compiled together, this is the biggest compilation
25:00of ozone related measurements
25:03that it's ever been put together from all around the world
25:06going back several decades, actually.
25:08So that was a huge undertaking, including, you know,
25:10calling up the government of Iran,
25:13and asking them that they would share their ozone data.
25:16There's a lot of work that went into that.
25:18I'm using a bunch of models that come out of what's
25:21known as the chemistry climate model initiative.
25:24And then we have a big team of people in all especially
25:27mentioned, Marc Serre, who's a space time statistician
25:31who works in my department, and will use his methods here.
25:33I'll explain that in a minute, okay.
25:35So Kai-Lan led our first study which was published
25:38this year, where we're combining,
25:42again, the observations and output for many models,
25:46and we're using here this health related metric,
25:48we're doing an average of several years.
25:51And the health related metric was requested
25:53by the Global Burden of Disease Assessment,
25:56because this is how they'll assess human health.
25:59Okay, so the big the picture we take tour observations
26:03this is what those look like.
26:04Again, a lot of observations in a few places,
26:07but other places very sparse observations.
26:11We have the, this is the multi model average,
26:14the average of all the models that we're using,
26:16you see that this is biased high,
26:18so we wanna correct that bias.
26:20Then combine these together, I'll talk about the steps
26:23that we go through to do this, to create this output map
26:27that was delivered for the
26:28Global Burden of Disease 2017 Assessment.
26:32So I'll go through the steps that Kai-Lan did in this study.
26:35First, he did a spatial interpolation
26:37of all the measurements which is shown here.
26:41He looked at all of the models, these are the models listed.
26:45And he did a full evaluation of each model with respect
26:51to all of the observations.
26:54Here is really the key to what Kai-Lan did,
26:56he found in each region of the world, so for North America,
27:01Europe, East Asia, et cetera.
27:03The combination of models that best represents
27:08the measurements, the best reproduces the measurements.
27:11So he is like an optimization routine that he goes through
27:15to find the linear combination of models
27:17that best reproduces the measurements.
27:20And he's correcting bias while he does that,
27:22that gives us this multimodal blend.
27:25And the last step is where we have observations,
27:28then, we're gonna correct within two degrees
27:31of those observations.
27:32The two degrees is fairly arbitrary,
27:35and I'll talk about that choice next.
27:38But we correct for the observations within two degrees
27:42of the observation.
27:44And this is our final product.
27:46So in the U.S where we had a lot of observation stations,
27:50it's going to because of this last step, basically
27:53be based mainly on the observations
27:57in a place like Africa where we have very few observations
28:00our output is going to be based mainly on the models.
28:04Okay, so that was our first attempt at it,
28:07which was produced for the Global Burden of Disease 2017.
28:11And we just finished our work for the new
28:14forthcoming Global Burden of Disease 2019.
28:17Here we did quite a few steps to improve upon that.
28:20We're now producing ozone maps for all years,
28:22from 1990 to 2017.
28:26Where you perform a new data fusion method
28:29that I'll explain in a minute, which is Marc Serre's method
28:31known as Bayesian Maximum Entropy.
28:34We add new observations from China and elsewhere.
28:36China really started measuring in 2015 or so.
28:40Now there's hundreds of stations in China operating
28:43which were not up operating before.
28:46And when we do this, we have really the observations
28:52if there's a lot of observations
28:53that can give us spatial information on a fine scale,
28:56such as around an urban area,
28:58but again, many places in the world
29:00have very few observations.
29:02So what we did is the last step was to use this
29:05NASA model that simulated the whole world at one eighth
29:10of a degree resolution.
29:12To add that find space or spatial structure
29:15(mumbles) output product is for the whole world,
29:17each year over this period, at .1 degree resolution.
29:21So we've delivered that to GVD, they're gonna use that.
29:25I'll explain the Bayesian Maximum Entropy method.
29:28So we use the output of the multi model blending
29:32that Kai-Lan Chang did.
29:34So we're now doing that and each year,
29:36that becomes in this framework, a global offset,
29:39which is shown in blue, the BME method would
29:43in suppose these are observations around it.
29:46So the BME method would exactly match an observation
29:50at the location of the observation
29:53and the influence of this observation.
