Following numerous disruptions to the upstream data flows that feed into the Berkeley Earth observational record, we committed to rebuilding our ingest pipeline to ensure a more robust, resilient record, one less dependent on any single source of data. That rebuild, our Pipeline 2.0, is now live.

Show Post

The continuous satellite record for Arctic sea ice monitoring started in late October 1978, but rescued Nimbus data sets, available from the NASA NSIDC DAAC, go back to the 1960s. https://nsidc.org/data/nimbus

Show Post

I've not done any analysis of the piecing together of different data, but from my quick read of what was done I'd say that Darwin=f(Jakarta) is likely to be more reliable than Tahiti=f(Santiago).

Show Post

More on the HadSST v4.1 to v4.2 changes: Improvements to the Engine Room Intake biases during World War 2 and addressing the digitasation truncation bias in the Japanese Kobe SST collection identified by Chan et al. (2019). Sandford & Rayner (2026) doi.org/10.1002/joc....

Show Post

New version of the HadCRUT global temperature dataset released @colinmorice.bsky.social HadCRUT5.2 uses HadSST4.2 (with improved adjustments for SST biases during 1933-1961) and CRUTEM5.2 (with updates to stations in China, USA and Canada). Here is the difference from the previous version 5.1

Show Post