Air Quality in Southampton (UK)

Exploring the data (using AURN & openair)

1 About

1.1 Contributions

Please note that authorship is alphabetical. Contributions are listed below - see github for details and who to blame for what :-).

• Ben Anderson (b.anderson@soton.ac.uk @dataknut)

1.2 Code

• https://github.com/CfSOtago/airQual/commits/master

1.3 Citation

If you wish to refer to any of the material from this report please cite as:

• Anderson, B., (2019) Air Quality in Southampton (UK): Exploring the data (using AURN & openair) , University of Southampton: Southampton, UK.

Report circulation:

• Public

Report purpose:

• to explore (official) Southampton City Council Air Quality data (http://southampton.my-air.uk) sourced from
• to test the openair (Carslaw and Ropkins 2012) package

This work is (c) 2019 the University of Southampton.

2 Introduction

dfW <- openair::importAURN(
  site = "SA33",
  year = 2019,
  pollutant = "all",
  hc = FALSE,
  meta = TRUE,
  to_narrow = FALSE, # produces long form data yay!
  verbose = TRUE # for now
)

# fails. it worked before
# dfL <- openair::importAURN(
#   site = "SA33",
#   year = 2019,
#   pollutant = "all",
#   hc = FALSE,
#   meta = TRUE,
#   to_narrow = TRUE, # produces long form data yay!
#   verbose = TRUE
# )

dtW <- data.table::as.data.table(dfW) # we like data.tables

Data downloaded from http://uk-air.defra.gov.uk/openair/R_data/ using ōpenair::importAURN().

Southampton City Council collects various forms of air quality data at the sites shown in 2.1. WHO publishes information on the health consequences and “acceptable” exposure levels for each of these.

lDT <- data.table::melt(dtW, id.vars = c("site", "date", "code", "latitude", "longitude", 
    "site_type"), measure.vars = c("no", "no2", "nox", "pm10", "nv10", "v10", "ws", 
    "wd"), value.name = "value"  # varies 
)

# remove NA
lDT <- lDT[!is.na(value)]

t <- lDT[, .(from = min(date), to = max(date), nObs = .N), keyby = .(site, variable)]

kableExtra::kable(t, caption = "Dates data available by site and measure", digits = 2) %>% 
    kable_styling()

Table 2.1: Dates data available by site and measure

site	variable	from	to	nObs
Southampton A33	no	2019-01-01	2019-12-20 23:00:00	8211
Southampton A33	no2	2019-01-01	2019-12-20 23:00:00	8211
Southampton A33	nox	2019-01-01	2019-12-20 23:00:00	8211
Southampton A33	pm10	2019-01-01	2019-12-20 22:00:00	7724
Southampton A33	nv10	2019-01-01	2019-12-20 22:00:00	7724
Southampton A33	v10	2019-01-01	2019-12-20 22:00:00	7724
Southampton A33	ws	2019-01-01	2019-12-20 23:00:00	8064
Southampton A33	wd	2019-01-01	2019-12-20 23:00:00	8064

3 Summarise data

Summarise previously downloaded and processed data… Note that this may not be completely up to date.

skimr::skim(dfW)

Table 3.1: Data summary

Name	dfW
Number of rows	8760
Number of columns	14
_______________________
Column type frequency:
character	3
numeric	10
POSIXct	1
________________________
Group variables	None

Variable type: character

skim_variable	n_missing	complete_rate	min	max	empty	n_unique	whitespace
code	0	1	4	4	0	1	0
site	0	1	15	15	0	1	0
site_type	0	1	13	13	0	1	0