29:56If you're very far away from the observations,
29:58you're gonna use the global offset which is this
30:01model output, so basing it on the models,
30:04and the influence of these observations falls off
30:07with distance from the observations.
30:11And that function by which it decreases with distance
30:16is a function of the spatial correlation
30:19of the observations themselves, okay.
30:22I'll talk more about that in a minute.
30:24So, the features of the output is
30:26that we're gonna match observations,
30:28where we have observations and far from the observations,
30:31we're gonna tend toward what the models are telling us
30:34after we bias correct them.
30:37I should say, though, this shows it in space,
30:39but we also do this in time actually.
30:41So we use information in different years.
30:44It's often the case that a monitoring station
30:47will come online in a particular year.
30:49We can use information from those monitoring values
30:54and use that to inform the years before that, okay,
30:58in a statistical sense, where
31:01again, the further we get away from it,
31:03the influence of those observations falls off
31:06with distance or time.
31:08This is what those correlations look like
31:10this is a covariance function with distance
31:14from the station.
31:15So spatially, it drops off quite a lot such that
31:19by the time we're one degree away from an observation,
31:24we've lost a lot of useful information.
31:27But in time, it drops off actually very slowly.
31:29So, one, this is to say that one year's observations
31:34is useful for informing the years around it,
31:36and (mumbles) we make use of that here, okay.
31:40So our final product, I'm just showing you results
31:42for a single year, but we've done this
31:44for all years over this period.
31:46We started with the observations.
31:48This is a multi model average which is bias high.
31:52If we go through this step of
31:55our first sort of methods of combining together
31:58the different models in an optimum sense,
32:01this is an correcting for bias.
32:03This is the result that we get.
32:05And then our final product, which doesn't look all
32:08that different from that one.
32:09But if you look at details around in urban area,
32:13for example, especially where we have measurements now,
32:16this is doing a lot better at reproducing
32:18those measurements, okay.
32:23So the most recent global,
32:25we've delivered that to Global Burden of Disease
32:27that's gonna come out in the forthcoming assessment,
32:29but these are the results that I'll take a bit
32:31and talk about to put air pollution related
32:34deaths into perspective.
32:35This is from the 2017 assessment that was done.
32:38So, Ambient PM2.5 pollution, that's 2.9 million deaths,
32:44Ambient noise, pollution, about half a million deaths.
32:47The third one here is household air pollution
32:49from solid fuels.
32:50That's people burning coal, and straw and wood,
32:55within a home environment,
32:57often where there's very poor ventilation.
32:59So this is not in the United States,
33:02but in the poorest regions of the world where people don't
33:04have access to electricity and things like that.
33:08So that's 1.6 or so million.
33:10If you were to add up PM2.5 and ozone,
33:13that's one of every 19 deaths globally.
33:16And what the Global Burden of Disease Assessment does
33:19is they assessed a whole bunch of different risk factors
33:23such that you can compare them against one another
33:25and here's Ambien PM2.5 pollution,
33:29coming in at number 10th in this list,
33:31if you only looked at death, it would be the number eighth
33:35most important risk factor but look at the things around it.
33:38So I think if you were to ask people, what's most important
33:41for health first, you know, PM2.5 pollution
33:45is the number one, it's shown in green,
33:47environmental risk factor.
33:49Here's unsafe water coming in after that,
33:52but around this are a lot of things
33:55that have to do with diet.
33:58Have to do with you know, obesity,
34:00high blood pressure, right?
34:02And here's PM2.5 pollution being comparable to all those
34:07many of those other sources.
34:08So that's been really very influential in
34:12changing people's minds about how important air pollution
34:15is globally as a driver of global public health.
34:20I'll mention as well that in the past year,
34:22there's been another study come out where
34:25they looked again at the epidemiological functions
34:27that they're using, constructed a new function,
34:31which gives us much greater number of deaths.
34:35So 8.9 is quite a bit bigger than 2.8.
34:38If they're gonna use this function in the forthcoming
34:42Global Burden of Disease Assessment,
34:44we should expect much bigger numbers to come out of that.
34:48Okay.
34:50How will climate change affect global air pollution
34:52and air pollution related deaths, okay?
34:55This is a figure from Arlene Fury that I collaborate with.
