37C3 – Numerical Air Quality Modeling Systems

a journey from emissions to exposure

High performance computing (HPC) in environmental science is usually associated with research on climate change, investigating the impact of atmospheric greenhouse gases (GHG) over the next century. Besides these GHGs, there are many other gases and aerosolos in the atmosphere, which have a much more direct and immediate impact on human health: air pollutants.

The World Health Organization (WHO) considers air pollution to be the world’s single largest environmental health threat, accounting for approximately 7 million deaths worldwide every year. That’s why in this talk we want to speak about how the problem of air pollution can be understood and predicted using HPC pollution modeling and its application based on general concepts and our own research.

We are Dr. Johannes Bieser and Dr. Martin Ramacher, both working at the Helmholtz Zentrum Hereon in the field of numerical pollution modelling. While Dr. Bieser wrote his Dissertation on emission modelling and its application, Dr. Ramacher wrote his Dissertation on pollutant transport and exposure modelling.

In our talk on numerical air quality modelling systems, we want to introduce basic principles and share our personal knowledge in the field of numerical pollution modelling, covering the entire pathway from emissions, transport, transformation and human exposure. Each of these steps relies heavily on large amounts of data from many different sources – satellite data, activity and meta data, measurements and many more – and skills in computer science. By default, environmental scientists are often not trained in computer science and high performance computing which implies a challenge of its own (and allows Nerds like us to excel).

Our talk will be enriched with practical, technical and partially political examples to demonstrate the difficulties scientist face during their quest to improve air quality for everyone: from TB of wasted data due to historically grown data formats to counterproductive policy decisions to „improve“ air quality. We’ve seen it all and after participating in the CCC for many years now, we decided to draw attention to some state-of-the science approaches for solving one of the world’s single largest environmental health threats: „air pollution“.

ottopaul
Johannes Bieser

https://events.ccc.de/congress/2023/hub/event/numerical_air_quality_modeling_systems/

#37c3 #Science

Okay our next talk is numerical air quality modeling systems uh our two speakers will explain to us how scientific methods can be used to model the quality of of um pollution or air quality in general uh our speakers are Martin Raa uh he is with the Hol Centum for hon and his main

Research subject is the um exposure of pollution concentrations and um on his um on his left there is johanes beer main research interest is the meta transport of in the ecosystem um and he’s with the hams Institute as well please welcome them both with a huge round of [Applause]

Applause check check ah okay hi everyone we are johanes Biza and auto Po and today we will take you on a journey um on numerical air quality modeling systems from emission to exposure actually we realized that it’s more like a roller coaster ride because there are

So many topics that we have to cover that um we need to be really quick and will be bumpy but the idea is that after our presentation you have a general understanding of what we are doing why this is important uh what are state-of-the-art Concepts and applications and what it can bring for

The future so let’s dive into it and we will cover all of these topics we will talk about air pollution where it comes from why is it important measurements then of course the numer iCal air quality modeling systems applications exposure we have a little finale and

Then you can ask us anything if there’s still time left so we dive right into the first topic which is air pollution and we will start with a little lesson in history so this is a picture of the great smok of London in 1952 and as you can see you can’t see

it’s very foggy so uh this is basically the smoke that happened and it happened because um it was Winter it was cold there was not much wind and on top of it literally there was a warm layer of air so if you have a warm layer of

Air on top of a cold layer of air you have something we call inversion in Metrology and that means basically that all the emissions are trapped below this warm layer of air so we have all these pollutants trapped inside and what happened here was that there was lot

Lots of coal burning so people burn coal to eat their houses and for industry and this resulted in Smoke um as a result of this mock many people died so um there are some estimates from 20 2009 from the UK Met Office that correlated the the smoke

That happened to uh the death that occurred during these days which you can see here on the left and the actual numbers are estimates of 10 to 12,000 people that died due to smoke and many more got sick but the big question here of course is what is smoke because smoke

Can be a lot that looks like smoke or fog um and in this particular case we are talking about winter smoke which I said it already due to the burning of coal there were a lot of emissions and in particular these were emissions of sulfur dioxide so you see here so SO2

Sulfur dioxide keep that in mind because we will have all these little um pictures of the the pollutants we will have them reappearing in the presentation often so there was sufur dioxide and there was parate meta but that was not everything because both of these are in high concentrations already

Toxic but what happened here was that also the air was saturated with water so we had a lot of high humidity and a lot of sulfur dioxide and in combination this leads to the formation of sulfuric acid and as the term acid says already it’s acidic so it’s you shouldn’t

Breathe it it’s just it’s dangerous for your health um another thing that happened during the 1950s were situations like this uh in Los Angeles so you had this um also smoke again but that’s a different kind of smoke because this happens when there’s a lot of sunlight when you have cars and Industry

And also trees all of these they emitting lots of substances and in combination with sun again you have the formation of smoke and if you breathe it it’s dangerous for your health actually it’s so dangerous that you are gathering with your friends and basement wearing gas mask and asking yourself what is

This smoke and as I said it already in this case it’s summer smoke so sosm is basically a product of nitrogen dioxide which is mainly happening due to combustion in cars but also in industry and other combustion sources uh plus volatile organic compounds which is actually a huge range of compounds which

