OrebitBlog
Announcement

GeoSuite Is Now Open Source — and Every Number Can Be Checked

Orebit GeoSuite is now free software under the GNU GPL v3: the code, the methods and two tutorials on real public drillhole data whose every number is re-measured from the app. What changed, and why it matters for a geologist's credibility.

By · · 4 min read ·847 words

A resource number is only as good as the steps behind it. A reviewer cannot audit a black box, and a geologist should not have to trust one. So from today, Orebit GeoSuite is free, open-source software under the GNU General Public License v3.0:

  • Code: github.com/ghoziankarami/geosuite. Core, Assay and Resource, every calculation from lab-code handling to kriging, in plain JavaScript you can read, test and change.
  • Methods: a methodology document that says what each step does, what it assumes, and where it stops being valid.
  • Tutorials on real data: two end-to-end walk-throughs on public drillhole databases, in English and Bahasa Indonesia, whose numbers are re-measured from the app on every change.

What you can do under GPL-3.0

Use GeoSuite for any project, commercial consulting included. Read the code, change it, run your own version. Share it, modified or not, under the same licence with its source. Your data and your exported results are yours; the licence covers the software, not what you do with it.

There is no paid tier, no licence key, no e-mail gate, no account. If GeoSuite saves you time, a voluntary donation via Saweria helps keep it maintained.

Mac, Linux, Windows — and offline

GeoSuite now installs as an app on any computer. In Safari on a Mac (macOS 14+) choose File → Add to Dock; in Chrome or Edge on Mac, Windows, Linux or ChromeOS click Install app. After that, all three modules open from your Dock or app list and work with no connection. The Windows Desktop Edition is still available, without a licence key.

Tutorials that cannot go stale

Most software tutorials are screenshots of a version that no longer exists. Ours are programs as well as text. A script drives the real app, step by step, records every number the tutorial quotes, and recomputes the key ones independently. A test fails if the app and the text ever disagree.

Vignette 01 — Thalanga VMS, Queensland. 717 holes, of which only 13 are in the deposit and none have lithology logging. You will see laboratory codes such as −995,000 ppm Pb that must not become grades, a 1,380 g/t Au value that must be blanked rather than “corrected”, why the software’s default search produced 202 Mt (43 times historical production), and how a constrained estimate compares with the 4.7 Mt @ 8.3 % Zn actually mined in 1989–1998.

Vignette 02 — Babbitt Cu-Ni, Duluth Complex. 399 holes and 35,616 assays in files that are in feet without saying so. 61 % of the core was never assayed, which moves the mean Cu grade by a factor of 2.6 depending on whether “unassayed” means unknown or zero. It also covers when a top-cut is justified, and why a geological inventory of several billion tonnes is not a Mineral Resource until it passes the RPEEE test.

The tutorials found bugs — that is the point

Driving real data through the whole chain exposed more than a dozen defects and gaps in GeoSuite itself. Among them:

  • samples below the last survey station were stacked on that station when the collar file had no depth column;
  • composites cut from one long interval shared one coordinate, which inflates the variogram nugget;
  • leave-one-out cross-validation found neighbours for only 10 samples of 200 on dense data;
  • the variogram and cross-validation changed from run to run;
  • variogram auto-fit reported a 60 % nugget on every dataset, because 60 % was where its search stopped; the nugget now comes from pairs within each hole;
  • and the largest: estimation had no domain boundary, so a domain’s grade was spread into every block within search range. Thalanga went from 202 Mt to a screen of 3.7 Mt @ 8.1 % Zn once blocks had to sit next to ore composites — against 4.7 Mt @ 8.3 % actually mined.

All are fixed, and each one that affects a number is now pinned by a test that fails on the old code. We also renamed the confidence tiers to high / medium / low confidence. They are a computational screen, and KCMI 2017 and JORC 2012 reserve Measured, Indicated and Inferred for a Competent Person’s classification.

What GeoSuite is, and is not

GeoSuite carries a drillhole dataset from raw CSV to a documented, reproducible estimate: validation, desurvey, statistics, domaining, compositing, variography, kriging, validation, grade-tonnage. It does not produce a Mineral Resource statement, and it will not pretend to. Its job is to make the numeric spine of an early-stage estimate fast, transparent and checkable, and to hand a Competent Person a clean, honest starting point.

Get involved

  • Try Core now — the Thalanga data is already loaded.
  • Read the code and the tutorials, open an issue, or send a pull request. A small anonymised file that reproduces a problem is the most valuable bug report there is.
  • Have a public dataset with a known answer (production, a published resource)? It could be the next vignette.

Try Core free → · Explore all 3 modules → — free and open source (GPL-3.0), no trial, no paywall. Find it useful? Support Orebit, any amount.

Tagged with

Part of the Orebit ecosystem — geological workflow tools for drillhole validation, resource estimation, and JORC/KCMI reporting.
→ Explore GeoSuite

Try it

Try the toolkit this article uses.

Orebit GeoSuite — single-file HTML, works offline, no install. From CSV to resource report in one afternoon.

Explore GeoSuite →
# From this article:
open geosuite.orebit.id
load(your_drillhole.csv)
apply(workflow_above)

# Done. Ship the report.