Temperature Shenanigans

A week or two ago we were treated to the news that July 2021 was the hottest month on record. Here’s how the BBC had it, as reported by Paul Homewood here:

“July was the world’s hottest month ever recorded, a US federal scientific and regulatory agency [NOAA] has reported.

The data shows that the combined land and ocean-surface temperature was 0.93C (1.68F) above the 20th Century average of 15.8C (60.4F).”

Paul then went on to debunk the story by showing that the month of July 2021 clearly did not have the highest anomaly of any of the temperature series. Here is NOAA’s own graph of temperature anomalies as presented at Notalot, with this July arrowed:

It seemed pretty clear that July 2021 was nowhere near the hottest month on record. I was left scratching my head about this, until it was explained at WUWT. July is the hottest month every year in terms of absolute temperatures rather than anomalies, so a smaller anomaly for July can still result in an overall warmer measure than a higher anomaly in say January. (You might ask why, in that case, we are using anomalies at all; I have no answer to that.)

However, there is more to this story even when NOAA seem to have been proved right in a narrow sense once seasonal signals have been removed from the data. The new anomaly is not just lower than some past months. It’s also lower than at least one past July. That means it’s not a record, right? Was this announcement a straight lie from a US national agency?

Here’s what NOAA said of July 2021:

Note the anomaly: +1.67 F (BBC said +1.68 F, but we’ll let that slide).

Now, here’s what NOAA said in 2019:

Note the anomaly: +1.71 F, which is considered by most people to be higher than +1.67 F.

But (referring to the 2021 statement above) 2019’s record-setting anomaly has since been trimmed from +1.71 F to +1.65 F, making this year’s +1.67 F the new world record. Hooray! We’ve got there. Someone better phone 2019 and ask it to return its gold medal: it’s now in a three-way tie for silver.

Note that I do not imply any nefariousness on the part of NOAA’s data crunchers here. It just isn’t a good look if a record can be silently cancelled, allowing a subsequent lower score to be later announced as yet another record.

Acknowledgement: I first saw the anomalies for 2021 and 2019 compared at Retraction Watch here. I can claim no credit for spotting the difference.