Um Biza will introduce to to you later and under the influence of sunlight you have the formation of ozone and ozone in high concentration is toxic and can endanger your health so as a result of these events and many other events around the globe there have been some

Clean air acts so-called and the first one was in 1956 that was a direct consequence of the smoke in London and it was a milestone in legislator air quality regulation or regulation of such things in general um and some of these measures like regulation of fuels and smoke free zones are still measur

Measures that we are applying today to regulate air pollution um in 1970 there was the uh Us Clean Air Act and at the same time there was also the establishment of United States Environmental Protection Agency that’s not a coincidence and today uh the the EPA

Does a lot of other things and not just regulating air quality so very two important acts that happened in history with respect to air pollution so when I talk about regulating air pollution we are we can come from two different perspectives here uh we have for example the the

Guideline values which are international guideline values these are concentration limits or values that you should not exceed because they endanger your health and this is based on many studies that were gathered by the wh it was last updated a couple of years ago and is actually um yeah it’s it’s a guideline

Value so we shouldn’t exceed these values then there are limit values for example the limit values which are established in the European Union by European commission they Orient they are orienting them um along with the wh values but as you can see they were last updated in 2008

Currently they are under revision but the existing limit values we have right now are way higher compared to the ones we have from the wh nevertheless these limit and guideline values are success story because if you look for example what’s happening in the European Union with the population in urban areas for

So SO2 you can see that less than 1% of the population living in urban areas is exposed to values of so SO2 considered to be dangerous for your health today or in 2021 but for other pollutants you can see a clear difference between uh what’s what’s going on because you have here

Particles in different sizes you have ozone and you have NO2 and if you apply the limit values by the EU they are still a problem for particles and ozone but if you apply the guideline values by who it looks like we have a problem almost everywhere and this is just this

Thing that the European commission values are not updated for a long time they are envisioned right now and they will follow probably the wh values but they will be probably not be as strict as the wh values although if this is a success story um if we look in other regions in

The world like where there’s no such um regulation for example here in Kazakhstan you can see very well we have winter we have smoke and this is what I what talked about before this is the so-called inversion so the clear sky above is warmer there’s warm air masses

Below we have cold air masses everything’s trapped inside so it’s you don’t want to be there same thing in China sometimes smoke occurs so there are many regions in the world where this is still a big problem but not only in other regions coming back to London we

Also um had something very unique again in London there was a ruling um that a 9-year-old girl from from lsam in South London she died in 2013 as a direct result of air pollution and this ruling stated that air pollution made a material contribution to her death so

There was the first ruling of that kind ever and I think this introduces the new yeah new way of dealing um with air pollution if we go back from the individual to the global perspective today air pollution is considered to be one of the world’s largest environmental health threats accounting for 7 million

Deaths around the world every year by The Who and 4.6 attributable to uh ambient air pollution so I I guess we we can settle that we have a big problem here that we should start to tackle and actually there’s a lot of work to

Do thank you um and um the thing is most of the time the idea is you can only manage what you can measure so we need measurements how are these measurements done for regulatory uh purposes so there are these regulatory measurement networks and what you can see here is um

An example of such a classic air monitoring station that is useful such purposes is an Essence and measuring the urban background it this um containers they look more or less the same everywhere probably you if you’re from Hamburg you have seen some of them they look similar it’s always a container and

There’s a lot of equipment inside which is standard equipment it’s calibrated and maintained and these equipments cost more than €100,000 so it’s very expensive to do that but they measure more than 20 pollutants continuously so it’s quite good to have these to get an idea of what’s the air quality situation

Like for example in Europe here you see the network the European air quality monitoring Network by the European environment agency and these are results for p 2.5 um the dots are measurement stations measuring in annual mean which is displayed here and on the first side it

Looks like um we have a good spatial coverage so we get an idea of what’s going on here but imagine each of these points is one station and the size with respect to the scale of the map you can get an idea okay there’s lots of space

In between so if you zoom in you see all of these Mees stations are far away sometimes they are even that far away that the data comes from Brandenburg and is instead of searching for Internet searching for for air quality uh measurements but can’t find them um so

What I want to say here is the coverage of measurements is sometimes really sparse so we need more measurements to fill all these gaps for example um I will skip this one so what can help here is for example do it yourself and locost S on networks

And I’m pretty sure or I hope that some of you in this room are dealing with this topic because I would like to get in touch with you uh there’s many different networks I just named some of them sensor Community former Lu daten info um lu D at open cologne and so on

And we did a little study for Hamburg where we wanted to evaluate um some modeling we did which I will also show later but just to get an idea of the spatial coverage in Hamburg you see here all these dots are measurement stations where these um these measurement pipes are located and they

Are giving us um yeah values into a database of pm2.5 so it’s it’s really useful to use this to get um the spatial distribution of the pollutant concentrations in a city like Hamburg better so it’s it’s really an update compared to the 16 measurement stations we have um

