Congrats to Shang Tian and Dongfeng Li, our paper on Arctic river sediment it out today! We highlight how small Arctic rivers have increased sediment and changed disproportionally faster (over the satellite record). New machine-learning approach on landsat data. www.nature.com/articles/s41...
😁
And Orbán just congratulated his opponent and admitted that his party lost the elections 🎉. To be fair, that is something that certain politicians on this side of the ocean are unable and unwilling to do
Things are looking promising in early results from the Hungarian elections. I can’t believe that the Orbán regime is about to end.
Today, Hungarians are voting. These elections loom large the world over. Viktor Orbán is a central figure in the international far right and his loss would be blow to familiar figures like JD Vance and Donald Trump, who are his pupils.
snyder.substack.com/p/the-hungar...
Our new paper in Nature takes a critical look at what happens when you drill into a magma body. We find that by understanding that response properly, you can unwind the processes and get back to what the conditions of magma storage really were.
www.nature.com/articles/s41...
What an amazing antidote to doomscrolling. Never thought that complex logarithms could be this interesting
A worldmap of the spatial distribution of extracted flood events in the Groundsource dataset. The map displays the total number of flood events extracted by the LLM-based pipeline aggregated per grid cell. The data are visualized using a Robinson projection, with event counts represented by a logarithmic color scale. Red points indicate the spatial centroids of reference flood events from the GDACS database.
Excited to announce Groundsource - an open-source dataset of historic flood events! This has easily been one of the coolest projects I've worked on recently!
Thread 🧵 for details and all relevant links. 1/n
Both of those are predictions- right? I guess I was mostly thinking about short-term predictability; in the long term, there will be a limit beyond which the channel location is not predictable, regardless of whether it is a chaotic system or not.
That was my reaction as well
Gorgeous thin section images of inversely graded grain flows in the Navajo Sandstone, by Jim Elder ⚒️🧪 smallpond.ca/jim/sand/sam...
I guess neck cutoffs wouldn't happen unless there was some other kind of platform change - right? And, if you are able to model the cutoff process itself, combining that with the 'normal' planform evolution model gives you predictability beyond T.
Yep. Can’t wait to see high resolution DEM coverage in that part of the world
Rio Javari, Brazil / Peru
Well I looked at the actual map (always a good idea!) and it turns out that the low-curvature segment actually has a pretty high sinuosity. These are a couple of large, mature bends that migrate slowly. Shows how I still don't have a good intuition for reading these plots correctly...
Low sinuosity and *low* curvatures.
I think it just happens to be a segment with low sinuosity and curvatures. The correlation between curvature and migration rate is also the worst there as there is not much signal.
The fact that you have to ask means that the data viz is not good enough 🙂
A plot showing measured curvatures and migration rates along a meandering river and correlated using dynamic time warping. The two curves are quite similar, although the migration rate curve is shifted downstream by about two channel widths.
"migration rates are a highly nonlinear function of channel curvature and they cannot be predicted"
migration rates:
The rig is needed when you want to extract longer cores
Sometimes the vibration is not enough and a bit of pushing and twisting helps
A 4x4 vehicle is needed to drive on the beach
Sawing off the core barrel above the top of the sediment
Did some vibracoring over the weekend on the barrier island between the Gulf of Mexico and East Matagorda Bay, to study how barrier islands are impacted by hurricanes
3D-printable time: An evolving meandering river as a 3D model where time is the 3rd (vertical) dimension 🧪⚒️
Here is a piece of the Yazoo River:
Nobody told him that saying “we remain committed to serve all of humanity as best we can” doesn’t sound as good as he think it does
This does not mean that I am doing less work and less thinking and I am not learning. I am just able to delegate the more boring stuff to an assistant that knows more about coding than I do; and I have more time to think about more interesting questions and to explore more ideas.
I know that there a ton of significant issues with AI and the companies that are behind the technology, but this is an excellent writeup of how thoughtful use of AI tools can speed up scientific work. I have been using Claude Code since January and it has been incredibly useful.
The diffusion-assisted particle reconstruction model (DPRM) framework for 3D particle reconstruction from a single 2D image - Figure 1 from Zhu et al., 2025
Good to see 'segmenteverygrain' being used as the starting point for a model that reconstructs 3D grain shapes from 2D data onlinelibrary.wiley.com/doi/full/10.... ⚒️🧪
The trove of files released by the Department of Justice, illuminates Epstein’s deep interest and entrenchment in the scientific community.
But the files also underscore how he used his power and money in ways that kept women out of places where they might succeed. https://bit.ly/4qWgPLz