34:59There's all kinds of ways that climate change as it occurs
35:03is expected to affect air quality.
35:06So climate change affects meteorology.
35:08Meteorology, rainfall removes pollutants
35:11from the atmosphere, higher temperatures and more sunlight
35:16make chemical reactions happen more quickly
35:19that increases air pollution.
35:22If there's stronger winds that can ventilate
35:25polluted region taking pollution elsewhere,
35:28that might decrease air pollution.
35:30There might be influences we expect of climate change
35:33to increase the amount of organics the trees put out.
35:39The if we look at wind blowing dust,
35:42if we look at forest fires, all of these things will
35:45be affected by climate change, okay.
35:48So there's a lot of different pathways here, physical
35:51ways that climate change could affect air pollution.
35:54And we're again looking at a bunch of different global
35:57models that have addressed this so what.
36:00The experiment that they ran was to hold a mission constant
36:04at present day levels and then look at 2030 experiments
36:08with future climate change versus today's climate.
36:12And then 2100 with future climate versus today's climate,
36:16so there's singling out the effects of climate change.
36:20When we look over these different models,
36:22we get different answers from each model,
36:24including some models here, a few models for which,
36:28you know, depends a lot on how the spatial distribution
36:31of where air pollution is increasing because
36:33of climate change, and where it's decreasing
36:36overlays on population, right?
36:37So if we happen to have big increase that happens right over
36:41India, which is densely populated,
36:44that's gonna be really important.
36:45Okay, so our multi model average year is positive,
36:49not hugely positive, and there's big uncertainty
36:52that comes about from the spread of the different models,
36:55nonetheless, most of the models suggest increase
36:59of air pollution due to climate change,
37:02a few suggested decrease.
37:05And for PM2.5 we have fewer models that reported changes
37:09in PM2.5, but only one of them showed a small decrease.
37:13And now we have more of the models showing a big increase
37:18the magnitude here by 2100, 200 or so thousand
37:22deaths per year attributable to climate change
37:26by this mechanism.
37:28If we look at all of the ways that climate change
37:30could affect health, this actually is pretty important.
37:34Okay, you might not have guessed that climate changes
37:37effect on air pollution would be one of the important ways
37:41that it would affect health.
37:42You might think as well, right, heat stress,
37:46spread of infectious diseases,
37:48access to food and water population displaces.
37:50There's all kinds of ways that affects health
37:52but when we've tried to put numbers to it, this number,
37:56you know, of deaths, puts it in the same ballpark,
37:59as many of those other factors.
38:01So, maybe not what you would have guessed at first,
38:03but again, because air pollution kills a lot of people
38:06that becomes important here, okay.
38:11What are trends in air pollution related deaths in the U.S?
38:15I'm gonna, we've only got two topics left.
38:17I'm gonna try to wrap this up somewhat quickly.
38:20This is the work of Omar Nawaz and Yuqiang Zhang.
38:23Omar was a master student with me.
38:25Yuqiang was a PhD student and postdoc.
38:28Omar created this nice animation for you.
38:32This is from a satellite data set looking down
38:35in North America of PM2.5 concentration.
38:38This goes from 1998 I think it was to 2012
38:44I think that's right 2011.
38:47And we've taken steps in the United States
38:50to dramatically decrease air pollution and that's sort
38:53of actually a public health success story
38:55that hasn't been talked about quite as much as it could be.
38:58We still have a severe air pollution problem (mumbles)
39:01I'll talk about in a minute.
39:02But nonetheless, we've taken you know,
39:04it's mainly EPA regulations that have driven
39:10air pollution levels down.
39:11And the effects of that are pretty dramatic
39:13when we look at it in terms of concentrations.
39:16So we wanna look at that in terms of health as well.
39:20We're using three different data sets,
39:22they give us concentration.
39:23So one is this 21 year simulation using
39:27the CMAQ regional air quality model
39:30that was conducted at the EPA.
39:32So that's pretty unique resource we're using here.
39:35That's sort of extended here using another data set
39:38the North American Chemical Reanalysis.
39:41This is like air pollution forecast models
39:45that archive their results, and we're using them here.