Officially here in Hamburg what’s kind of a problem is um the quality of the sensors but there’s lots of stuff going on to to improve the quality with some algorithms and what is also sometimes a problem is the temporal coverage so not all of these sensors are really

Recording the entire year but some of them have some some data gaps that we need to fill them but anyway it’s very useful to have these measurements um coming back to the global scale we have here a screenshot of the world air quality index project and actually they are gathering all of

The measurements available I think this is the most comprehensive database um I know which is accessible all the time online and you if you zoom in you see all the lowcost sensors you see the official monitoring sites you see everything so it’s very nice but what

You can also see on the first side is that we have a global North bias so us covered Bell China India that’s all fine and Japan and Europe and Australia there are also some measurement stations but like South America and Africa okay Siberia too but uh we have some some

Gaps here which should be filled um yeah so coming to an end for measurements or observations they are still the gold standard uh when we talk about high-end instruments the problem is just that they are very expensive um because of Technology calibration operation maintenance but they are highly reliable and

Compatible problems or things that we could improve are the limited spatial and temporal coverage so as I said there’s a global North bias especially and then for the high-end instruments uh here do it-yourself sensors can work the problem is also that we can only measure what happened and what happen so we

Can’t predict the future with measur not yet not that I know and we also have very limited source source abortion capabilities which is also um a problem is that these measurements they have detection limits so if you have very low concentrations it’s hard to measure it and also we have errors which are

Connected with measurements and these can deviate from the True Values and with this I would like to welcome you to the model welcome to the model you made it here I’m amazed after Martin told you that we have a problem and that we can measure it I mean you

Probably think what what benefit if I find out if my food is poisoned after I eat it so wouldn’t it be nice to know beforehand I mean wouldn’t it be nice to know how air pollution is tomorrow or next year or to know what some policy decision or some regulation would

Actually do once we enact it and huh lucky we can do models there’s just one problem I mean with the models is that they’re basically all wrong um but sometimes luckily they’re still useful and then that’s the point where you would start defining what a model is and that’s difficult because basically

Anything is a model um really anything can be a model so now let’s find out which model we actually want to use to answer our question at hand and our question is that we want a model that has everything from the emission to the exposure so we need

Stuff being emitted into the atmosphere we want things flying through the atmosphere and move we want them to react and change become more toxic less toxic and then at some point we want to know is the stuff actually going inside a human and okay that’s our goal so as a

Modeler of course we think in the Box um not outside the box today we’re in the box and so we imagine a box where we have these processes we have stuff coming in we have chemical reactions and we have transport and yeah maybe more boxes is nice so we start fing the

Planet with boxes and our atmosphere is basically just a huge bunch of boxes of different sizes and here just let me deviate a bit on the topic of scale because it it really depends what do we want to look at if we want to know like the global circulation how much

Pollutants come over the Atlantic from the US to Europe you can use crit cells that have a size of hundreds of kilometers that works fine um um if you want to know something about air pollution in a city you really need to go to the sub kilometer scale where

You’re 10 100 m basically you can see that on this fancy diagram that shows you on the x-axis actual scales distances on the planet and on the y- AIS you see the characteristic time scales of different physical processes and so they basically see we’re between a few meters to a few

Hundred kilometers in our crit cells so let’s just say we filled a whole planet with crit cells and now we want to start modeling and an air quality modeling system is is a modeling system because it’s basically just a bunch of different models getting thrown in a box and

Connected to each other and here we start with the transport so we don’t think transporting so that’s in our case I we call it the hydrodynamic model it could be an ocean model if you would be interested in pollution in the ocean but in our case of course it’s an

Atmospheric model and there was a really nice talk on the last chaos Congress so if you’re interested in um climate and meterological modeling there you can now if you watch Us online go hop over and come back um it’s just these models we’re using them and we’re building upon

Them and there’s just no time to speak more about them but in the end it’s just the physics it’s the temperature it’s the pressure it’s the movement the wind it’s it’s the water like the rain and the water vapor going around and um now here that that’s very are in in 2023 so

You see on the on the left side the original Apollo picture from the 70s and on the right side a a recent 1 kilom resolution Global atmosphere ocean run with the ion model that’s actually the model that’s used for the German by the German Weather Service for the forecast

So you have seen that and we’re really at a point where personally it starts being difficult deciding which one is the picture and um with that I can take some breath get the first mesmerizing video with the supplemental information into you and so basically we now have things moving through the

Atmosphere check first model is there so now we want to have stuff coming in so we need doing emission modeling and EM Mission modeling is just a nicer way of being an accountant of everything because it’s everything emits nature emits plants emits even the planet itself G geogenic volcanoes emit U of

Course I mean the humans are the big emitters with Mobility energy industry and agriculture we quite a deal but all those natural sources a forest might be nice but in a certain circumstance it can be an emission source of some pollutant and then it’s not only the pollutant U this the pollutants

Themselves but it’s also precursors you you’re not just modeling something it goes in the air transports really you need to reproduce at least some basic atmospheric chemistry which means you end up with something in the range of a few tens dozens of species of substances

To really being able to do that and here are the friends again we keep showing them so we secretly give you some chemistry lessons Without You noticing um those pesky blue things down there it’s basically hundreds of thousands of different carbon chains so we try to somehow not have hundreds of thousands

Of difference so we use some estimates and um but that that’s very high level so how do we actually do it and there we basically have have the two general approaches it’s coming from the top down or from the bottom up and basically it’s a top down approach is when you have

Some fancy statistics like imagine you know exactly how much gasoline was sold in Germany in the last year you know exactly how much cars have been sold or are registered and just from those numbers you can estimate like how much nitrogen oxide how much carbon monoxide

Did cars emit in Germany then you have this number and then you start Distributing this number into every crit cell into every time step typically emissions you look at like hourly values and then at some point from this one estimated value you come down to to

Actually a map looks a bit like that um you use some GE reference data population data socioeconomic data sets unemployment statistics there’s more pics in SCH hin than humans I found out um yes really it’s true and um and then you just use Bunches of profiles and different Tempo distributions to

Basically get your big total number into each and every of your little crit Sals um the opposite way is that you have no idea about how much is emitted but you have a pretty good idea about the sources and then you just come from each source and estimate each source and

Here’s an example for that that’s the uh shipping anti-collision AIS system so you get a global coverage like every five minutes of every major ship with the exception of those Russian black Fleet oil tankers that turn it off and then you actually can get it the data

For every oil tanker in the world how big is it how heavy is it what engine does it have when was it last rehauled what what kind of fuel does it use and you basically model every single ship and then equally you end up with an Miss Mission map I tend to forget

Breathing you need a break okay um Stay With Me um let me just give you a little example that’s just a what popped up when uh it it’s basically just an emission map for vehicle emissions in Germany for different Road types and the total emission there’s the same amount

Of stuff emitted in Germany but you you can also see that they look similar but definitely different you can see fancy artifacts and some strange things in the top down which shows you that you Interpol um and you can see much more detail in this this detailed model down there

Where basically there have been cars and destinations and people driving on actual roads and um yeah that’s maybe the first first time to complain um so you you model these cars and then of course one thing is you have an emission Factor so like look here2 G of NO2 per

Kilometer and I mean for a few years all the modelers knew that we have to multiply our uh vehicle emissions with some Factor at least two or three and then no one was really surprised when it came out that the emission factors weren’t right because actually models do

Work you wouldn’t believe it I mean they say everyone believes the measurement besides the person who did it No One Believes the model besides the model and um yeah trust me and having now this this monster of an emission model with all these sources and species we cannot only have things

Transported we actually can have stuff transported that’s from our colleagues in K the kit and you can see some Saharan dust that’s blown up and here you can see a nice episode actually the Saharan dust then is transported up to us um but it’s still not enough remember

We wanted also things to happen so we want chemistry and all our our species and pollutants to react with each other and um I mean it’s mostly nitrogen and oxygen what what people aren’t aware is like how how few of those air pollution molecules there actually are we’re

Talking about a few per million of air molecules with some toxic substances it goes even to billions and so what do we do Martin said I couldn’t do a complete atmospheric chemistry lecture no one would like it and we don’t get two hours additionally so um we compromised on two

Really interesting examples and I try to be quick and easily to follow and we start with tropospheric ozone so you think also and that’s nice we have it in the stratosphere it takes out the UV but we also the tropospheric ozone that’s down here where we live and breath and

It’s a completely different mechanism there’s basically that’s a simplified thing that tries to convey the message that you have two circles that interact so you have this n O2 circles and you have the big circle in the middle and here you can just follow like this little little VC how it’s getting into

The circle and then it interacts and now on the bottom you see oh the first time the circle went through so we have produced some ozone and the VC happily continues it hits the other cycle again and in in this way the idea behind that is that it’s a highly nonlinear process

Break one Circle the other doesn’t matter so one industry in one location could be a huge problem for oone take the same thing somewhere else it doesn’t matter because um if you don’t have one doesn’t matter if you add the other and then I stumbled uh randomly about this picture

On the internet and I thought I mean it’s nice I like trees but poas are actually one of the highest eocs so I don’t really see a nice tree I see like a deadly chemical trap no wait wait wait wait a minute wait a minute boring sight

Is coming in because actually we did some modeling studies on that and uh it really depends on where you are and when are there for example in Hamburg and in the rhin ru area the impact of plants on formating extra ozone is really low like 2% in Berlin during a heat wave there

Was an impact of up to 12% and sometimes even higher but if you look at the general findings based on literature you you can see that it’s really important what kind of tree you have on the left you have a tree that emits just a few

Voc’s and on the right a lot of voccs and then it also depends on the conditions you have so if there are already a lot of voc’s uh compared to nox um it’s it’s really important how much there is so if there are a m a lot of

These high VCS and there are a lot of these trees that emit a lot of VCS there will be oone formation so just to jump in here yeah yeah we still have chemistry to do yeah so um sorry for him interrupting um because now that that’s the that’s really interesting it’s

Secondary particle formation so isn’t it thrilling just the word and um basically what you have is that something is emitted as a gas sticks around in the atmosphere and once it meets another gas from a completely different Source they can actually build particles and that’s the example here with the ammonia on the

