Lightning lectures: Seeing the air we breathe
Okay uh firstly, good afternoon everybody. Um, i'm delighted to be here, uh to. At least give you an insight into some of the research, uh we only have 15 minutes so uh i'll have to be quite quick in both, talking and moving through the slides, um just a quick question how many of you traveled in by uh, tube. Okay, yeah how many by bus, no yeah a few, what about. Cycling, walking. One or two okay. Good car. Yes, that'd be a bit tricky, going here right now moving car that would be something. Okay but uh let's ask another question um. Out of those four options, uh quick show of hands how many of you think. Which which option, you will expose you to the most air pollution, so what about, a tube. Okay. What about. Buses. Okay, a few, what about. Cycling or walking. A few again what about cars. Okay, one or two a few, okay, so you can see. That we're all quite confused, we don't really know which which, mode of transport, actually limits the amount of pollution, that we're exposed, to but there's a reason for that and it's partly because. We don't really adequately, measure, pollution. With enough resolution, to really understand. The dynamics. For you know for individuals. So um. Just to put pollution into context it is a global issue. About 10 15 years ago, 16, of the 20 most polluted cities were in china. More recently, 14, of the most 20 polluted, cities, are in india, so it's definitely a global. Phenomenon. And in terms of london. This was something that, near the, start of my research as an as a postdoc. This was developed by a model it's a model of london and the pollution levels obviously red means very very, high. And this is what they hope to be by 2010.. Does anybody know what this blob is on the left-hand side by the way. Heathrow, airport absolutely. Now the reality, is they didn't actually this didn't improve, over that period of time, and it led to a lot of. Different headlines. Regarding, air pollution. Threatening, with fines from the eu, etc. And part of that is because this model. Was actually built. Or created. Based on about, 120. Sensors. To cover about 600, square miles, so there's about 120. Fixed, air, air pollution air quality monitoring, sensors. Uh, throughout london. And you think about that 120. To cover 600, square miles about one every, five, square, miles, is not enough for you to adequately, know what the pollution is on your street or, or even in your, local area. So. Let's just think about pollution, why that might be well, first of all. Pollution. Very rarely can you see it it's very rare that you see. Most of the pollutants, that we are subjected, to so, for example, you might see a little bit of smog. And possibly, some aerosols, some particles, if they're big enough, but most of the rest of the pollutants. You know the sulfur dioxides. Nitrogen dioxides, etc. You will not see them even these small, particulate. Matter. Pm10s. And pm 2.5, it basically means, the size of them is about. Either, 10, or 2.5. Millionths of a meter in diameter, so they're extremely, small. And so as a result. We have to find ways of measuring things that you can't actually see. Because if you think of maybe. India or china you probably there's lots of images of the thick smog so you can clearly see it and there's a problem but, in somewhere like london you won't necessarily, see all that that haze. So part of my research was involved, in. Actually. Uh. Using, transport, or at least. Monitoring, transport, activity. And trying to correlate, that with the resultant, air quality, that's produced, locally, and this is all based around urban pollution, in cities like london. And the main reason for that is because poor air quality if you sort of follow the throat with a flow chart. Shows that it will lead to an adverse health effects, and ultimately, lead to. A social, impact in terms of health generally and also an economic impact in terms of the overall burden. On, the nhs, services, for example. So the project was called message. It's actually an acronym, it stands for, mobile, environmental. Sensor systems, across a grid environment. Luckily it's it's spells message right. But basically. What it meant was that we would develop sensors, that you can place on moving, platforms, like buses, and, and uh taxis, at cars etc. Um, and then, you would then gather, all that data so effectively, what you create, is an ad hoc network, of sensors, that are moving, and then the idea is to develop the infrastructure, that will allow you to gather that information. And then use that to interpret. In real time, what the dynamics, are of pollution, in a with much more resolution, than your, you're probably. You'd get from just modeling, a sparse. Distribution, of fixed, sensors. I suppose i should, mention that it is underpinned by some fundamental. Science. Just very briefly. When atoms and molecules, which, most pollutants, are, when they absorb.
