Another hextri example

The data are 8 years of daily air pollution measurements from Seattle.

seattle<-read.csv("~/Documents/TEACHING/517/seattlepm.csv")
head(seattle)

##    date mo yr dow       pm1 temp stagno admyng gasyng
## 1 10000  5 87   3 0.2642518   52    7.0      2      0
## 2 10001  5 87   4 0.4894978   55    7.5      1      1
## 3 10002  5 87   5 0.3777940   59   10.5      1      1
## 4 10003  5 87   6 0.5317807   55    8.0      6      1
## 5 10004  5 87   7 0.4188140   58    7.5      4      2
## 6 10005  5 87   1 0.3956474   58   11.0      3      1

• mo is month
• dow is day of the week
• temp is the mean temperature in degrees Fahreneheit
• stagno is the number of hours in the day with zero or low wind speed
• pm1 is a measure of light scattering by the air due to fine particle air pollution

We can turn the pm1 variable into the standard micrograms per cubic meter units, and also recode into groups

seattle$pm1mass <- 21*seattle$pm1
seattle$pmgp <- cut(seattle$pm1mass, c(0,6,12,18,Inf))

And now some plots

coplot(pm1~temp|stagno,data=seattle)

## 
##  Missing rows: 172, 173, 174, 175, 489, 490, 691, 692, 693, 713, 716, 718, 719, 720, 721, 722, 723, 1624, 1625

coplot(pm1~stagno|temp, data=seattle)

## 
##  Missing rows: 172, 173, 174, 175, 489, 490, 691, 692, 693, 713, 716, 718, 719, 720, 721, 722, 723, 1624, 1625

And from another viewpoint

library(hextri)
hextri(stagno~temp,data=seattle, class=pmgp, colour=c("blue","darkgrey","orange","sienna"), style="size",nbins=25, diffuse=TRUE, sort=TRUE, ylab="Air stagnation (hrs/day)",xlab="Temperature (F)")
legend("topleft",col=rev(c("blue","darkgrey","orange","sienna")),pch=19, legend=rev(c("very low","low","moderate","high")),bty="n")

Fine particle pollution is higher with stagnant air (obviously) and at lower temperatures (because wood fireplaces). Air stagnation tends to be higher when it’s cold – Seattle gets some winter temperature inversions.