News and Notes

It all started a few weeks ago, an IEM user was wondering what the heat index record was for Des Moines. As with about everything, I responded with that I have an autoplot for that. Namely autoplot #153 that produced this image.

The user was curious though as to why some of the hottest weather in Des Moines history back in the 1930s was not showing up on the plot. Including a rather gross reading of air temperature 111 and dew point of 70 on 25 July 1936. I responded with "Oh dear" and down the rabbit hole I went.

The heat index is not an observed variable, but something computed based on available raw atmospheric data like temperature and humidity. There are at least 21 equations used and each have various assumptions. [As an aside, the community seems to be coalescing around Wet Bulb Globe Temperature as a better measure of sultriness than heat index.] The NWS equation was what I thought my code was using to compute a heat index. After some investigation with the MetPy (python meteorological) library, it was found to not compute a heat index when the relative humidity was below 40%. So that gross reading of 111 over 70 evaluated to an undefined heat index and thus not included in the plot.

I submitted a code patch to the MetPy library, which will be available in an upcoming release (version 0.11). The IEM is now using this updated heat index equation and has repaired the archive of stored heat index values. You can download the computed hourly values or daily summaries.

And so with the correction, we end up with this:

And the answer to the original question is a sweltering 122F! The fun never stops while attempting to curate all this data and my apologies if you need to re-ingest the updated information into your archives.