39:49And then the satellite derived product that comes from
39:52Randall Martin's group, he's at the Dalhausser University
39:56in Canada, what we're using as well
39:58as we're using from the CDC county level population
40:02and baseline cause specific mortality rates
40:06to assess air pollution mortality, and each year.
40:08So we're gonna do air pollution related deaths in each year,
40:11over this whole period, using this information
40:14to also account for how population
40:16and other causes of death are changing.
40:21So the results that we get using our three different
40:23data sets all should have a pretty dramatic decrease
40:26this for PM2.5.
40:29The three different data sets over in the years they overlap
40:32disagree by quite a lot, unfortunately.
40:35And that's of course, because they're reporting
40:37different concentrations, but they all show
40:39a similar trend, okay.
40:42And that's it's itself sort of an interesting finding.
40:46Because we use this county level mortality rate,
40:49we were able to then separate out the total change in death,
40:53which is in black here with uncertainty around it,
40:56and then the deaths that would have come about
40:58from only the concentration change.
41:01If we held the population and the baseline death rate
41:05at 1990 levels, and then what the effect of population
41:09and base, of course, population is growing over this period,
41:13but fewer people are dying from heart attack and stroke,
41:16which are the things that air pollution affects.
41:19So that goes down over time.
41:20But the bigger influence is really
41:22this concentration change.
41:24So we can use this simulation to estimate
41:27that PM2.5 reductions since 1990 or so,
41:31have these decreased death in 2010,
41:34by about, this is using only the EPA data set,
41:38by about 35,000 deaths or so.
41:41Okay, we did it for ozone too, only the satellite data set
41:46doesn't apply to ozone.
41:47So we have air pollution, ozone related deaths getting worse
41:53than perhaps better, but quite a lot of year to year
41:56variability here as well, okay.
41:59And again, in this case, the baseline
42:03death rate is going up.
42:05So, without concentrations decreasing,
42:10air pollution related deaths would have gone up,
42:12but in fact, they have stayed about the same
42:15or have gone down a little bit, okay.
42:20This is my public service announcement
42:23since I have your attention.
42:25I've worked on different ways of talking about air pollution
42:28related deaths and how it's important.
42:29I use the number one in 19 deaths globally
42:35from the Global Burden of Disease Assessment.
42:37For the United States, it's about 110,000 deaths
42:42from our work about 47,000 deaths.
42:45This helps translating it to one in 25 deaths
42:48or for the United States, one in 60 or so deaths.
42:52But what I think helps more as compared against other
42:54causes of death.
42:56So in, when I talk with the public about air pollution
43:00related deaths, I try to go out of my way to say,
43:04you know, air pollution is more than all transportation
43:08accidents and all gun shootings combined.
43:11Or it's a breast cancer plus prostate cancer, okay.
43:16I think for a lot of people that gets their attention
43:17and puts it in a different light.
43:19Why is it so important?
43:21Because at the top of this list, this is just the causes
43:24of death from the CDC is heart attack and stroke,
43:28being, you know, a very large number of hundreds
43:32of thousands of deaths every year.
43:34And air pollution modifies that, air pollution affects
43:37those deaths, which means that at the end of the day,
43:40air pollution is really important here, okay.
43:44Let me skip over that.
43:45Okay, so, last question.
43:48If we slow down climate change,
43:49what are the benefits for global air pollution and health?
43:53This is known in the literature is that as CO-benefits
43:56so let's say the world listen to
44:00the teenagers marching on the United Nations this week,
44:03right, got their act together and reduced greenhouse gas
44:09emissions to solve climate change.
44:11Many of the actions that would be taken would be
44:13to shift us away from fossil fuels.
44:15We know that fossil fuel combustion is the major source
44:17of air pollution that we care about
44:20that influences our health.
44:22So there ought to be called benefits associated with that.
44:25And there ought to be health benefits.
44:27Actually, Michelle has worked in this area too.
44:29If we look back historically at these studies,
44:31a lot of those studies were done by public health people
44:34that maybe didn't take is a very sophisticated look
44:37at the atmospheric science part of the problem,
44:40or by economist, right?
44:41That we're motivated to understand,
44:44how big is this code benefit compared to the costs
44:47of reducing air pollution in the first place?
44:51When we take action to reduce emissions of greenhouse gases,
44:56that reduces greenhouse gases
44:57but also slows down air pollutant emissions,
45:00that's good for air pollution and human health.