Particular, Wavelengths, of frequencies, of light, they will become excited. Their states will become excited. And then when they relax, back to their natural. Ground state, they release. A photon, of energy. And that energy is in the form of, a wavelength, of light and by light i don't just mean the visible light that you can see, but light can be right across the macro, electromagnetic. Spectrum. And in particular we worked in the uv. Range, because that is particular. That energy range is particularly. Suitable, for monitoring. Air pollutants. And the result is that you what you have. Is that. Different pollutants. Have different signatures. In this spectrum, so for example this is uh sulfur dioxide. Which seems to have this sort of all these sharp peaks. Near at near near well this is 200, to 230. Nanometers. But again. You've got other pollutants, like no, no2, benzene, which have completely different profiles, and we use this difference to determine. What types of pollutants, are in the air, and how much of it is there. And how we did that was combining, two techniques. My own research, background. My phd was in atmospheric, physics, so we were quite well aware of a technique, called doas. Which is another acronym. But basically. What it does it uses a sunlight. As the the source of light, and as that light travels through the atmosphere. Different pollutants, will absorb or different chemicals, generally, will absorb, different frequencies. And then we, basically detect the light, at the other end if you like, and compare that to what you started with, and then we use that to then determine. Uh how much pollutant, is there and and and uh. How much of it is uh what type it is, and. The thing about the atmospheric. Process, doas is that you've got this very long, optical, path. Of several kilometers. Through the atmosphere. And that allows you to pick up very. Low concentrations. Of chemicals, because they'll give you a stronger, signal, over a longer path. And then, in order to measure air pollutant. On the ground so to speak we don't, we can't, be afforded the luxury, of measuring, over very long path length so we we we reduce it from say, a few kilometers. To maybe a few meters, across, a roadside, for example. But in order to keep the sensitivity. Of the instrument. We then combined it with some, techniques, from, high energy physics, because they go into high particle, collisions. Trillions, of collisions and they're looking for just a few hundred. Uh, sort of signature, signals, and so they're really looking for a weak signal in a noisy background, so by using. A very robust. Noise reduction, techniques. You can actually reduce, the amount of noise, in your system, and what that means is that your signal to noise ratio, remains the same, as you go from a long path to a short path. Unit. And as a result. You keep the sensitivity, of sensitivity, of your instrument so that you're able to monitor. Air pollution and trace quantities, at very low low levels like parts per billion, levels for example. What this means in practice, um. In the in the the early research. I mean if this looks like a violin case well it actually is, um, and that would seem to be the most suitable case to to house our unit and we wanted to replace, the capability, of say a fixed.
Monitoring, Station. This is a typical fixed monitoring, station. Apparently this is a mobile, station looks very similar in size except it's got wheels. But basically what we wanted to do was then use that same capability. But in a unit that you can carry around. Walk around, put it on you know cycle. In a very sophisticated. Way put it to a vehicle. And drive around and so you can start to map pollutants. I should also mention that there is, before you it becomes a sort of a viable, product, there is an engineering, phase, that is involved. So as a scientist, we're quite happy once we can prove the principle that something, could work. But an engineer actually makes sure it does work, uh in the environment so uh i've, got to give credit to the engineers as well. What this means is that you actually get, you can generate, maps so this is, around. The campus of imperial, college that's actually the royal albert hall there, and this is when we did a few circuits, around and actually. Aggregated, the data to get an understanding, of where the the key pollution, sort of levels, peak, and you can see they sort of peak at the junctions. Which is probably understandable, because you've got more traffic in, in, in both directions. We've done similar exercises. For example we put a unit on a long boat, and, took it all the way from west london right across east london, and again mapping for different pollutants, and you can see the height. Represent. Where. The sort of elevated, levels, are, and this type of visibility, you wouldn't get that level of resolution, with the current. Fixed set of monitoring, stations. Another interesting, um. Uh sort of investigation. We did we we drove, a, sensor or vehicle, and monitored pollution, from sort of south london all the way through to, earth's court, and we actually identified, at the time, two, potential. Pollution, hot spots, one was across vauxhall, bridge. That may be understandable, if you've been in that region there's lots of multiple, lanes of traffic in, in either directions. But this one is quite interesting, because at the time. The hot spot was identified, near harrods and harvey nichols. And that poses some interesting questions for society, let's suppose, and i'm not saying it is but let's suppose there really was a permanent. Pollution, hot spot, near harvey nichols and harrods. What sort of information. Would. The authorities, feel was was adequate, to release to society. And this is these are discussions, which you have to think about because they can affect. Things like i don't know house prices, for example. If you if you've got young children, that dream penthouse, that you wanted in knightsbridge. It may not seem as attractive, anymore for example. So these are things which you may want to. Which which if you like the powers that be have to consider. Another example of how pollution can vary quite quite considerably. So if we if we take the sort of legal limit, roughly, is about 100, parts per billion, so it's down here somewhere. But this is where we went on a we took two units we were walking side by side with them and you can see. Initially they follow us very similar trend, but at this stage one, one decided to stay at roadside, or curbside. The other one was in the central reservation. Of a busy, busy main road and you can see there's a huge spike. Uh in the middle of the road as opposed to on the sidewalk. So going back to the question about. Uh which, which mode of transport are you more subjected, to well it depends i mean technically if you're in the middle of the road, which you are if you're in a car, you may well be, subjected, to much more, pollution, than if you're at the side of the road. This is another quick example of, pollution, spikes near traffic lights.
