Beijing Air Quality : Statistical analysis using R
▻http://www.garethkennedy.net/AQIStats.html
Beijing is known for being one of the most polluted cities on Earth. Often the government clamps down on private vehicles or barbecues in an attempt to control the air quality. However the effectiveness of these measures depends on the source of the pollution, in particular of particles smaller than 2.5 microns. These particles are the major cause of pollution related health issues in humans (e.g. here) and are mostly produced by combustion processes.
The concentration of these small particles is measured by the PM 2.5 index which will be examined here using hourly data from 2008 as measured by the US Embassy in Beijing (available here). To determine the origin of the pollution and examine how the city is cleaned the weather data (available here) for the same time period is examined. The statistical analysis is conducted using a self-written R code which is available here.
Since both the slope and coefficient of determination are useful in examining the effect of the wind on the PM 2.5 pollution level then both of these were combined into a single plot (right). The colour being set by the slope with green being maximum (negative) slope and red being close to zero slope (so no effect of increasing wind speed on decreasing pollution). The transparency (visible as how pale each colour is) shows the coefficient of determination, with wind directions with no correlation to pollution appearing as close to white.
From this figure wind from the NNE direction is the most efficient at cleaning Beijing of PM 2.5, but any wind from the W to the NE direction can also clean the city. The strongest individual cleaning events are from the NW where the wind is the strongest (see above).