45:02This is a pathway that is immediate local,
45:05but I also told you that climate change as it occurs,
45:09so in this context, we're slowing down climate change,
45:12climate change effects, air pollution.
45:14So we're slowing down that influence too.
45:17So our study was the first to look at
45:19these two different pathways,
45:21such a you could add them up together, okay.
45:24I'll show you some results of that study.
45:26So again, we're using our global atmospheric model.
45:29In this case, I've worked with a team of energy economics
45:33modelers using the what's known as the G-Cam,
45:36energy global energy economics model.
45:38So in doing this, they simulate what the future
45:42would be like under, you could say a reference case
45:45or a business as usual case without climate policy.
45:48In their model, then they apply to a global carbon tax.
45:53That was pretty aggressive,
45:54aggressive enough to really actually
45:57have a big effect of slowing down climate change.
46:00Within their model, the model is choosing the
46:03most cost effective ways of reducing greenhouse gas
46:05emissions, we were then able to see
46:08what is each of those actions have
46:11mean for air pollutant emissions,
46:15and then put that into our global atmospheric model
46:18overlay that on the global population.
46:21So these are global changes in global PM related deaths
46:25the solid lines in the reference case,
46:28and in the emission reduction case.
46:30So it's the difference between the blue and the red
46:33that is the CO-benefit.
46:35That is attributable, in this case of the climate policy.
46:39We're getting numbers that are half a million deaths
46:41or so by 2030.
46:42So immediately, we get a pretty big benefit by 2100.
46:46We're at one and a half million deaths avoided
46:49by this climate policy.
46:51For ozone, we also get by 2100, pretty big number.
46:54This is in part because the climate policy
46:57is reducing methane and I told you
46:59that methane is important for reacting to contribute
47:03to the globalism background, okay.
47:06When I put numbers, dollar signs associated with this
47:10I'm using here, red is using a high value of a life,
47:13blue is using a low value of a life
47:16looking at it in 2030, 2050, 2100,
47:19the different world regions and the global average here.
47:22So you get, you know, regions like
47:25that are densely populated,
47:27that have severe air pollution problems now,
47:30having pretty big monetize benefits that come out of this.
47:37Some regions here like Australia with a very low population,
47:41and it's gonna be the CO-benefits are gonna be much smaller.
47:45The green shows using 13 actually different
47:51global energy economics models
47:53that all ran a similar experiment,
47:55the cost of reducing emissions per time.
47:58So this is all normalized per ton of carbon dioxide.
48:01So, cost per ton, the solid line is the median
48:05of the 13 models and the dashed lines
48:09give you the full range of those models, okay.
48:11So that's shown here, the benefits outweigh
48:14the cost in 2030.
48:16Also for most world regions in the global average in 2050,
48:20by 2100, we've taken advantage of all the
48:24very cheap ways that we know about reducing
48:27greenhouse gas emissions and are moving up the cost curve.
48:30And there's quite a range of estimated costs
48:33here from this point to this point,
48:35nonetheless, the CO-benefits are still pretty
48:38comfortable with that.
48:39So, we found here then that the CO-benefits are comparable
48:43to or exceed the cost of reducing emissions
48:46in the first place apart obviously,
48:49from other benefits of slowing down climate change itself.
48:52And all the reasons that you go on to that.
48:56When we looked at the CO-benefits literature,
48:58so the the entire range of CO-benefits literature
49:01is here in yellow.
49:03Dollars per time, these are studies that were done
49:06in all kinds of using different methods over
49:09a couple of decades, in all many different world regions,
49:12but most of these studies were local, or for one country.
49:17And one of the novelties of our work,
49:20we put it into this global framework,
49:22we're now accounting for if the United States,
49:25for example, reduces emissions, that affects health
49:28in Europe, actually in Asia, because part
49:31of that air pollution reduction affects
49:34air quality elsewhere and benefits human health elsewhere,
49:37by putting this in a global framework,
49:39where accounting for all of those trans boundary
49:42and influences, okay.
49:44so that's our global CO-benefits study.
49:49Yuqiang Zhang is my PhD student then did quite a lot of work
49:52to downscale that to the United States,
49:54and I'll show you a couple of the results from that.