So These spikes, represent the pulsing, of traffic lights and and the engines revving up as they drive off, and again. You can see you've got some very high spikes. And the, the body has a non non, linear response. To uh pollution. Spikes so in other words, being exposed to a high dosage, over a very short period of time is far more damaging, than being exposed to say, the same dosage, but spread out over a longer period of time, so again it makes certain. Jobs if you like like i don't know selling flowers on a street corner or newspapers. Actually, could be quite dangerous. Especially if you've got respiratory. Issues. So, that's about one sensor. And one sensor can do quite a lot, but the real aim i'd almost compare it to for example. Whoever, you know designing, the first pc. A pc in itself can do, a huge amount but if you have, a complete i.t, infrastructure. Then you can do a lot more, and that's really. Where the key, areas of the research, are so this is a scenario. Around east london. And basically we've got about 140. 150. Sensors. This is based on a scenario so there weren't real sensors deployed. But we said imagine if we had a, deployed, sensors in this nice fixed sort of manhattan. Grid style. What would the data look like if it was monitoring. Continuously. And, you'd get something. Oops. What's happened here okay, so you get something like this for example. During the morning the main sort of archery is the main road, our trees are pretty polluted. And then. As rush hour starts to die down. These arteries, you know these pollution levels start to fall, you may find pockets, of pollution. In different areas, and this is a type of visibility, that would be really useful, for local authorities. Okay because then they can make real-time, decisions, like rerouting, traffic, and limiting the impact, on human health. And also you could do things like. Analyze, the impact of the school run. On the local. Air quality, around the school, in this particular, snapshot, it kind of demonstrates, that the school run actually, does impact. On the on the the local pollution, so at 3 30, it's still polluted. And it's outside of rush hour, for example. And so these are the types of, details, that we really like. And we've also done studies for example. We did a project in brighton where we covered three primary schools, which were located, near to main roads, we monitored, the pollution. From the main road and the emissions, from vehicles as they drove past but we also. Simultaneously. Monitored. Uh pollution, within, the school grounds itself, to see what what the correlations. Are. Again. It's no surprise, and, it's quite quite interesting that a few years later. We had headlines, like this. In london, where, you know they now identify, that locating, schools right next to main roads, is probably. Uh not the best, best idea especially. Because some you know many young people, have have. Respiratory. Issues. So in terms of the research. The real challenge, now. Is really well okay. It's not really realistic, to deploy, 150.
Sensors It to that level of density. In, a part of london. But the question is what happens if we could replace, these sensors with maybe, fewer moving, sensors. And then could we still get the same level of, of detail, the same level of interpretation. From fewer moving sensors, because that of course will be much more cost efficient. So how we approach that, we say took the original 140. Sensors. And then initially said well let's assume now that we've got seven sensors that move in this sort of cyclical, cycle. And then look what the data will produce, and compare that to what we we had originally. And then we, would make more complicated. Irregular, paths. And then finally we sort of said well let's tap into the bus network and assume that we had them. Fixed to buses so that now the routes are only limited. By, the bus network, and what you end up with is, this is the original, plot that you would have seen from a fixed. Sort of data, fixed, grid of sensors. And this is the the data from the mobile, sensors. And this is the difference between the two in other words it's sort of highlighting. Uh what information, is missing and you can see there's very little so it gives us great hope that we can actually. Use mobile sensing. To actually generate, these pollution, maps. By using, intelligent, sort of sampling. And reconstructing. The data, with fewer sample, points, and that's really where the researchers. As, led to at this, present day, i should also mention a couple of other benefits. Of research. One of which, is. Innovation. So in fact even though we did most of our research based on air quality monitoring. In fact what we created was a, multi-multi-species. Gas analyzer, like a platform, that can analyze, lots of different types of chemicals, in the air, and so what you find is that different industries. May have an interest in different types of chemicals. So for example the oil and gas industry. They're quite interested in hydrogen, sulfide, because it can lead to cracking of pipelines, and so on, so again. Our units are able to use the same technique. And tailor it towards, monitoring, for hydrogen, sulfide. And there's lots of other areas, for example defense, might look for biohazards. And so on. So that's one benefit, of fundamental, research, i would say, is that it can lead to innovation, in many different, areas it's a platform, technology. And the other benefit, which might be less surprising. Is creativity. So in fact, i would argue that if you have a creative pursuit. You could actually. Use your technical, knowledge, to enhance, what you do creatively. So that's so a personal, example. For many years, i was a part-time, dj known as dj, chemist. Fair enough. But, the thing that i used to like was remixing. So hopefully you know maybe you know what a remix is where you take elements of one track, you take elements of another track and you fuse them together, and hopefully create something that sounds fresh and cool and hopefully better than the originals. Now in order to do that what you're really doing is combining, audio wave files, in a particular, way and making sure they fit in a certain way. And when a lot of the remixing, went digital. I actually realized that i was almost doing the opposite, of what i was doing in my research, where we were taking, light waves, and effectively. Disentangling. Them so that we could find out the the the sort of contributions, from all the different, uh chemicals, so you're sort of breaking apart, light waves, into its separate components. Whereas with a remix you're combining. Separate components. Albeit. Sound waves. So what it what it meant in practice. I found that you could really push the envelope. In terms of, what sort of genres, of you can fuse together, so i sort of used my technical.
