Four open notebooks test climate data against measurements, from Oslo to Sydney
Northflow's four open climval notebooks now run on public data only, from weather stations in Oslo, Madrid, Nairobi, Toronto and Sydney to Sentinel-3 over eastern Spain. Each one is a worked example of validating climate data with climval, our open Python library.

Four questions
How close is ERA5 to measured station hours in five cities, 2019 to 2020? How do five CMIP6 models compare with ERA5 over Europe, 1985 to 2014? How does Sentinel-3 land surface temperature match ERA5 over eastern Spain in spring 2025? And how well do published NeuralGCM forecasts hold up over ten days?
No account, no substitute data
Every notebook downloads its data from public sources with no account needed, and every figure comes from that data. If a source cannot be reached, the notebook stops with an error. Each one ends with its full citations.
The notebooks are open under Apache 2.0 on GitHub, and each opens in Binder with one click.