One hand and the sulfur and nitrogen dioxides on the other these are gases that get emitted let’s say there’s a ship emitting NO2 there’s a cow emitting NH3 and the stuff oxidizes and it basically turns into mini magnets and then it starts being transported with the atmosphere and when they actually

Meet the mini magnets will engage and you form particles and um that’s really interesting and I’ll show you a nice video in a second to show how interesting it is but um with all the talk I think I managed to get through three of our four models so we have

Transport we have like a metrological model the physics and all the transport we have an emission model that gives us all the stuff getting into atmosphere and then we have the things reacting in the atmosphere so I guess we can start with some interesting applications that we did like

Randomly we have this video um and again that shows how nice models are so here basically you run the model with everything and then you add a source or remove a source and by looking looking at the differences between your runs here we have the secondary particles

Formed from ship emissions so it’s only the emissions from ships and so in the English Channel you will have all the NO2 going into the atmosphere and when the wind gets it over the the coast line it reacts with the ammonia and then you can get these nice blobs and that’s a

Specialty of the pm2.5 because it’s so small it’s it’s not heavy enough doesn’t fall down so these blobs basically persist until they hit a rain and if it doesn’t rain it will just stay there and oh a 7 microgram blob just bent over Hamburg that’s already like half the limit value

That also shows you that on a given day a source far May might ruin your air quality nothing you can do and another thing with these particles is that especially with all those diesel Vehicles a discussion was really strongly about there’s the particles there’s too much NO2 coming from the

Vehicles and then with a a student Anna a few years ago we thought yeah but what about if we removed the ammonia instead of the NO2 and basically you just see on the left side a map for Europe with the secondary particles in our standard prediction what it should be and then we

We didn’t even make people vegetarians we some World Health Organization diet with reduced food consumption and meat consumption and yeah just having less manure and animal husbandry some areas really have the particles already that was quite interesting we have some examples Yes actually I would like to

Catch up on what you introduced um about the advantages of modeling because uh as as you said there is the chance that we can do Source a portion man and we did a little study here for hurg in 2012 and you can see the total PM 2.5

Concentration on the left and on the first side you see some hotpots there but the question is now where is it coming from so what are the responsible emission sources and on the right you see industry Residential Heating ships and traffing with all their relative contribution to that and this is

Something you can easily do with models um in this case you can see that for traffic for example we have a big problem like everywhere then when it comes to ships and Industry it focuses on specific reasons and this maybe you keep that in mind for a

While um and what we also can do with models as you introduced uh we can do future scenarios so uh I know 2020 isn’t the future anymore but at that point where we did this study that was in 2018 it was published then in 2020 so um this was still the future and

We made some prognosis with scenarios what happens in 2020 and modeled it and then if you do the difference of these you can see what’s what’s the impact where is the biggest impact what are the sources where we can really start to to regulate and you see again it’s

Traffic so then um there have been some traffic bands in Hamburg maybe some of you know maybe some of you can’t remember and I um yeah these are two streets here in Hamburg where have been banned for diesel trucks and banned for diesel cars and if you exclude some

Specific trucks or cars from these roads to red use the measured concentration then it might appear that you don’t have a problem anymore but if you would have run a model you would have seen that the air pollution would just have shifted to other regions so it’s really a question

How useful are such such meurs in the entire perspective of a city okay still alive okay good um and yeah we can clearly identify such such such things with models it doesn’t um Narrows down the effect that happened there of course the concentration went down um what’s going on

M yes okay um coming to an actual idea for scenario we can model we do often and we like to do some Maximum Impact scenarios for example here what would happen if he would introduce only electric cars um in the inner city of Hamburg and and also we fire them with

Coal so the dirtiest source of energy we know right now but still what would happen if we would do it and you could see that even if we in the year 2020 would have exchanged all the cars in the inner city by electric cars there would have been substantial reductions of NO2

Up to minus 20 microgram per cubic meter and actually that’s the uh existing limit value right now so this would have a really nice impact and the higher emissions from power plants at that point there was still the the power plant and more Borg active right now

It’s not anymore but like the impact of additional energy to power all these cars was very small so we are talking here about ranges of 0 to 2 microG per cubic meter so also very nice model application and what we did in the last year we um created an urban air quality

Forast for the city of Hamburg so up there is the URL you can all look it up on your iPhone it’s public and it’s based on all the measurements uh all the models you introduced um so it’s really based on science we are doing on a daily basis

And it’s improved all the time but it’s still very ugly so if some some of you here might help us with making it a bit nicer I would happy would be happy to talk to you after the talk yeah who thinks that scientists make beautiful products I don’t get it no and we’re

Also not good talking to people anyway um okay so I think we now have the full chain of modeling system the air quality numeric modeling system almost because we still have to talk about the exposure so what is exposure um basically that’s the amount of air pollutant concentrations that you

Are breathing so we are here in the middle we have now our air quality modeling system and now we want to know what are the health effects and there’s a step in between that’s the exposure and I really love this topic because it’s sometimes it’s so simple and then

