The featured plot is an attempt to illustrate the difference between July low temperatures during a thirty year period that includes the 1930s ("Dust Bowl") and the most recent thirty year period. The top panel presents the data as a kernel density estimate, which attempts to provide a smoothed histogram. The bottom panel differences the two plots. While the y-axis values are a bit obtuse, a larger value indicates a higher frequency. So the difference in distributions between these two periods is more readily apparent. This plot is for Des Moines, so its urban heat island effect is likely somewhat to blame for the increased temperature, but increased humidity levels are also likely to support the increase in frequency for values in the upper 60s and lower 70s. This autoplot can be run for other locations in the country and arbitrary periods of your choice, what changes can you discover?

Voting:
Good = 7
Bad = 0
Abstain = 1