Sensing an Airport Environment - SNAQ Heathrow
Ben - Sensors are essential for our understanding of the environment around us. Weather stations for example have been collecting data on wind speed, rainfall, pressure and humidity around the world since at least the 1940s. And now, a new project led by the University of Cambridge with a host of university and industry partners, and funded by the Natural Environment Research Council, seeks to deploy a network of sensors around Heathrow airport in order to study the atmosphere with unprecedented resolution. I met Doctor Iq Mead from Cambridge University to find out what it is that they hope to learn.
Iq - This project is really looking at urban air quality in the first instance and then we're trying to tie that into specific science questions within that; specific airport questions such as the dispersion of the buoyant aircraft plume or distribution of sources around an airport, what does it do if you increase the use of one side of the airport versus another, these kinds of things. One of the beauties of this kind of network is that you can disperse it over wide area and in a fairly dense mode, to get a lot more information than you would actually need to answer that particular question, so you can tie the data into other issues as well around it.
Ben - Why airports in particular? Why is an airport more interesting than a city?
Iq - This project is built on previous projects, looking at urban and rural concentrations in towns and cities in the UK and abroad. So we've deployed in Cambridge, London, Kuala Lumpur, and Lagos. The idea being that we want to understand what's happening in that environment for those particular questions. For Heathrow airport, we have a rather unique situation in that we have a very large airport surrounded by a very large city, and we need to understand what's happening before the airport, as things get into the airport, and what the outflow is from the airport, now that it affects a wider area.
Ben - So what is it that you're actually sensing? What are you picking up?
Iq - Broadly speaking, we're looking at a number of gas phase pollutant species such CO, NO, SO2, ozone, but we're also looking at carbon dioxide, speciated particulates and the wind speed and direction for like particular measurement site. What we can do with that is really understand how the pollutants are moving through our environment, directly tied to where we're making the measurements as opposed to a bulk, "the wind speed is this number of knots and in this direction" and then crudely linking it.
Ben - So each one of your sensors is obviously capable of measuring a large number of different factors. Why do you need to have them in this network? What's important about the structure of the network?
Iq - Actually, the whole thing is a very modular system in as much as it's a collection of sensors in one "sensor node". Those nodes are then put out in a dense network. The key to the network is that we can then start to understand at an appropriate granularity in time and space, the things we're trying to measure. We can do a lot of interesting things with dense networks we couldn't necessarily do with more dispersed networks. For example, the validity of the data we're collecting is essentially tied in to its calibration. If we have a dense network, you can then rollout a calibration to very specific nodes and then they will tell you a lot, or you can infer the behaviour of nearby nodes. So you can take your study for validation across the network without actually having to maybe go to all sites and calibrate them in a traditional way although we will be doing that to some of them.
Ben - So you could for example, if you have one node in between two others and it doesn't quite agree, you know that that one needs calibrating differently because presumably, there will be a gradient across those three.
Iq - Of course, depending on your local sources. It could quite correctly be varying differently to the ones around it. However, if you know that site A to site C is roughly the same source pattern and then site B is in between, you should be able to understand what's happening in site B using site A and C.
Ben - What sort of density are we talking about? How regularly spaced are these sensors?
Iq - It's an irregular network. I mean, in part, we have to be practical. We can't just turn up at an airport and say I want to put a sensor on a lamp post here - it might be in the middle of a runway. What we're going to be doing is locating a number of sensors around the periphery of the airport and then also, dispersed within the airport at strategic points including on the ATC main towers so we get a vertical transect as well. Put it into context, across the UK, there's about 120, 125 static monitoring sites used for air quality as part of the main DEFRA AURN monitoring network and we're going to put in 50 plus sensors across Heathrow alone.
Ben - So that's clearly far more dense than what's gone before it. Presumably, these are all capable of talking to each other and there are ways to collect the information remotely. You don't need to send somebody to every single sensor to pick up its data.
Iq - Well absolutely. I remember when I first started, if you put a sensor out, you had to go and collect the data from it with a cable. These sensors, we're collecting data autonomously and then sending it to a remote server at 2-hourly intervals, I think is the current plan. The idea being that we have all that data coming in semi-real time to us. We can start our analysis on day 2 of deployment and you can really start to understand what's happening, and then you can actually look at your network and say, "Well this sensor is redundant here. Maybe we should move it there, or this is an area of particular interest, we're looking at a particular question, let's see if we could put up a few more sensors there." Augment the network as well as tie it into the existing network because they do work together. Our sensors are not as intrinsically sensitive as the traditional static sites but they more than make up for that in their mobility, and their ability to be put out as this dense networks. It's all about the appropriate scales of measuring for the particular species in question.
Ben - So do you have to design very specific sensors in order to play this multifaceted role of picking up lots of different things and communicating it, and acting as part of a network?
Iq - I suppose what we've done is taken advances in various technologies, and put them together into a sensor node. So, within that upcoming deployment, what we have are a series of electrochemical sensors which were originally designed for a much higher concentration in the industrial sphere. And we started to work with those. They're now coming down and they're just at the point where they're sensitive and selective enough for this kind of application. We also have a photoionization detector, there's a small folded path spectroscopic device looking at CO2 and a sonic anemometer for our wind speed and direction. Put all these things together and the intelligence needed onboard to control all these constituent parts is a large part of the network design, then how to deal with all that data afterwards becomes an even larger part.
Ben - It seems appropriate that you're putting this at an airport because presumably, you face some of the same challenges of the people who build and design aeroplanes where they have to miniaturise and condense all of their sensing equipment as well.
Iq - We do face constraints on things like power and weight. However, they're not quite as acute in as much as it's not going to cost us more in fuel, for example, to carry a larger power pack. But in terms of practicality of a network, you have to take into account the operation and the lifetime of the network. You have to understand that it's going to take a certain amount of power, are you going to be able to scavenge power from local sources, do you have to put power there for it - i.e. big battery. You are under a series of constraints regarding how often you can send data, for example, because switching on an antenna takes quite a lot of power, and each one of these things essentially has a mobile phone on a chip onboard internally. So you have to control how much power you're using and then you have to think about how much data you're sending, the cost associated... These things snowball a lot.