49:57When he did that for the United States.
49:59Again, we're similarly (mumbles) a global climate policy,
50:03but he ran a couple of experiments to separate out
50:05the effect of domestic within the United States
50:08emission reductions, right here
50:11versus what comes from foreign emission reduction.
50:14So when we look at PM2.5,
50:17most of the benefit is from domestic reductions
50:20that makes sense PM2.5 has a rather
50:22short lifetime in the atmosphere it doesn't move very far
50:26from it's source.
50:27So, most of the benefit is domestic
50:29with some influence for example, from
50:33the reductions in Mexico and Canada
50:35that effect in the United States.
50:38When we looked at the...
50:41when we looked at ozone, however, most of the emission
50:45most of the benefit actually came from
50:47actions that foreign countries took
50:50and the global reduction in methane.
50:52Okay so that was an interesting.
50:54(mumbles) Yuqiang, then looked at the
50:56health benefits associated, finding that most of the benefit
51:01for reduced PM2.5 came about
51:05from domestic reductions shown here.
51:08And most of the benefit for ozone related deaths came about
51:12from foreign reductions
51:14affecting health in the United States, great.
51:20I've covered a lot of ground today.
51:21I hope it wasn't too much for you.
51:23But I hope each of you maybe took away some nugget
51:26that you will carry with you.
51:27There was a lot of people that contributed a lot
51:30of work to this.
51:31Several graduate students over many years,
51:33I really highlighted the work of Yuqiang Zhang
51:36and Raquel Silva, over my PhD students
51:39and did a fine job doing this,
51:41and a lot of collaborators over these many studies.
51:44So thanks a lot for listening
51:45and I'm happy to take some questions.
51:48(students applauding)
51:58Yes, right here.
52:01- [Female Student] (background noise drowns out speaker)
52:05I have a question about the definition
52:07of ozone layer mortality or PM2.5, related to mortality.
52:14I mean, how do you define (faintly speaking)?
52:17- Right, so what we're doing here
52:19is we're using results of an epidemiological study
52:23that would have related PM2.5 and ozone to mortality.
52:29And then using our model, we come up with different
52:33estimates of concentration depending on the application.
52:37And then we apply that function.
52:39So it's the function, the epidemiological function
52:41and the epidemiological study.
52:44I should have made this clear up front more
52:46that relates PM2.5 and goes on with health.
52:51The studies that we're using are the big cohort studies
52:54that are from the United States, largely okay.
52:56So the American Cancer Society Study.
53:00So it's a bit of a leap of faith to say that
53:02that function applies elsewhere in the world.
53:05And we're also in some of our applications,
53:08assuming that, that function applies throughout
53:10the whole century to come, right?
53:13We don't know that that's true.
53:17And we don't know that they apply elsewhere.
53:19Now we're getting better information about
53:23air pollution related deaths in China and India
53:25and elsewhere, but still not the same quality
53:30and number of participants in the study
53:33as we have for the big cohort studies in the United States.
53:35In other words, I'm not sure what else you would assume
53:39about what happens elsewhere in the world
53:41or from the future.
53:42But we should acknowledge and I didn't say it,
53:44but I'll say it now that there's big uncertainties
53:47and assuming that those functions apply spatially
53:51and through time like that, and hopefully that helps
53:53with your question, yeah.
53:55- [Male Student] So particulate matter can be very diverse
53:59it's just size of a matter that you contain chromium six or
54:03(background noise drowns out other sounds)
54:06so how do you take that difference in the heterogeneity
54:10of this substance across different countries?
54:12Or is there a plan?
54:14Because you don't have the data, right?
54:15You have (faintly speaking).
54:16- Well, we don't have the epidemiological studies
54:19that tease out those relationships.
54:21I know Michelle is working in that area,
54:23and other people are as well.
54:25If we had that we if you know, give me a function,
54:27and I'll use it.
54:29But you know, short of that, it's a real question.
54:33And from an air pollution management point of view,
54:37you know, if we knew that it was the sulfates
54:40or it was the organic carbon,
54:42we could just regulate that rather than the mass.
54:44So the limiting factor is really actually
54:47where I started off the presentation talking.
