Atmospheric air pollution is one of the most important environmental concerns in many countries around the world. It is strongly affected by meteorological conditions. In this set of exercises, we will use the
openair package to work with and analyze air quality and meteorological data. This package provides tools to directly import data from air quality measurement networks across the UK, as well as tools to analyze and produce reports.
In the previous exercise sets, we used some functions in the
openair package, along with some geospatial packages to spatially analyze and visualize air quality data. However, sometimes it is difficult to customize or modify those functions according to our interest. So, it is also very useful to gain advantages of using default function in
R. In addition , so far, we mostly practice visualization functions and do not pay much attention to statistical approaches to analyze air quality data and also analyze the effect of meteorological parameters on them. Therefore, from the last three exercise sets, we started to analyse individual air pollutants using functions we had practiced before, along with using default functions available in
In this exercise, we will continue to analyze air pollutant data using some new functions and some of the functions we have already practiced in the previous exercises in more detail.
Answers to these exercises are available here.
For other parts of this exercise set, follow the tag openair.
For this exercise, we will need to import the following data from the
my1data <- importAURN(site = 'MY1', year = 2016, met = TRUE)
In the previous exercise, we investigated the potential sources of emission for PM concentration using polar plots. Now it would be useful to investigate how source characteristics can be analyzed in more
depth by considering their temporal variation. We first consider how concentrations of PM10 vary by the hour of the day and the day of the week using the
timeVariation function. This function is very useful for assessing different source influences, particularly when used with data filtering (ex. by wind speed and direction.) By default, the plots are shown in local time, which helps to give a clearer indication of variation by the hour of the day. In this exercise, we will first look at the results from the previous exercise and after identifying the direction and time of which PM10 and pm2.5 are elevated, use them as a filter to plot and discuss the temporal variation of PM10 and pm2.5.
It is useful to consider whether any other pollutants also vary in this way. Because NOx
is typically from mobile sources in urban areas, it would be interesting to know whether concentrations of PM10 vary in a similar way. If they do, this would provide a strong indication that the source of PM10 is also likely to be similar, ex. from mobile sources. Repeat the previous exercise using NOx concentration and try to compare the results from those of PM10.
Perhaps a better approach for comparing PM and NOx would be to plot them
together on the same plot. Unfortunately, the problem is that they each have very different
scales. One approach is to “normalize” the concentrations by dividing them by their mean
values. In this exercise, first normalize the concentration of PM10 and NOx and then plot their diurnal and weekly profile in the same plot.
In this exercise, we will try to investigate trends. This can be done using the
Theilsen function in the
openair, which calculates monthly means and fits a smooth line through them. This also provides a shaded area either side of the trend line, representing the 95 % confidence intervals in that trend. Use this function to plot the trend of PM10 and discuss the variation in trend.
In the previous exercise, we investigated the trend in PM10 concentration over different months. However, there could be effects of seasonal change on the variation in PM10, which makes it difficult to investigate the trends. Therefore, in order to remove such effects, there is an option in the
trend.plot function to deseasonalize the trend. Repeat the previous exercise using the deseasonalize option. Also, it is important to identify if there is any statistically significant change in the trend.