← Founder Notes
Archive

The ground is now being asked to file its own settlement report

Yethikrishna ROriginal on Threads

the ground is now being asked to file its own settlement report: a physics guided model predicts surface deformation with a root mean square error of 0.43 millimeters and beats lstm and transformer baselines, a unet trained on interferometric phase patterns finishes full scene analysis in minutes instead of days, and a graph network tracks 13,505 ground points over six years.

the earth now files its own tilt log.

Context

A paper on InSAR surface deformation monitoring and prediction, a case study of Lishi District, Shanxi Province, China, describes a physics-guided deep learning method and reports an RMSE of 0.43 mm and an R2 of 0.92, outperforming time-series prediction models such as LSTM and iTransformer. It carries a search-metadata date of 13 May 2026.

How it compares

The 0.43 mm figure matches this paper's abstract. It comes from a single regional case study and is the authors' own result. The note's unet trained on interferometric phase patterns and the graph network tracking 13,505 ground points over six years were not traced to a source here, and the paper names iTransformer rather than a plain transformer, so beats transformer baselines is only partly matched.

Watch next

  • The unet and graph-network items in the note.

Sources

  1. A Physics-Guided Deep Learning Method for Temporal InSAR Surface Deformation Monitoring and Prediction: Lishi District, Shanxi (listing)exa.ai

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 4 October 2026 at 17:22 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/the-ground-is-now-being-asked-to-file-DeEkhLUALKm" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="The ground is now being asked to file its own settlement report"></iframe>

More notes