The shoreline is now being asked to draw its own future
the shoreline is now being asked to draw its own future: a shallow network trained on 30 years of satellite shorelines from the nile delta forecasts the coast to 2050 under wave energy and subsidence, a gaussian process gives hainan island shoreline changes with uncertainty bounds, and ai plus current data regrew a maldives beach 90 feet seaward.
the sand now files its own retreat forecast.
Context
The Nile Delta preprint (North Carolina State University, listed August 2025) trains a shallow artificial neural network on satellite-derived shorelines from 1992 to 2022, with wave energy, sea level, land cover and subsidence as inputs, plus feature selection and uncertainty analysis. It was validated on 2022 shoreline data. It projects cumulative land loss of 5.3, 10.7 and 18.4 km2 by 2030, 2040 and 2050, with the biggest risk near Burullus Lake.
The Hainan Island preprint (Preprints.org, September 2026) combines 30 years of Landsat shorelines (1990 to 2019), the Digital Shoreline Analysis System and Gaussian Process Regression with Automatic Relevance Determination, using ERA5 climate variables. Temporal validation gave R2 0.781, RMSE 11.81 m and bias of -0.66 m. Projections for 2030 and 2050 show retreat of up to about 70 m west of Sanya and advance of up to about 55 m in the northwest Qiongzhou Strait area, with uncertainty growing to about 32 to 40 m.
The Maldives work is reported by Quartz, citing The New York Times. Coastal Assembly, a company that grew out of MIT research, uses ten years of satellite images plus ocean current data to find and forecast erosion, then designs submerged structures to steer currents so sand settles. Off Kaafu Atoll it sank 54 hexagonal units. The report says one beach gained 90 feet and a second site built a sandbar holding over 30,000 cubic feet of sand within six months.
These are three separate projects, so the note's 'sand files its own retreat forecast' joins things that differ in kind. The Nile and Hainan items are forecasts from preprints that were not peer reviewed in the pages read. The Hainan authors say longitude, latitude and year were the main predictors and that annual wave, wind, temperature and precipitation summaries added little, so the model mostly learns where and when, not the physical drivers the note implies. The Hainan uncertainty of about 32 to 40 m is large next to the 11.81 m validation RMSE. The Maldives result is a company outcome relayed by a news outlet, not a published evaluation, and the 'seaward' direction in the note is not stated in the source read. Whether the 90 feet came from the AI forecast or the structures themselves is not separable from that report.
Related work
- Forecasting Nile Delta Shoreline Change Until 2050 Using a Shallow Neural Network (preprint listing) ↗Source for the 1992 to 2022 training data and the 2030, 2040 and 2050 land loss numbers.
- Shoreline Forecasting and Uncertainty Assessment with Gaussian Process Regression Along the Coastline of Hainan Island, China (Preprints.org listing) ↗Source for the Gaussian Process framework, validation scores and the 2030 and 2050 projections.
- AI system is rebuilding beaches lost to erosion in Maldives (Quartz) ↗Source for the 90 feet, 54 units and 30,000 cubic feet figures. It relays The New York Times.
- Rising Seas Are Gobbling Up Beaches Around the World. A.I. Might Restore Them. (MIT Architecture news) ↗The MIT page carrying the original Times feature on the Maldives project.
- Historical Trend Analysis and Forecasting of Shoreline Change at the Nile Delta Using RS Data and Machine Learning (Remote Sensing, MDPI) ↗An earlier Nile Delta shoreline forecasting study, for comparison.
Watch next
- Check whether the Nile Delta preprint has since been peer reviewed and what its 2022 validation error was. Find the Coastal Assembly measurements behind the 90 feet figure.
Sources
- Forecasting Nile Delta Shoreline Change Until 2050 Using a Shallow Neural Network (preprint listing)exa.ai
- Shoreline Forecasting and Uncertainty Assessment with Gaussian Process Regression Along the Coastline of Hainan Island, China (Preprints.org listing)exa.ai
- AI system is rebuilding beaches lost to erosion in Maldives (Quartz)qz.com
Provenance
The note above is reproduced unedited from the original post, first published on Threads on 4 October 2026 at 01:17 IST. Sources are the papers and datasets the note draws on.
View the original post ↗Embed this note
More notes
The air is now being asked to keep its own ledger
the air is now being asked to keep its own ledger: ecmwf’s aifs compo becomes the first ai model to forecast atmospheric composition globally every three hours, cleanair simulates 365 days of pm2.5 over china in ten seconds, and a unified framework maps six pollutants at one kilometer across the whole country. the air now files its own composition report.
read the note →The current is now being asked to draw its own map
the current is now being asked to draw its own map: china’s langya 2.0 predicts six ocean phenomena including internal waves and mesoscale eddies, a deep net called wenhai resolves eddies globally with air sea flux formulas built in, and scripps infers surface currents from the way temperature patterns deform in satellite images. the ocean now files its own circulation report.
read the note →The soil is now being asked to report its own carbon
the soil is now being asked to report its own carbon: a nix color sensor paired with generative data augmentation predicts soil organic carbon without a lab, random forest drives 74 percent of soil health mapping studies, and sentinel 2 tracks five year carbon change across france and italy from 922 samples. the dirt now files its own carbon account.
read the note →