The UK has a lot of air quality measurement sites! We use these to track national compliance with AQ objectives, as well as to inform more local-scale policies.
Facilitate internal & external training to allow others to better understand their own data.
Dynamic Reporting
We create a lot of dynamic reports using rmarkdown (although we’re slowly transitioning to Quarto!).
Air quality data lends itself to automatic reporting as it…
…is always in a consistent format (date, pollutant concentrations, met data).
…is added to our databases continuously after ratification.
We also create ad-hoc reports in response to pollution “episodes” (particulate matter episodes on bonfire night, ozone episodes during heatwaves, etc.).
Dynamic reports are an excellent middle ground between traditional reports and data dashboards, providing a strong narrative and extensibility.
The {openair} package
Introduction
openair is an R package developed for the purpose of analysing air quality data — or more generally atmospheric composition data. The package is extensively used in academia, the public and private sectors.
It has been around for a long time — over a decade! This means some things are a bit odd for newer R users (e.g., use of lattice over ggplot2!). Key question - how can we best leverage it in this rapidly developing world of R programming?
Figure 1: A ‘calendar plot’ - a heatmap in a familiar calendar format.
Figure 2: A ‘Theil Sen trend’ - a robust trend fit to a timeseries.
The {openair} package
Data Access
Even if you don’t really care about air quality, openair (and worldmet) can still be useful to you as they provide access to lots of timeseries data!
The purpose of openairmaps was to combine openair’s directional analysis with the leaflet package (& javascript library) to create interactive air quality maps that help us triangulate pollution sources and understand local air quality in its geographical context.
Once I got started, I was hooked! Every time you think you’re done, you’re dragged back in to expand on the functionality further.
Being in the position to teach the package also allows a developer to understand what users are actually interested in.
Case Studies
Trends Over Lockdown
Once nice use of the easy layer control options is to view changes over time. For example, we can clearly see the huge impact of the COVID-19 lockdown in Oxford City Centre!
oxford_road<-importAURN(site ="OX", year =2018:2022, meta =TRUE)polarMap(oxford_road, "nox", # data, pollutant control ="year", # create control menu for the year upper =20, limits =c(0, 250), # set limits so markers share them d.icon =300, alpha =3/4)# semi-transparency
Case Studies
Source Apportionment
If we have a whole network of sites, we can triangulate potential sources. For example, where are likely sources in a nearby industrial site?
library(dplyr)# get datanlincs_local<-importLocal(c("SCN6", "SC12", "SC10", "AMVL"), year =2021, meta =TRUE)nlincs_aurn<-importAURN(c("SCN2"), year =2021, meta =TRUE)# combinenlincs_all<-dplyr::bind_rows(nlincs_local, nlincs_aurn)# reuse modelled met for local datanlincs_all<-nlincs_all%>%select(-ws, -wd, -air_temp)%>%left_join(select(nlincs_aurn, date, ws:air_temp), by =join_by(date))# polar plot map!polarMap(nlincs_all, "pm10", alpha =3/4,# use multiple providers provider =c("OpenStreetMap", "Esri.WorldImagery"))
This is a surface-level analysis of a complex air quality situation - please do not draw any lasting conclusions about Scunthorpe from this map!
Case Studies
24 Hours in Central London
Using different plots as markers can reveal even more interesting features - for example, the polarAnnulus() plot can show how pollution typically varies over a 24 hour period.
london<-importAURN(site =c('CLL2', 'MY1', 'KC1'), year =2019, meta =TRUE)annulusMap(london, d.icon =300, c("nox", "pm10", "so2"), # multiple pollutants! provider ="CartoDB.Positron", # just to demo - a different provider cols ="viridis")# and a different colour palette
How does this work “under the hood”?
The {leaflet} package
The leaflet package makes it really easy to construct interactive web maps. Look closely around the web and you’ll find leaflet maps everywhere. An example of a map is shown below.
library(leaflet)oxford<-data.frame( lat =c(51.754474, 51.758900), lng =c(-1.260699, -1.259626), label =c("White Rabbit Pizza Restaurant", "Department of Statistics"), popup =c("Where all the tasty pizza comes from!", "The home of OxfordR!"))leaflet(oxford)%>%addTiles()%>%addMarkers(lat =~lat, lng =~lng, label =~label, popup =~popup)
