Well, here is a graph I found at luftkvalitet.info, and it's a report over the air quality in Bergen. There are two automatic air-samplers in Bergen, one at Danmarksplass and one by the old town hall. The NO2 level and concentration of particles in two different sizes are measured. Each of the components is classified into four different categories, depending on the concentration (as shown in the chart below). The (total) air quality is the reported as one of the categories, depending on the most polluted of the three components. The chart below is from month of November 2009.
Some interesting data can be extracted from this graph. Like that November 2009 where approximately 30 hours longer at Rådhuset compared to Danmarksplass, but on average the years 2003 and 2009 where longer on Danmarksplass.
What I'm getting at here, is (again) that this isn't a very good graph. Personally, I think that they are missing a "no data" category or something, since November has a total of 720 hours, which none of the bars have. And, as I understand the data, the total of all categories isn't interesting, since they are all the same (only depending on how many days in the month). In fact, the interesting data here is the hours with elevated level of air pollution. And due to the large size of the normal (green) bars, the graph does not do a good job in presenting these data. So, here is my suggestion of a redesigned graph, focusing on comparing the data with elevated level of pollution. I opted for a three panel vertical Trellis chart:
To even better compare the individual values, a 1x4 vertical Trellis chart could also been used, wiyh panels Rådhuset Nov 2009, Rådhuset average 2003-2009, Danmarksplass Nov 2009 and Danmarksplass average 2003-2009. Each panel with three horizontal bars with each of the different level of pollution.
Another option would be to include a bar for the total hours with elevated pollution (sum of some, much og severely polluted categories).