Link to slides/notes
I’ve not studied geography since year nine, but am interested in our environment and visualising data, so opted to study a Geographic Information Systems module for my Masters degree. Bellow is the report* I submitted, where I have had a go at estimating the size of the population affected by poor air quality in the London borough of Camden’s five designated Air Quality Focus Areas. In the conclusion and appendices at the end I make some of my own suggestions about what we can do to improve this air quality. I didn’t know about https://www.londonair.org.uk/ until after submitting, but it is very relevant. If I had time, given there is data on the quantity of traffic on certain roads, and we have the number of children on roll at school/nursery, I would try and estimate the number children who come near to the high use roads on their way:
* Note because the module might set a similar assignment in future, I have removed content at the colleges request. It now displays ” # Removed content was here…” .
1. An introduction with background information for this report and the aims of the report
Air Quality Focus Areas in London
“There are 187 Air Quality Focus Areas in the Capital (2013). These are locations that not only exceed the EU annual mean limit value for NO2 but are also locations with high human exposure. The Focus Areas were defined to address concerns raised by boroughs within the LAQM review process and forecasted air pollution trends.” ((Elliot.Treharne@london.gov.uk), 2013)
The brief for this report is: “Camden Council would like a preliminary analysis for estimating the approximate number of people who are affected by poor air quality, as defined by the London Air Quality Focus Areas. In particular, they are interested in if and how the locational relationship between the Focus Areas and other features such as the road network, underground stations and the green spaces in Camden can be used to explain the size of the affected population.” Camden has five focus areas which are:
1. Kilburn High Road from Willesden Lane to Grevill Place (ID: 30 | Old ID: 108 )
2. Marylebone Road from Edgware Road to Euston/King’s Cross Junction (ID: 31 | Old ID: 127)
3. Holborn High Street and Southampton Row Junction (ID: 29 | Old ID: 134)
4. Swiss Cottage from South Hamstead to Finchley Road Station (ID: 32 | Old ID: 136)
5. Camden High Street from Mornington Cresent to Chalk Farm (ID: 28 | Old ID: 137)
Emissions and their impact in Camden
Journalists at the Telegraph, a national newspaper looked at the government published statistics, and present some of these in an article designed to alert and catch their readers attention. The article is called “Mapped: Where is air pollution killing the most people?” For Camden, the map on website http://www.telegraph.co.uk states “In Camden there were 87 deaths due to air pollution in 2010. A total of 1157 years were lost. The mean anthropogenic PM2.5 (µg m−3) was 13.8, which is a measure of the air pollution in the area.” (Telegraph Media Group Limited: Sarmad Jawad, 2015)
The main pollutants of concern, with London emissions recorded in 2013 are: Nitrogen Oxides (NOx), particulate matter (PM10 & PM2.5), Carbon dioxide (CO2) . The sources of these omissions and how they are recorded spatially in GIS data systems can be seen from this slide of the London Atmospheric Emissions Inventory (LAEI) – workshop:
(Greater London Authority (GLA): LONDON ATMOSPHERIC EMISSIONS INVENTORY (LAEI) 2013, 2016)
Aims of report, sourcing & exploring data and making decisions
The aim of this report is to estimate the number of people affected by poor air quality in the five focus areas that fall mostly in the London borough of Camden. To do this, the different sources of population entering, exiting and living in the focus areas would have to be established. To get started, I searched for the best quality and up-to-date data from a variety of sources (see this reports list of data used) and used ArcGIS to aggregate these datasets to get a high-level overview and compile descriptive statistics:
# Removed content was here…
Decision 1 (Location of resident populations)
Notable was that the five Camden focus areas intersected with four other London Boroughs: Brent, City of London, Islington, Westminster. See figure 3, “LSOA base population”, row two. For this reason, the decision was made to source and merge population data for all these boroughs with spatial data, into one GIS layer.
# Removed content was here…
Decision 2 (Apportionment method of resident population to use)
# Removed content was here…
Decision 3 (100m high risk zone and applying a 50m buffer to Marylebone road 50m)
In his website article Bill Adam’s cites research and states: “On average, PM concentration is significantly higher within 330 feet (100 meters) of major highways than it is further away. (Zhu, et al., 2002)” (Adams, 2015). For this reason, it was decided to add a 50m buffer to Marylebone road focus area edge, as is in some places the distance between the road an this edge was considerably less than 100m. See section 3 for more information
Decision 4 (Population sources not included in estimations)
Other populations exposed, but where further research needed to determine the number of people affected, and where there is no readily available data are:
- Bus and Coach users entering & exiting the focus area
- Cyclists – NO data available and Cycle roots in a ITN Urban Paths data outside all FA
- Pedestrians walking into FA e.g. walking to school/work within catchment area but starting beyond FA boundaries
For possible cyclist’s exposure, some effort went into converting ITN Urban Paths data (GML format) to an ArcGIS compatible format using Feature Manipulation Engine (FME). However, it was only to find to my surprise that there are no cycle paths intersecting with focus areas.
