Why are some states so much colder than others?

Recently I saw that the weather in my home state of Minnesota was colder than the average surface temperature on Mars (as measured by the curiosity rover).
Static choropleth chart that shows the average 2013 temperature in the U.S. by state.
New year is here and this brings a great opportunity to measure things such as weather. I used a combination of Python and JavaScript to create this image of average temperature by state. In the graph above, dark blue represents the coldest temperatures while dark red represents the warmest. It looks like both North Dakota and Maine are worse off than Minnesota. Florida and Texas are where you would want to live for consistently warm weather. Often in data science, it's more important to use the correct graph than it is to get highly involved with reporting statistics. Instead of showing a chart by state this post could instead have shown that Florida and Texas were significantly warmer (which they are) than North Dakota and Maine. However, this number intensive approach ignores the human element and makes data harder to interpret.


What's more is that it is easy to use this approach to communicate complex ideas such as the average temperature over time. Take a look at this moving picture showing the average temperature by state for each month of the year:

Moving choropleth chart that shows the 2013 monthly average temperature in the U.S. by state.

This image could have been a line graph showing average temperature by month for each state (or more likely for each region). But seeing how all of these states change in unison together through the use of color takes advantage of the human ability to visually interpret moving colors. Notice how clearly this graph illustrates that the coldest temperatures occur in Jan-Feb and the warmest occur in Jul-Aug.


The data used in this analysis is provided by the diligent scientists at the National Weather Service. They have small solar powered stations throughout the country that provide data such as temperature, humidity, wind speed and direction, air pressure, and precipitation. This complicated set of telemetry devices then stores these data along with date, time, and location into a central repository. This approach gives people interested in weather advanced analytical capabilities. A few weeks ago I ran across one such curious looking machine, take a look:
A lonely weather station located amid a turbulent sky

Finally, since the Sun causes warm weather and the Sun is strongest in June and weakest in December, why would months such as July and August be warmer than June and why would February be colder than December? The answer is that like many other complex systems, there is some latency between cause and effect. For various reasons, it takes times for the effects of the Sun to translate into changes in weather. To illustrate this, take a look at this image showing the monthly percent change in temperature:
Moving choropleth chart that shows the 2013 monthly average change in temperature in the U.S. by state.

Red represents an increase and blue a decrease. Notice the power of this visualization to clearly explain why it is still warm in August. The change in temperature is still increasing and does not begin to decrease until October. Also notice strong regional variation. The Midwest seems to experience the most dramatic temperature changes while the Coastal regions experience hardly any change. The coast benefit from the Maritime effect that keeps these areas warmer in Winter and cooler in Summer. The Midwest experiences the Continental drift where cold weather from Canada or warm weather from the Gulf explain temperature variations outside of the normal seasonal changes.


Things are starting to heat up! And I'm not just referring to the weather. What's next?

Tags: