I only know one guy who's written in depth about this, Tony Heller. Here's one example article but click through the titles at the top and there are others. At some point I downloaded the raw data tables from a NASA (or was it NOAA? I forget) FTP site and re-ran one of his analyses, which was valid, so that increased my confidence in the graphs he presents.
He shows the data has been heavily altered in various ways. Sometimes he shows it by finding and downloading "raw" data tables and graphing them over the top of the publicly released graphs. Sometimes by searching newspaper or web archives for old temperature graphs and showing that the historical data in modern graphs has been altered vs what was once being published.
Very occasionally other people notice this too. Retraction Watch is a blog that normally just writes about retracted scientific papers but here they noticed that July was reported as the hottest on record, but that the same government agency had previously reported another record breaking temperature that was higher than July's:
How is it possible for a temperature to "break records" when it is lower than previously reported temperatures? NOAA explains:
"NOAAGlobalTempv5 is a reconstructed dataset, meaning that the entire period of record is recalculated each month with new data. Based on those new calculations, the new historical data can bring about updates to previously reported values. These factors, together, mean that calculations from the past may be superseded by the most recent data and can affect the numbers reported in the monthly climate reports."
This is normal for climatology. When a new month's data is added they don't just append it to the prior data tables. They re-calculate every temperature in the dataset using the latest modelling software. Climatology, as a science, cannot actually tell us "how hot was it on this day 10 years ago" because there are many different answers.
This is already a terrible, terrible problem for their credibility. But the worst thing of all is that these adjustments are so large that they materially change the entire shape of the 20th century temperature history. Decades ago, global temperature graphs showed a long period of declines starting around WW2. Hence why global cooling was such a big topic in the 60s (incidentally, Heller's site shows ample archival evidence that this was indeed mainstream climatology back then). But if you look at modern temperature graphs, you cannot see any such decline. It's been erased from the historical record.
https://realclimatescience.com/alterations-to-the-us-tempera...
He shows the data has been heavily altered in various ways. Sometimes he shows it by finding and downloading "raw" data tables and graphing them over the top of the publicly released graphs. Sometimes by searching newspaper or web archives for old temperature graphs and showing that the historical data in modern graphs has been altered vs what was once being published.
Very occasionally other people notice this too. Retraction Watch is a blog that normally just writes about retracted scientific papers but here they noticed that July was reported as the hottest on record, but that the same government agency had previously reported another record breaking temperature that was higher than July's:
https://retractionwatch.com/2021/08/16/will-the-real-hottest...
How is it possible for a temperature to "break records" when it is lower than previously reported temperatures? NOAA explains:
"NOAAGlobalTempv5 is a reconstructed dataset, meaning that the entire period of record is recalculated each month with new data. Based on those new calculations, the new historical data can bring about updates to previously reported values. These factors, together, mean that calculations from the past may be superseded by the most recent data and can affect the numbers reported in the monthly climate reports."
This is normal for climatology. When a new month's data is added they don't just append it to the prior data tables. They re-calculate every temperature in the dataset using the latest modelling software. Climatology, as a science, cannot actually tell us "how hot was it on this day 10 years ago" because there are many different answers.
This is already a terrible, terrible problem for their credibility. But the worst thing of all is that these adjustments are so large that they materially change the entire shape of the 20th century temperature history. Decades ago, global temperature graphs showed a long period of declines starting around WW2. Hence why global cooling was such a big topic in the 60s (incidentally, Heller's site shows ample archival evidence that this was indeed mainstream climatology back then). But if you look at modern temperature graphs, you cannot see any such decline. It's been erased from the historical record.