I would have liked to include data quantifying pedestrians walking from homes outside the focus areas, into the focus areas and returning, particularly children of school age, as they are at the highest risk to health due to their height and stage of development. Instead “Important Buildings” data (see list) and symbology were used to clearly present proximity of schools and hospitals:
Decision 5 (Time: minimum daily exposure of affected population)
One general assumptions made, is that to be counted as affected population, a person should have a minimum of two minutes’ exposure to pollutants emitted in one Focus Areas, in a minimum of 1 day, in a year. My hypothesis is that two minutes is enough time for someone to exit a station in a focus area and then leave the FA by the shortest path. Because all estimations are the number of people effected daily, children could be affected 190 times – the number of days in a school year (Long, 2016).
Decision 6 (sources of population entering, exiting and living in the focus areas)
Removed content was here…
My hypothesis and ideas about the relationship of geographic features to emissions and populations affected
I have some local knowledge of the five focus areas, as at some point in my life I have either, lived, worked or studied within approximately 100-500 meters of them, and they all to some extent form a street canyon (preventing dispersion of emissions). Because I am only counting affected populations if they are 100m or less from the emissions source (roads) and all significant green spaces are further away than 100m, they will not factor in estimations. I say more about proximity to green spaces in the conclusion.
# Removed content was here…
Note: All maps also show features which can reduce vehicle emissions such as cycle paths and car recharging points.
This map is a combination of a reference map, with Choropleth mapping to show AADT daily traffic flows on the roads. Traffic flows are classified using standard deviation
Detail of (reference) map: showing population input features for Kilburn High road and Swiss Cottage
Much of the process of analysis has already be described in the introduction, so the following descriptions build on top of these.
# Removed content was here…
5. Conclusion – discussion on the analysis output and possible limitations arising from the methodology used and the data provided.
I think as a range, “848,814 – 1,840,279” as an estimation of people affected, is probably reasonably accurate. However more data (collection) and research is needed to accurately measure and map the populations affected. This could provide:
- More information on the long-term health impact of air quality
- More up-to-date data (i.e. data used old: census was 2011 and LAEI data is 2013)
- Cyclist, bus, coach and pedestrian exposure geo-data (e.g. how many cycle for more than 10minues a day in a focus area?)
- Construct more accurate, and less arbitrary (as I have used) methods of weighting data
The figure below shows the more detailed breakdown of methodology and calculations, than presented above:
# Removed content was here…
The range, “848,814 – 1,840,279”, does not distinguish between a person who is exposed for two minutes in a year, and a child who walks for 30 minutes to and from school 190 days a year on Finchley road, receiving 5,700 minutes exposure. For this reason, future estimations should use a methodology where population affected is ranked according to exposure time and maybe concentrations. Using school enrolment and catchment area data could be explored with a view to advising parents on routes to school.
It can be argued that the proximity of the focus areas to the greenspaces is not close enough to mitigate the effects of air quality pollutants on populations because many live, work, wait for buses in the focus areas, or walk through the middle of them on a daily basis. You could argue that there is a deficiency of green spaces, but this would not in itself improve the air quality unless it reduced the source and quantity of the emitted pollutants. This could be achieved by pedestrianizing (routes in) the focus areas, greening the focus areas themselves, and there by partially increasing green space. (Interesting article https://blogs.esri.com/esri/arcgis/2011/06/30/park-analysis-and-design/ )
For thoughts on how to reduce human exposure to pollutants and the concentration of pollutants please see appendix item 3.
(Elliot.Treharne@london.gov.uk), G. L. A., 2013. Air Quality Focus Areas. [Online]
Available at: https://data.london.gov.uk/dataset/air-quality-focus-areas
[Accessed 27 January 2016].
Adams, B., 2015. What is a safe distance to live or work near high auto emission roads?. [Online]
Available at: http://sandiego.urbdezine.com/2015/05/28/what-is-a-safe-distance-to-live-or-work-near-high-auto-emission-roads/
[Accessed 3 1 2017].
Greater London Authority (GLA): LONDON ATMOSPHERIC EMISSIONS INVENTORY (LAEI) 2013, 2016. London Atmospheric Emissions Inventory 2013 : Workshop_DataStructure – Description of the data structure in the LAEI2013. [Online]
Available at: https://data.london.gov.uk/dataset/london-atmospheric-emissions-inventory-2013
[Accessed 27 December 2016].
Long, R., 2016. The School Day and Year – Commons Briefing papers SN07148. [Online]
Available at: http://researchbriefings.parliament.uk/ResearchBriefing/Summary/SN07148
[Accessed 6 1 2017].
Telegraph Media Group Limited: Sarmad Jawad, a. A. K., 2015. Mapped: Where is air pollution killing the most people?. [Online]
Available at: http://www.telegraph.co.uk/news/earth/environment/11991350/Mapped-Where-is-air-pollution-killing-the-most-people.html
[Accessed 27 December 2016].
Some effort went into converting ITN Urban Paths data (GML format) to an ArcGIS compatible format. Eventually after looking at various conversion tools, this data, including cycle paths, was converted using Safe Software’s: Feature Manipulation Engine (FME). However, it was only to find to my surprise that there are no cycle paths intersecting with focus areas.