Know-how, If you like to push, and stretch the limits in terms of what sorts of things you can remix. Now in the interest of time, i can't actually. Play you anything, however, you can check out, dj, chemist. On soundcloud. And there are plenty, of uh tracks that you can uh you can enjoy in your own at your own pleasure. So. Just to summarize. Just to summarize, this particular. Segment. Mobile sensing. It does have the power, to really give you the visibility, that you need. Um, i mean a lot of the the legal averages are based over hours of data, 24, hours but of course we breathe in every few seconds, so we want to know much much more about the dynamics. Also, whilst you may have the general, trend of, what the pollution level is in your general, area you actually want to know how what you personally are exposed, to. And mobile sensing certainly offers that capability. And you can. Capture most of the the trends that you would have captured with a fixed dense network. And then the other thing is i suppose. Is that in order to change policy. Because ultimately you want local authorities, to adopt this type of methodology. But they are not going to do that unless effectively, they are mandated. Mandated, to do that because they don't have a spare budget to buy, lots of monitoring, equipment. And so, they are influenced, by what the what what the legal requirements, are. And the legal requirements. They say it's based on best available, technology, but often it lags behind, the technological. Development. And the only thing that moves things forward is often, public concern. And so that the policy makers, or politicians. If you like, actually start to take note so i would bear that in mind, and then on a personal, level. My only research, interests. For any of the youngsters amongst us were maybe thinking of going in a career in research. I started off as an undergraduate, chemist, and then i did a phd, in, physics. And then in terms of research. Pretty much nowadays, is more maths and computing, and modeling. Of of sensors and not so much on the instrumentation. So, research is a really good. Career, to sort of evolve into many different fields, and then finally i'd say to tackle global, challenges like air, air pollution. You actually need elements, of what i've called three things. Conveniently. Condensed, into an acronym, ice. Which is innovation. Creativity. And enterprise, and if you combine, those in the right way then i think uh we can tackle, most uh global challenges. And it also means that there's always something, for somebody to get involved, in, so, thank you for listening. And, yes any questions i'm happy to feel them. Okay. You asked. At the beginning. You know which which way of transport, might be the most. Polluted, but then you didn't mention the tube so how is how's the tube with the er air quality, well i'll be honest with you there is very little pollution, monitoring, on the tube itself.
