... Why would you use model reanalysis numbers to base the discussion on when real world physical temperature measurements are available (if you can wait a few days for the collation for monthly figures to be published)?
...
You obviously want to use data of the highest quality in different respects. Satellite data are considered to be higher quality than 'physical' data.
First, because you cover the entire globe in a more or less equally weighted grid. Physical data is only measured at certain spots, often near populated areas. We have very good coverage where many people live, but very little coverage where no people live. SH has e.g. very few measuring points compared to NH. Oceans in SH have very few points.
Secondly, also physical data has issues. Measuring points are moved. The urban heat island is a persistent issue that is not easily resolved. Measuring points are influenced by urban crawl.
Thirdly, if you want to track aerosol masking effects, the more relevant data are those that relate to the amount of heat energy in the atmosphere, i.e. sat data.
All reanalyses are built to a higher or lesser degree on sat data.
As regards the quality issue with NCEP it seems to be a model related temporary problem. Moyhu writes: "Although I noted doubts about NCEP/NCAR's recent results, the pattern of anomalies was qualitatively similar to that of TempLS."
https://moyhu.blogspot.com/2020/08/july-global-surface-templs-down-0039.htmlYou can check for yourself that this was a temporary issue on the page provided by Karsten Hausten,
http://www.karstenhaustein.com/climate.phpIf you scroll to the very bottom of that page, you can compare the monthly means (CFSR-GFS, NCEP reanalysis, GISS temperature). You can follow the issues with NCEP , they are clearly noticeable since April of 2020.
And, finally, to come back on topic, there is still no signal at all to be seen from the diminished aerosols due to corona lockdowns. My conclusion is that there might be such signals on a regional level, but we will have to wait for the researchers to dig it out.
Moyhu btw provides a really nice zoomable trackball Earth temperature presentation, built on a mix of physical data and sat data:
https://moyhu.blogspot.com/p/blog-page_24.htmlExample: If you zoom in on China you would expect to be able to see some lockdown effect on temperatures in e.g. April and May this year, but April was actually a cold month in China. May was warm, but not exceptionally so.
