3 ms·
I encourage people to look into the ERA5 dataset provenance especially when you approach the observations made toward the "pre industrial date" of 1850 . Remem
by tonymet 3y ago
I encourage people to look into the ERA5 dataset provenance especially when you approach the observations made toward the "pre industrial date" of 1850 .
Remember that modern global surface temperatures are collected by satellites, and the dataset is comingled with recordings observed visually & made by hand using buckets by sailors who were not primarily academic researchers. Segments of high resolution, low noise data (satellites) are mixed with low resolution, low coverage, high noise records (hand records on a boat made surrounding the united kingdom).
My point is to be in awe of the technical aspects of this effort but also keep in mind that we are only making copies of low resolution, noisy manuscripts from sailors 170 years ago.
- relaxing 3y agoOk? And what’s your point in pointing out that?
- tonymet 3y agoThe data is noisy so be careful when using it for research. Always account for the provenance of the records when working with "data".
- tonymet 3y agoone of the project's goals was to load the data and make predictions. The page covered the data loading part, but not the methods and error tolerance in the predictions
- shoyer 3y agoERA5 covers 1940 to present. That's well before the satellite era (and the earlier data absolutely has more quality issues) but there's nothing from 170 years ago.
- tonymet 3y agoSimilar noise issues apply. Most of the other surface temp models have to cover 1850