Um We. We, monitored lots of things but we didn't really get to the stage where we could we we did much. Research, on, pollution, in the tubes but if you think about it. The tube itself where does it get its air from it's not, fresh clean air coming from somewhere so it's going to be. And also it's probably going to be less likely to dissipate, so in theory. You know the tubes could be just as, as polluted, as as anywhere else but it doesn't require more research, and as i said. My main focus was to develop the capability. Of doing that but to enable. Other groups interested groups to actually start. Looking into into those areas. Okay, there's a question at the. Back. Comment on the question my comment is um, my group in cambridge we actually have a soundcloud. Song called super photons. Excellent so i think other people also do music and my question is how long does it actually take or, do you actually get your. Research, implemented. Into, policies. Right well uh so. In terms of. And it's a good question, i think, the short answer is it takes a long time to change. Legislation. Even if you have the technology. And you prove that it can work so. Part of the, part of my research involved commercializing. The technology, so i'm co-founded, a co-founder, of a, business. Technology, business called duvas, technologies. And at that time when it was first formed, probably nearly 10 years ago. I was uh, tasked as the business development, and i spoke to lots of local authorities, and interested, parties both home and abroad, and what i found is that local authorities, whilst they buy into the vision, and those on the ground in the environmental, section, would love to have this equipment, whenever it came to applying for budgets, and so on and applying for funding. The message would be well we're not mandated, to do it so. You know apart from it's a nice thing to do. There's nothing more that forces local authorities, to start prioritizing. These areas. So. The way how it works you have to influence the air quality expert, group. And then they hopefully will influence the legislators. Who will then change policy, but what really makes a big difference, is when there is an increase in level of public concern. And that has happened over the years when i started this pollution, tend to, research, 10 to 15 years ago, it really wasn't. Something that that. Captured. The public's, sort of. I don't say imagination. Because i mean that makes it sound grand but it didn't really capture. The public in the way that it does now now if you go on google and put air pollution, in the news you will see lots and lots of different types of articles. So i'm hoping that that legislation. Will now finally, catch up, with technology. And maybe the business will sell a few units, luckily as i said it's a global. Phenomenon, so there is lots of uptake, in other parts of the world, where perhaps it's not so bureaucratic. For example. Okay. Yep there's a question, there. When you are doing, comparing. Your static. Sensors with your mobile ones you did. Different pathways, did you find like an optimum. Pathway. Um. Yeah. Um. Absolutely, yeah it's a good question, so, when was looking at the different scenarios, for the for the moving sensors. What we was trying to establish, so we we took if you like the fixed grid as our standard, to say this is ideally, the type of information, we would like to see, but then we generated, new sets based on moving sensors and saying well what would they measure, if they were moving in a particular, way, and what we found was that, in fact. An, ad hoc network. You lose very little of the key. Information. However, it is always, useful. To. Supplement, that with a small, number of fixed stations. So in fact, the current existing monitoring, stations we're not. Sort of advocating, we replace, them it's just that we can now augment, what we already know, and also.
Validate. Often what the models predict, by actually being to being able to measure, much more. Specifically. In terms of time in space. Okay. Oh i think there's a question. Thank you i was wondering. What's the current, cost, of a mobile, sensor, and are there any plans to make them more affordable. Right okay. So um. The current i can't say the current cost but, the the units that we developed, of the order of several thousand pounds, each. Okay, but just to put that into context, at the time the fixed stations, those fixed units that looked like sheds, they were about 120. 000 pounds to construct. Um. Since then there has been development, in in in pollution sensing not just by. Our own, uh company duvas, technologies, but other, uh, entities. The issue is depends on what technique you use so you can get some very cheap sensors. But they won't be as accurate. So it's almost like you want something. Fast, accurate, and cheap and you can always only get two out of the three. And, that's so we went for, accuracy. And and, and you know fast, sort of high throughput, which inevitably, means it's going to be slightly more, more. Expensive, than maybe going for something less accurate. Okay a. Couple more yeah sure. Okay. I'll go faster. Um, so kind of following on from her question the current sense is the size of a violin, case which is obviously, pretty remarkable, on a car, maybe less noticeable, on a bus, yeah but you said that, networks, are the best do you think that they might be miniaturized, made even smaller in the future. Um i think they they probably will, but there are some. Limiting, factors i mean you notice that we're using a technique, which originally, relied on a huge path length of, as i said several kilometers. And every time you reduce, that path thing so if you reduce your path length from several kilometers, to say several meters. You're reducing. The strength of your signal by a factor of a thousand, so it's only because we employed, some very robust, noise. Noise, reduction, techniques which were developed. In high energy physics, and applied them to this, particular. Unit that we're able to keep the sensitivity. So the limits are that if you go to shorter and shorter paths using an optical, technique. Then in fact you're going to lose the sensitivity. And because you're monitoring. Pollution, unless it's extremely, high. You want to monitor, ambient levels as well as, when they you know exceed. So i think. There are. They will get smaller, technology, always tries to make things smaller and faster. But i don't think it's just simply a case of you know, as chips, get smaller. The units themselves will get smaller because there are some hard limits in terms of, the setup you can use other techniques as i said, chemical, techniques, for example. But then the problem with them is that they have to be. Stationary. For several minutes, before you can start measuring so therefore. Having a chemical, measuring technique on the move, means that it's never stationary, which means it wouldn't be accurate, again so there are these trade-offs, but i believe they will get smaller, and ultimately.
You Know the dream would be that you know all of us would have one, somewhere, either on our phone as a, sensor, or some something you could wear. And then use all that metadata. To create this large, network so that we really get a good understanding, of the dynamics. In real time. But we're i think we're still a little way off although i did notice the mayor, had commissioned, put some funding into, having schools. School children. Come into school with backpacks. And with, pollution, monitoring. Sensors on them, and then start to monitor the different routes in which they they come into school and what they were exposed to so going back to our original, question maybe it will be answered by, by school children. In due course. Okay, well, thanks again thanks for your. Time.