54:51It were limited by measurements of air pollution
54:55that then could be used for epidemiology,
54:57that then could divert derive a function
54:59that then we could use for this kind of application,
55:02but, you know, we're learning more about
55:06using those different measurements and now becoming
55:08more creative combining satellites, you could use a model,
55:11for example, to estimate the contributions
55:14of different emission sources
55:17or different chemical components
55:19to an air pollution mixture, and then do epidemiology
55:22based on the model, right?
55:24Okay, so we're coming up with a lot of new and creative ways
55:28of approaching that question, but yet great question.
55:33Yes, please.
55:36- [Male Voice] I have a question about the,
55:38about your model versus the
55:40Global Burden of Disease model currently.
55:42So the estimates that you had for air pollution
55:45related deaths with something like 40,000
55:50versus, no...
55:52A 100,000 or so, versus by 40,000 with the
55:56Global Burden of Disease,
55:57What is the key differences in your model versus that?
56:00- Yeah, so one is the function that's used
56:03for to relate air pollution with health.
56:05The other is where we're getting exposed
56:07like concentrations from, is it from a model
56:10or from some model measurement blending.
56:15The factor of two is more than a greater difference.
56:18And we would ideally like to see (mumbles)
56:21I mean, we're really working to try to continue
56:23to tease out those differences
56:25and see if we can resolve them.
56:27I know the satellite people have now produced a new
56:31for PM2.5 the satellites have been really very important.
56:35Satellites can see ground level PM2.5,
56:38but they can't see ground with a low ozone.
56:40That's one of the important distinctions here,
56:41we didn't have the benefit of satellite
56:44providing information on ozone.
56:46And they can see it with the satellites
56:48can see it with very fine spatial resolution.
56:51So in the PM2.5 world, you know that
56:54it's actually the satellite that provides a fine spatial
56:58resolution whereas we used to fine resolution model
57:01to do that anyways, that goes beyond your question.
57:04But your question is a good one.
57:06And it troubles me that it's quite as different as it is.
57:10But, you know, I think we need to just continue
57:12to work on it.
57:13See if we can work out the differences
57:16between the different studies.
57:17- [Male Voice] So for the 2019 (faintly speaking)
57:21this model is gonna be adopted,
57:23because it is a very large change.
57:24And this is something I've noticed with other
57:27updates of the GDP numbers for the same years
57:31get updated dramatically as a result.
57:33- Yeah.
57:34- [Male Voice] And so depending on when you actually
57:36access the data, you might get pretty large
57:39in the estimates, so do you have a sense for
57:43what's gonna be done in the 2019 study?
57:46- Well, I know that I know what they're
57:48doing for concentration.
57:49So they're using a similar method for concentrations
57:51and then our ozone estimates, I mean,
57:54for PM2.5 concentrations and then our ozone estimates
57:57are gonna be used that I know well.
57:59I don't know what risk functions they're planning to use.
58:03And it's a good question.
58:04But that's, you know, they have a team of people,
58:07you know, some of the best epidemiologists in the world
58:09reviewing the literature.
58:11So I leave that up to them to use.
58:13I try to, you know, not push the envelope there
58:17our studies are pushing the envelope just by bringing
58:20different information from different fields together.
58:23That's why I (mumbles) so we gained nothing
58:25by using some epidemiological study that, you know,
58:29the people who really understand it,
58:32let them choose it, right?
58:35- [Robert] Okay, so I think we'll wrap
58:36it up, two announcements.
58:37So there's lunch in the LAPH 108.
58:40And also for students who are interested in available,
58:43Jason's gonna be having an informal discussion
58:46starting around 11:15 in room 101.
58:51So thanks again Jason. - Thank you.
58:52(students applauding)
58:58(students chattering)
59:22- Hi, I'm (speaking off mic)
59:23- Oh, hi.
59:24Thanks a lot
59:25- (faintly speaking) or maybe already know, right?
59:34- Oh, that's true.
59:36- Yeah, have you ever heard about that?
59:38- I saw (mumbles), I saw (mumbles), the day (mumbles).
59:44- There you go. (laughs)
59:45but we still need to hear (faintly speaking) (laughs).
59:48- So I wanted to ask, like do you your models initially,
59:53what sparked the question was when I saw (mumbles)
59:56one of the earlier ones your (mumbles) over.