USGS Mineral Commodity Summaries
Case study · Infrastructure
A Data & Geopolitics Exploration
- What
- ~50 commodities from the USGS Mineral Commodity Summaries, every figure page-cited.
- Stack
- Python PDF extraction with per-commodity regression tests; static sortable table.
- Status
- LIVE
- Dates
- Published May 2026 · Updated June 2026
- Links
- Live site · Source code
Background
As trade policy and geopolitics shift, what do the fundamentals actually say about U.S. mineral supply and import reliance?
This viewer makes the USGS Mineral Commodity Summaries readable across commodities instead of one PDF at a time: net import reliance for any material, which countries those imports come from, and what USGS itself flags under "Events, Trends & Issues" — every figure traceable to the page it was printed on. Gallium, germanium, the rare earths, cobalt, and antimony sit at the center of export controls, tariff fights, and reshoring debates, and this is the authoritative source on all of them.
A Look Inside
Each view shown on mobile and desktop — tap any image to open the live site.
How It Works
Every value links back to the exact page of the official PDF it came from. The tool surfaces what USGS published and doesn't add anything on top. I vibe-coded the extraction pipeline and then pinned it down with per-commodity regression tests so the parsing stays honest:
- Data source: the public-domain USGS Mineral Commodity Summaries (MCS 2026) PDFs, covering ~50 commodities — lithium, rare earths, platinum-group metals, gallium, germanium, cobalt, antimony.
- What's extracted: the salient-statistics tables (2021–2025 imports, exports, net import reliance, price), import sources by country, world production and reserves, the prose blocks, and every footnote — units preserved exactly as printed.
- Where it comes from: each record carries the source URL, the PDF's SHA-256 hash, and the capture date; the viewer shows the source page inline so any number can be checked against the original.
- UX: a fast static sortable table with a category taxonomy (rare earths only, PGM only) and CSV export, deployed to GitHub Pages via Actions.