2) Pedestrian exposure to poor air quality and presenting proximity of schools to emissions carriageways
I would have liked to include data quantifying pedestrians walking from homes outside the focus areas, into the focus areas and back, particularly children of school age, as they are at the highest risk to health due to their height and stage of development etc. What I have done is use “Important Buildings” data (see list) and symbology to clearly demarcate the proximity of schools and hospitals etc. to the focus zones. See figure below:
- Have real-time information in road signs, like on motorways, but showing concentrations of pollutants in the air and for the benefit of pedestrians, cyclists and motorists.
- Greater political and voter effort required to speed up Public road transport (Buses) and Taxis/Cabs becoming totally low emission (electric) fleets of vehicles
- More Electric recharge points outside congestion zone. i.e. the density of charging points north of Marylebone road lowers dramatically
- More support, regulation and pressure on industry (vehicle & electric battery manufactures) to produce vehicles of viable cost
- Reduce public transport congestion by promoting flexitime e.g. later & earlier starts to working day
- Continue enhancing the cycle lane network. To this end, re-engineering existing canal paths, railway and underground paths to include covered cycle paths. For example, the physical space used to bring intercity trains into the heart of London, could be more efficiently used by re-engineering to including cycle paths or tunnels (in both directions) to run alongside the train tracks and enable cyclists from the suburbs to commute into town on bicycle. Perhaps straight into Euston or Kings Cross for example. This would reduce the load on Tube and bus networks
 Where, if a full chemical analysis of the body before and after exposure were performed, they might for example, find an increase of of PM10 deposited in the respiratory tract or NO2…
 Annual average daily traffic: “Traditionally, it is the total volume of vehicle traffic of a highway or road for a year divided by 365 days. AADT is a useful and simple measurement of how busy the road is. Newer advances from traffic data providers are now providing AADT by side of the road, by day of week and by time of day.” https://en.wikipedia.org/wiki/Annual_average_daily_traffic
 For example, walking from Euston to University College hospital through focus area 31.
My colleague at LSBM Karen Stepanyan, and I spent the week (April 25-29) at LAK 16 in Edinburgh. Also taking in the ending Key note for Learning@Scale. Here I record a few notes under three headers, and re-present slides that resonated with me. As part of the week I participated in these workshops:
- Learning Analytics Across Physical and Digital Spaces – Cross-LAK (proceedings)
- LAL Workshop: Learning Analytics for Learners – LAL (proceedings)
Analytics on a live performance (presenting slides in this case)
Somebody at a music college gave me some feedback on their institutions progress with the Canvas LMS. I asked them what might be of interest at LAK16, and they responded: the analytics of performance. Being a musician of sorts myself this was interesting. This paper from Cross-LAK was the closest I encountered to the analytics of a performance: Towards a distributed framework to analyze multimodal data
Vanessa Echeverria, Federico Domínguez and Katherine Chiluiza
Abstract: Data synchronization gathered from multiple sensors and its corresponding reliable data analysis has become a difficult challenge for scalable multimodal learning systems. To tackle this particular issue, we developed a distributed framework to decouple the capture task from the analysis task through nodes across a publish/subscription server…
What was most of interest to me: To see a system with automatic analysis and reporting on the quality of a persons “presentation” or “performance” – based on analysis of audio and video streams and textural analysis of presentation slides.
Metacognition in learning and Experience Sampling Apps @Scale
These notes are not about one paper, but about what a number of presentations/papers made me wonder. In short: why am I not aware of an Apps/projects working at scale (~thousands of schools), to help learners, and gather data related to meta-cognition (and maybe the learners emotions as well) – perhaps using an experience sampling method (ESM)? After all, governments around the world spend a considerable amount of money on educating young people, and it is often less effective than hoped. This seems like a great albeit difficult project to increase the usefulness of Learning Analytics.
This was a short conversation on twitter related to the thought above. Prompted in part by discussions at the LAL workshop, this tweet and the App described by Carrie Demmans in her short paper “English Language Learner Experiences of Formal and Informal Learning Environments” … Keywords: ubiquitous learning, experience sampling methodology, informal learning, formal learning, language learning, English language learner, communication, affect, analytics. Youtube video of presentation
Also I have always thought one of the values of being introduced to the concept of ‘Preferred Learning styles’ was it gets us thinking about how we learn, and the most efficient ways of learning for our selves – meta-cognition. However “There presently is no empirical justification for tailoring instruction to student’s supposedly different learning styles”. This is a slide from Prof. dr. Paul A. Kirschner’s keynote lecture
…So this is what got me thinking about Metacognition in learning, and developing Experience Sampling Apps @Scale
New Forms of Assessment
A Dispatch from the Psychometric Front by Professor Robert J. Mislevy. This was another interesting lecture…Examples applications using new forms of assessment discussed are shown in this slide below: SimCityEDU; Packet Tracer – Cisco Networking Academy….
Doug Clow’s Live blogs: https://dougclow.org/lak16-liveblogs/