Recently I tried updating this map using the current data file ("climdiv-tmpccy-v1.0.0-20220804.txt") and it didn't work. I found that NOAA is now including data from Alaska in the master file, whereas previously the Alaskan data was in a separate file. I reworked the notebook to account for this change. You will still have to change the statement beginning cdata = SemanticImport[ to reference the directory where you have placed the file of historic temperatures.

The updated notebook, and the map, are attached.

Attachments:

USClimateChange.nb

Hare's the link to the two follow on projects that I did with my kids. The first was my 7th graders idea to look at percent change rather than absolute change. The 2nd was my 9th graders idea to see how we could make predictions (which I modified to be a project about linear and quadratic approximations to the data).

https://mikesmathpage.wordpress.com/2019/06/23/follow-up-2-to-john-shonders-us-weather-data-visulaization/

Thanks so much, John, for sharing your work. Using it with my kids has been really fun. I'm really happy that this sort of data analysis can be made accessible to middle and high school kids.

Very interesting Doug! In later versions of this map I too decided to use shades of blue for negative changes. It makes sense visually. You also see that the starting point makes a difference too. The map at the top of this page shows temperature changes between 1991 and 2012 compared to the average temperature between 1901 and 1960.

Mike, how nice to know you were able to use the code and data to teach your children. That's why I posted it: hoping people would extend the analysis, try out different smoothing techniques, use different starting and ending points, etc. I watched your videos and it sounds like your boys have some good ideas. I'll be interested to see what they come up with. Happy Father's Day.

John -

I used your post for a project with my kids this morning. Thanks for sharing this amazing work - they loved it and had a few ideas that they'd like to explore a bit more. Can't wait to work on these ideas with them. Here's what we did:

https://mikesmathpage.wordpress.com/2019/06/16/using-john-shonders-amazing-us-temperature-visualization-with-kids/

The question I thought someone was going to ask is where did they get temperature data for counties that didn’t even exist at the time. Arizona for example didn’t become a state until 1912 and yet there are the average monthly temperatures for each of its current counties back to 1895. The answer is through kriging.

Thank you. Must be my error in calculating the FIPS.

-Ed

John, I may be mistaken in decoding the file, but I believe the file shows about 17 county entries for Michigan and it has 83. For Arizona, the state code is 04 and has 15 counties. The file shows 58 counties being reported for 04. FYI. Thanks for the information as well as your analysis. Very Interesting.

Excellent analysis! A quick question, what was the standard deviation in temperature measurements? I am asking that because some Delta T are small and I wonder to what extend we can interpret them as a meaningful shift in T?

My mistake. I originally used a moving average of 10 years for the smoothing, and then changed it to an exponential average. The values in the legend correspond to the moving average case. Looking at both figures, there really is not much of a difference. This is a problem when using a bespoke legend as I did.

I'm sure it would be possible to generate the legend automatically with the correct limits -- and perhaps someone will be interested enough to show us how to do that -- but for now I went back and edited my original post to correct everything. The figure and the file now use the moving average.

Try https://www1.ncdc.noaa.gov/pub/data/cirs/climdiv/