Also so there’s so many impacts um so what we actually do is shown here on the left and it involves some things that are not happening at home so we are not at home all the time we go out we are at the communication Congress we are on our

Bikes we are doing whatever inside outside different locations but the problems with the established exposure assessments is that they think we’re doing this so they basically assume that we are living on our roofs or at our Windows all the time every day every hour of our lives and based on this

Assumption health effects coming from Air pollutants are calculated and I’m pretty sure that each of you now understood already that there’s probably a big uncertainty and bias in it because that’s simply just not true but of course there are scientific methods to tackle this problem and um on the left

You can see again this this distinction between a static individual staying at home all the time but the concentration changes every hour and on the right you have this mobile individual that moves through the city so we can identify all these areas in the city and the concentrations but it really doesn’t

Help if we get if we don’t get the population activity right so where are you and when and there are of course measurements you can use you can use mobile phone data agent based modeling but there’s also something called the time micro environment activity approach it sounds really aggressive I would say because

It’s whatever um but it’s it’s easy it’s simple and it’s nice I really like to use it so the idea behind that is you classify uniform areas in a city you say these are work environments these are home environments these are transport environments and then you can go and say

Okay where are these environments then you use satellite products for example the Copernicus Urban Atlas you have satellite images they are produced then to land use data and then you can just cluster them and map your environments so you really know where are all these environments in any region you can

Imagine in Europe and there are also some other projects for um other countries in the world of course what you then do is you do some temporal mapping so it’s basically a bit like a mission modeling but with the population so you know a certain fraction of the

Population is at work right now at a specific uh time during the day so or in traffic or at home or wherever and you can also focus it on specific spefic groups of population and so on so you just combine this information and on top you also have the chance to take into

Account the infiltration of the concentrations that are outside to indoor environments based on these clusters you define so you have measurements that tell you only 90% is coming in and you can take this value and combine it and then you combine all of this and you get a dynamic population

Instead of assuming that we are at home all the time we now have some Dynamics and if we compare this now with is the static assumption that we are at home all the time um we did again for hurg because we just love Hamburg and we

Are close to Hamburg so we just do everything we do for Hamburg as a test case um you see here in green residential areas these are areas in Hamburg where people live and in red these are areas where not that many people live depends on what data you

Using um the green shows now um um reduction so that means we moved the people away from the green areas to the red areas and all these red areas have not been considered before and if you remember I showed you the sourcer portion M before like these areas are

The highly polluted areas in hurg so there’s industry and there’s shipping so it’s really important to take into account that people are working there that they are in transport and so on so we can imagine that this will have an impact on the overall exposure and actually we did many many studies in

Europe in different cities to identify what’s the signal here what’s the bias what are we under estimating and it’s really a significant signal in many urban areas so we have up to 1% higher NO2 21% higher pm2.5 and more recently we did a study to tackle this Dynamic problem for all

Um the European 27 countries and we also saw a signal there that it will be underestimated and if you now think about that um the EA the European environment agency calculates the exposure and based on that the health effects and that they say each year 600,000 people die earlier so premature

Death because of um air pollution and you add these five and 3% probably the health effects are worse than we when than we know right now so I think it’s really important but also very easy to take this into account and there was a great um I

Should call it great but for a scientist was kind of great this pandemic situation because um was a real real world experiment so there was a lot of data collected during it and we had some kind of a Baseline and and the difference so we now know because Google

And apple they collected a lot of data and we use this data to to identify what’s the impact of changing activities on emissions and air quality and then eventually exposure so people were less at uh work more at home there were less in transport this all has effects on

Emissions and many many studies identified that the concentrations went down but there were just very few studies that also took look at the exposure so how many were we exposed to how much were we exposed to and we did this and of course when you are not

Assuming that people are still at the same locations as they were before the pandemic the exposure goes even lower so we are happy that this kind of data exists and that we can use it and we are pretty sure that there will be more of

These nice data sets that we can use to improve our emissions that we can use to improve our exposure and so on and with that um we we have to come slowly to an end and this your part my mic working oh yeah um we’re slowly approaching our destination the cryostatic sleeping

Chambers will be deactivated soon um you might have noticed we really would have liked to keep talking and we can lock the do so there’s no way out but um the thing is I guess we could give you an overview of what the thing is this air and this pollution and what

Kind of models we use so for last but not least it’s where are we doing that actually and um so you can run these models on a workstation so I mean I know phds that got a workstation and can do simulations there but we’re approaching a point where really you

Need especially if you want longer time scales you want to do scenarios and go now higher resol Solutions a global coverage like to call it gloal because there used to be this modeling community on the high solution scale and This Global people and with the computational capacities capabilities it’s really uh

Basically becoming one where we can model globally with high resolutions and um it’s actually one thing that’s a bit special about all this this um climate and chemistry modeling is that it’s huge we we can have a a domain with a million crit cells and then we have tens

Hundreds of variables and um it just means we have crazy uh requirements concerning the data um so we can in theory calculate our stuff on any HPC machine but uh that’s a really nice thing that um here in Hamburg we have uh comes on the next

Slide we have the the lant system yeah let me go there um that’s that’s a super computer in Hamburg it’s a German climate Computing system where our Institute is one of the shareholders and I just checked we 74 on the top 500 list but um it’s as I said not about the

Computational power it’s really about the the space so you easily fill like 30 50 terabyte with a year worth of data and then you might have 10 or 50 and so it gets really huge and uh our system is actually slowly approaching the exabyte I’m always

Waiting when I get prompt with like an EB because the PB is still confusing me and I don’t know but that’s something I just want to mention why is there no 16bit floating point in in data formats like net CDF I I get the suspicion that

Half of the dis space is just filled with insignificant digit um but then no one wants to interpolate on some integer uh value and have an an a scaling factor or stuff like that um but yeah that’s what we’re doing we’re calculating uh and using up space and energy and hope that we

Produce less air pollution than we uh prevent with our work and um that’s that’s he could model that no yeah we should be there in 10 years though um yeah with that just so I hope what you got away is that there’s a quite interesting multi-disciplinary topic that covers all

Kinds of natural science and computer science but also has strong links to social science and policy making and um that is actually relevant and although people aren’t that afraid of briefing the air like for example some people flying it’s it’s what probably will kill you and um what’s now happening is that

I mean we’re just wondering that there was no talk about this subject at all we thought there something on climate models but these air pollution models that there has been really a development over the last 20 years that these things are now really getting to a point where

You will probably are already starting to be confronted with that and I mean may it be that you’re an allergic person and you get your Poland forecast or um that maybe you benefit that an urban development planning project was based on on a Model study um I’m just very

Positive that these things that are now moving towards Earth system models we don’t only have the atmosphere but we have the ocean we have ecosystems uh we have even a model where we have like fish and um so actually we were coming to a point where we have like a digital

Twin of the complete Earth system where we can track pollutants where they going where they coming from and can I go outside of my house tomorrow with that thanks for sticking with us um yeah if there are any question well we we have a big problem um I think we

Didn’t mention it yeah we are we are no computer scientists so we apply all of these models on supercomputers but uh sometimes it’s hard and uh yeah we can we we need people like you to support us and we also try to help people entering this topic so for

Example we did a little hecaton here in 2023 and except of us these are all people who measure stuff so they have no clue about modeling and we Tred to explain how to build a model how to run a model on a HPC and yeah um I think

With this that was fun that was fun yeah so but now we already done this is our literature please ask questions [Applause] thank you Martin ra and thank you Jann bizo um if you have questions please queue up at the microphones um and also we take question

From the internet I would start with the signal Angel the internet is asking about two sources of pollution what about vehicle tires and what about if we all switch to hydrogen and there’s amonia everywhere how does how do you see that impacting the world yeah actually that was the

Reason why I just showed um results for NO2 concentrations because for example when you introduce electric cars or hybrid electric vehicles you you will still have Tire where you will have um particles from breaking from the tires themselves so by introducing electric cars we will only partially tackling the

Problem of particulate metal pollution so I think that’s answers a question yeah and it’s that tire and uh break wear is I mean there are estimates in the emission model so we have some factors for different depending on the size and weight of cars but these are really really uncertain

And it’s still a big question mark thank you uh microphone number two your question hello hello first of all thank you for a great talk talk very interesting one uh my question is is when it comes to modeling air pollution in Germany one of the biggest topics is

The shutting down of nuclear power plants and then dependence on gas and then shutting down of gas pipelines and depending on coal have you done or have you experimented with any modeling around like reintroducing nuclear power plants and how it would influence the air pollution in the current climate

Where the gas doesn’t flow that much anymore so that’s not a study we’ve done um let me think a second before I say silly things but um no we didn’t it it’s just that I mean of course I mean Cal fired power plants it’s the it’s the

Most polluting thing I mean it it emits more radioactivity than a nuclear power plant because of the cium in there and um it is definitely pollution wise and not wise decision um but I it’s nothing that would affect the German air quality because the the cold fired power plants

It’s it’s really the the sulfur and there uh we’ve had reduction of like 98% because all the cold fired power plants now have sulfur uh desulfuration so they they reduce it before it gets out of the thing so the coal is really a climate thing with the CO2 mainly but it it

Shouldn’t be like a significant part of what what’s actually polluting you that’s much more from traffic agriculture thank you thank you uh sign Andel your question uh the community is asking how about how are the models impacted by different measurement chesses is it better to buy like one

Very good measurement station or many small DIY almost on and can people get involved helping with that that’s actually a good question and the the an is we need everything in the end because it’s really important to have these reference grade uh instruments which are really expensive uh you can use these to

Calibrate your your small DIY sensors so you get an idea on how they perform and how well they perform and then you can distribute them spatially so you get a better idea of of the spatial coverage but still there are gaps in between and then modeling comes into play so if

Everything works together and interacts together nicely then we get a really good picture of what’s actually going on because as Visa said um sometimes it’s hard to trust models but it’s also hard to trust measurements so we need in the end we need everything and I think we can brainstorm about possibilities how

To work together yeah sure thank you microphone number one what is your question I by Distributing the sensors um you get a really good real time model I think is there any researches regarding the blops you call them blops I think uh how to catch them

Or how to clean them up because if they are in metropoles maybe there is a technique to to work with them so you can collect the trash from floor but maybe you can collect also some trash from air with the blobs you mean the the ship emission particle video yeah but

More in the metropoles not on the ocean um yeah what’s happening is that there’s now um I mean it started that so let me let me start it’s really crazy ships they they drive with heavy fuel oil mostly and oil it’s trick it’s it’s basically at room temperature it’s not

Liquid they they drive with for solid and it’s full with sulfur especially but um a few years ago the the world Maritime organization started introducing sulfur reduction areas where you were not allowed anymore to drive with sulfur content and now actually is that that’s upcoming that’s not there

Yet there’s now the nitrogen emission areas where ships will be forced to actually switch onto a liquefied nitrogen gas and it’s a super interesting topic because uh having the sulfur emissions away from people that’s really good but if we now use the desized fuel that also means we’re not

Emitting sulfur above the Atlantic above the Pacific and there’s is super effective coolant as a as a particle so um yeah because that’s that’s really something that’s again from the polyc perspective a bit it’s I think it’s crazy you have these these areas and they are I don’t know how many miles

Away from the coastline but then there’s a point when you exit these these coastline this area you can again just burn whatever you want so the regulation is just valid when you’re close to the coast and this is as you as you saw that doesn’t make sense because the formation

Happens also far away from the coast and in land but in the opposite way that you have something like a smoking room where they clean the air so you have a area where you can create some pollution which is not should not be our future but um and then they have some technique

To clean to yeah but I mean you’re completely helpless it’s gets transported with the winds and diffusion and so you could have some fancy artificial trees everywhere I mean it’s it’s really if there wouldn’t be that massive amount of manure and ammonia emissions from the animal husbandry and

For the nitrogen oxides it’s really mostly the uh Vehicles traffic so if if your Cloud plume from the ship comes on coast and there is no pi Farm then also there would probably be less particles produced thank you for your answer uh microphone number two your question hello thank you for your

Interesting talk I also had one of those sensors in operation that you mentioned that measure um pm2 and um pm1 2.5 and pm10 and um what I saw in my data was also the traffic patterns that you mentioned it was very clear but um what

I also saw was um the high peak at the end of the year coming from the fireworks and I would um um ask you um what is your position on a um B of fireworks do you have any opinion on that and given your knowledge also of

Course I have a personal opinion on that and I I don’t like it I would ban it um from a air quality perspective and EM Mission modeling perspective it’s really interesting because we just don’t take it into account so whenever we we model the January and the 1 of January and

Also the 31st of the of the the last of the year then we Al always have this peak in our models when we not in our models but in the measurements and when we compare it we always we are far away from it because we just don’t take into

Account because we are using these temporal profiles to distribute emissions but such things as holidays and special occasions or like 500,000 people meeting to grill in the park or whatever it’s just not in our models because it’s so specific that we can’t model it in general well of course you

Can do it you can do a specific study about it but if you look at the whole city or country or even the world then such events just disappear um yeah so no it’s not in our model and it’s also very bad for cities that are shaped like a pit like like

Hustle or St and so on definitely yeah yeah yeah we didn’t talk about orography yeah thank you for your answer Mone number three your question hi so if you spend a lot of time online you quickly read about this new class of pollutants the per fluorinated ALU substances or

The pasas and I was wondering if that is relevant at all to your research in air pollution or if it’s rather on the turf of water pollution and if it’s relevant if you already have the ability to measure and model them actually we are running a little group who’s dealing

With this topic uh so we we have um did you pay him to ask the question I didn’t thank you thank you for the question um so as as we were introduced we are pollution modelers so we are not focusing only on air pollution we are looking at many pollutants and not only

Atmosphere but also in the ocean and yeah so for peas we actually uh published a global emission model for pasas into rivers oceans and atmosphere that covers the whole production period from the 1950s to uh 2020 that’s available in this echad database where there is also the ipcc emission um

Databases are and I really love the pe’s topic because as modelers it’s always like the initial condition what’s the initial condition and we just start 950 the substance didn’t exist it’s like I have an initial condition that thought it would never happen I can die happy

Now and um we we actually just uh we have two PhD students we’re supervising and one is working on the atmospheric icon model and the other on the the ocean icon model and we’re just um kicking them not to have vacation and finish the model space no no we’re nice

Supervisors and um so but yes we were actually working on having an atmosphere ocean coupled model where we can also look at things like because there acids once they’re in the ocean they shouldn’t come back but then you can have sea spray and it actually when the sea spray

Evaporates you have a mechanism of getting it wind driven back into the atmosphere that’s probably a factor in the long range transport and now there’s like new Europe regulations coming up that bends like 24 p and so just to add one more thing because we also do

Measurements on mainly in water but also some in the atmosphere and there are some other groups around the world that are doing that too so actually that’s a big problem with on on the agenda for measurements and modeling and yeah we’re on it thank you for your answer uh I’m

Afraid time’s up are you still around for people to contact you for for further questions yes awesome you never stopped talking yeah K thanks for your great talk thank [Applause] You