Maching learning in ocean observations

The world’s oceans take up twenty five percent of our annual CO2 emissions to the atmosphere. This helps slow down global warming. Keeping track of this uptake is essential for understanding climate change.
We have millions of verified data points. These are published as the surface ocean CO2 atlas, SOCAT. But, for every year, only a fraction of the vast global ocean is surveyed.
How do we go from these spotty measurements to an estimate of the carbon uptake of the full ocean? The clue is neural networks and artifical intelligence.
Animated and edited by Eli Kristin Muriaas. Produced with support from the Research Council of Norway (RapidVERIFY, #309571). Thanks to ICOS; Glen Peters, CICERO; Peter Landschützer, VLIZ: Corinne Le Quere, University of East Anglia; and the University of Exeter.
Read more here: bjerknes.uib.no/en/article/ne...

[@Lisohlavek]

Very interesting, thank you!

[@damien2198]

Estimate of an estimate of an estimate and add "adjustments", all in a black box, what could go wrong ;-)

[@Raivias4]

That's interesting, but why the increase in carbon uptake since 2000? Is it attributed to improved sea life, or... I don't know.