Variable type: numeric

skim_variable	n_missing	complete_rate	mean	sd	p0	p25	p50	p75	p100	hist
no	549	0.94	27.95	38.75	-0.22	4.12	14.20	35.60	444.91	▇▁▁▁▁
no2	549	0.94	32.74	22.47	0.00	15.11	28.03	45.55	158.31	▇▅▁▁▁
nox	549	0.94	75.57	76.58	0.84	22.77	52.03	100.15	771.31	▇▁▁▁▁
pm10	1036	0.88	16.74	10.98	-1.40	9.60	13.70	20.50	95.00	▇▃▁▁▁
nv10	1036	0.88	13.66	9.48	-9.60	7.40	11.30	17.42	99.70	▇▆▁▁▁
v10	1036	0.88	3.07	3.09	-14.00	1.30	2.70	4.30	22.80	▁▂▇▁▁
ws	696	0.92	3.76	2.03	0.00	2.30	3.30	4.90	13.00	▆▇▃▁▁
wd	696	0.92	201.12	104.40	0.00	116.77	229.75	284.10	360.00	▅▃▃▇▆
latitude	0	1.00	50.92	0.00	50.92	50.92	50.92	50.92	50.92	▁▁▇▁▁
longitude	0	1.00	-1.46	0.00	-1.46	-1.46	-1.46	-1.46	-1.46	▁▁▇▁▁

Variable type: POSIXct

skim_variable	n_missing	complete_rate	min	max	median	n_unique
date	0	1	2019-01-01	2019-12-31 23:00:00	2019-07-02 11:30:00	8760

Table 3.1 gives an indication of the availability of the different measures.

4 Analysis

In this section we present graphical analysis of the previoulsy downloaded data. Note this is just a snapshot of the data available.

4.1 Nitrogen Dioxide

yLab <- "Nitrogen Dioxide (ug/m3)"

t <- lDT[variable == "no2", .(mean = mean(value, na.rm = TRUE), sd = sd(value, na.rm = TRUE), 
    min = min(value, na.rm = TRUE), max = max(value, na.rm = TRUE)), keyby = .(site)]
kableExtra::kable(t, caption = "Summary of no2 data") %>% kable_styling()

Table 4.1: Summary of no2 data

site	mean	sd	min	max
Southampton A33	32.73677	22.47269	0	158.3072

Table 4.1 suggests that there may be a few (0) negative values. These are summarised in 4.2 while Figure 4.1 shows the availability and levels of the pollutant data over time.

t <- head(lDT[variable == "no2" & value < 0], 10)
kableExtra::kable(t, caption = "Negative no2 values (up to first 6)") %>% kable_styling()

Table 4.2: Negative no2 values (up to first 6)

site	date	code	latitude	longitude	site_type	variable	value

t <- table(lDT[variable == "no2" & value < 0, .(site)])
kableExtra::kable(t, caption = "Negative no2 values (count by site)") %>% kable_styling()

Table 4.2: Negative no2 values (count by site)

Freq

# dt,xvar, yvar,fillVar, yLab
p <- makeTilePlot(lDT[variable == "no2"], xVar = "date", yVar = "site", fillVar = "value", 
    yLab = yLab)

p

Figure 4.1: Nitrogen Dioxide data availability and levels over time

# p <- ggplot2::ggplot(dt, aes(x = obsDateTime, y = nox2, colour = site, alpha =
# 0.1)) + geom_point(shape=4, size = 1)

t <- lDT[variable == "no2" & value > 200][order(-value)]

kableExtra::kable(caption = paste0("Values greater than WHO threshold (NO2 > ", hourlyno2Threshold_WHO, 
    ")"), head(t, 10)) %>% kable_styling()

Table 4.3: Values greater than WHO threshold (NO2 > 200)

site	date	code	latitude	longitude	site_type	variable	value

p <- makeDotPlot(lDT[variable == "no2"], xVar = "date", yVar = "value", byVar = "site", 
    yLab = yLab)

p <- p + geom_hline(yintercept = hourlyno2Threshold_WHO) + labs(caption = "Reference line = WHO hourly guideline threshold")


if (doPlotly) {
    p
    plotly::ggplotly(p + xlim(xlimMinDateTime, xlimMaxDateTime))  # interactive, xlimited 
} else {
    p
}

Figure 4.2: Nitrogen Dioxide levels, Southampton (hourly)

Figure 4.2 shows hourly values for all sites. In the study period there were 0 hours when the hourly Nitrogen Dioxide level breached WHO guidelines. The worst 10 cases are shown in Table 4.3.

lDT[, obsDate := lubridate::date(date)]

plotDT <- lDT[variable == "no2", .(mean = mean(value, na.rm = TRUE)),
             keyby = .(obsDate, site)]

p <- makeDotPlot(plotDT, 
                 xVar = "obsDate", 
                 yVar = "mean", 
                 byVar = "site", 
                 yLab = yLab)

p <- p +
  geom_smooth() + # add smoothed line
  labs(caption = "Trend line = Generalized additive model (gam) with integrated smoothness estimation")

if(doPlotly){
  p
  plotly::ggplotly(p + xlim(xlimMinDate, xlimMaxDate)) # interactive, xlimited # interactive
} else {
  p
}

Figure 4.3: Nitrogen Dioxide levels, Southampton (daily mean

Figure 4.3 shows daily mean values for all sites over time and includes smoother trend lines for each site.

Clearly the mean daily values show less variance (and less extremes) than the hourly data and there has also been a decreasing trend over time.

4.2 openair tests

Wind rose

openair::windRose(dfW)

Figure 4.4: Wind rose for Southampton A33, 2019

Pollution rose

openair::pollutionRose(dfW, pollutant = "no2")

Figure 4.5: Pollution rose for Southampton A33, 2019, hourly data

We get a slightly higher % of high measures when the wind is from the SE?

5 Runtime

Report generated using knitr in RStudio with R version 3.6.2 (2019-12-12) running on x86_64-apple-darwin15.6.0 (Darwin Kernel Version 17.7.0: Sun Dec 1 19:19:56 PST 2019; root:xnu-4570.71.63~1/RELEASE_X86_64).

t <- proc.time() - startTime

elapsed <- t[[3]]

Analysis completed in 7.88 seconds ( 0.13 minutes).

R packages used:

• data.table - (Dowle et al. 2015)
• ggplot2 - (Wickham 2009)
• here - (Müller 2017)
• kableExtra - (Zhu 2018)
• lubridate - (Grolemund and Wickham 2011)
• openair - (Carslaw and Ropkins 2012)
• plotly - (Sievert et al. 2016)
• skimr - (Arino de la Rubia et al. 2017)

References

Arino de la Rubia, Eduardo, Hao Zhu, Shannon Ellis, Elin Waring, and Michael Quinn. 2017. Skimr: Skimr. https://github.com/ropenscilabs/skimr.

Carslaw, David C., and Karl Ropkins. 2012. “Openair — an R Package for Air Quality Data Analysis.” Environmental Modelling & Software 27–28 (0): 52–61. https://doi.org/10.1016/j.envsoft.2011.09.008.

Dowle, M, A Srinivasan, T Short, S Lianoglou with contributions from R Saporta, and E Antonyan. 2015. Data.table: Extension of Data.frame. https://CRAN.R-project.org/package=data.table.

Grolemund, Garrett, and Hadley Wickham. 2011. “Dates and Times Made Easy with lubridate.” Journal of Statistical Software 40 (3): 1–25. http://www.jstatsoft.org/v40/i03/.

Müller, Kirill. 2017. Here: A Simpler Way to Find Your Files. https://CRAN.R-project.org/package=here.

Sievert, Carson, Chris Parmer, Toby Hocking, Scott Chamberlain, Karthik Ram, Marianne Corvellec, and Pedro Despouy. 2016. Plotly: Create Interactive Web Graphics via ’Plotly.js’. https://CRAN.R-project.org/package=plotly.

Wickham, Hadley. 2009. Ggplot2: Elegant Graphics for Data Analysis. Springer-Verlag New York. http://ggplot2.org.

Zhu, Hao. 2018. KableExtra: Construct Complex Table with ’Kable’ and Pipe Syntax. https://CRAN.R-project.org/package=kableExtra.