*Scientific papers* 2/2

A #lowcost #deeplearning methodology integrating #OpenStreetMap for #urban planning tasks across multiple domains. #OSM #vectors serve as reference masks for #satelliteimagery to enable predictions via #CNN and #GNN. Accuracy >90%.
New #remotesensing products require #vector #geodata. Hence the value of investing in the completeness and consistency of #OSM data
https://www.
nature.com/articles/s41598-026

So there's a prerogative to be as audacious as possible. Move the needle as far as possible every time...

...because we never know at the time what heuristics we are bound by if we just do the known stuff.

And we might never go back. So it is imperative to think about what data we can collect, and how, to help paint in gaps we can't see.

/fin

#seaice #remotesensing #antarctica

...to avoid falling into snark I'll leave it there.

To recap: we did new and expensive and difficult things on sea ice because people on planet earth rely on accurate weather forecasts and want to understand seasonality. All this stuff would probably play some role in your insurance premiums even!

I've *always* held that in mind on the ice. Why am I here? to get a paper? well yes, however thats a by-product of [see above].

#seaice #remotesensing #antarctica

Unfortunately due to funding we didn't realise the full picture - direct coincident airborne and satellite observations, with airborne data tuned using spatially-appropriate validation points.

We did confirm a longstanding bias in observations 😬, which [is still] a heuristic interplay of field site selection, ship navigation choices, satellite data using those to tune results, and models using all of the above to tune outputs..

a heavy momentum to shift!

#seaice #remotesensing #antarctica

...we can't observe everything in situ, so those spaceborne instruments zooming around collecting large area data are super important (to your weather forecasts, your fishing trip, whether your roses bloom when you expect them to).

All the work above was about understanding what is integrated into a satellite data point. Are we seeing what we think we see?

Are our assumptions correct?

Are the numbers we use in weather and climate models still valid?

#seaice #remotesensing #antarctica

There's a lot going in oceans very few people ever see let alone visit. Especially in or near winter time!

Yes we can do things from space. This next diagram shows some relative scales. Those ice features in the photos above? Not visible sorry.

Things have improved a bit with ICESat-2, Sentinels, GEDI and some upcoming missions. However we're still more or less stuck at a very blurry picture.

Especially if products are also integrated over time...

#seaice #remotesensing #antarctica