The International Council on Mining and Metals (ICMM) is a mining industry group that was founded in London in 2001. In September, they released a Global Mining Dataset containing over 8,000 mines and other related assets around the globe.
Below is a heatmap of the assets they catalogued in this dataset.
In this post, Iβll explore ICMMβs mining dataset.
My Workstation
Iβm using a 5.7 GHz AMD Ryzen 9 9950X CPU. It has 16 cores and 32 threads and 1.2 MB of L1, 16 MB of L2 and 64 MB of L3 cache. It has a liquid cooler attached and is housed in a spacious, full-sized Cooler Master HAF 700 computer case.
The system has 96 GB of DDR5 RAM clocked at 4,800 MT/s and a 5th-generation, Crucial T700 4 TB NVMβ¦
The International Council on Mining and Metals (ICMM) is a mining industry group that was founded in London in 2001. In September, they released a Global Mining Dataset containing over 8,000 mines and other related assets around the globe.
Below is a heatmap of the assets they catalogued in this dataset.
In this post, Iβll explore ICMMβs mining dataset.
My Workstation
Iβm using a 5.7 GHz AMD Ryzen 9 9950X CPU. It has 16 cores and 32 threads and 1.2 MB of L1, 16 MB of L2 and 64 MB of L3 cache. It has a liquid cooler attached and is housed in a spacious, full-sized Cooler Master HAF 700 computer case.
The system has 96 GB of DDR5 RAM clocked at 4,800 MT/s and a 5th-generation, Crucial T700 4 TB NVMe M.2 SSD which can read at speeds up to 12,400 MB/s. There is a heatsink on the SSD to help keep its temperature down. This is my systemβs C drive.
The system is powered by a 1,200-watt, fully modular Corsair Power Supply and is sat on an ASRock X870E Nova 90 Motherboard.
Iβm running Ubuntu 24 LTS via Microsoftβs Ubuntu for Windows on Windows 11 Pro. In case youβre wondering why I donβt run a Linux-based desktop as my primary work environment, Iβm still using an Nvidia GTX 1080 GPU which has better driver support on Windows and ArcGIS Pro only supports Windows natively.
Installing Prerequisites
Iβll use Python 3.12.3 and a few other tools to help analyse the data in this post.
$ sudo add-apt-repository ppa:ubuntugis/ubuntugis-unstable
$ sudo apt update
$ sudo apt install \
jq \
python3-pip \
python3.12-venv
Iβll set up a Python Virtual Environment and install an OSM vector tile data extraction utility I wrote last year.
$ python3 -m venv ~/.osm
$ source ~/.osm/bin/activate
$ git clone https://github.com/marklit/tiles2columns \
~/tiles2columns
$ python -m pip install \
-r ~/tiles2columns/requirements.txt
Iβll use DuckDB v1.3.0, along with its H3, JSON, Lindel, Parquet and Spatial extensions, in this post. Normally I try and use the latest release of DuckDB but v1.4.0 has an issue where itβs Parquet files arenβt readable by many of the tools I use at the moment.
$ cd ~
$ wget -c https://github.com/duckdb/duckdb/releases/download/v1.3.0/duckdb_cli-linux-amd64.zip
$ unzip -j duckdb_cli-linux-amd64.zip
$ chmod +x duckdb
$ ~/duckdb
INSTALL h3 FROM community;
INSTALL lindel FROM community;
INSTALL json;
INSTALL parquet;
INSTALL spatial;
Iβll set up DuckDB to load every installed extension each time it launches.
$ vi ~/.duckdbrc
.timer on
.width 180
LOAD h3;
LOAD lindel;
LOAD json;
LOAD parquet;
LOAD spatial;
The maps in this post were mostly rendered with QGIS version 3.44. QGIS is a desktop application that runs on Windows, macOS and Linux. The application has grown in popularity in recent years and has ~15M application launches from users all around the world each month.
I used QGISβ Tile+ plugin to add basemaps from Esri to one of the maps in this post.
The dark, non-satellite map of ICMMβs asset locations above is mostly made up of vector data from Natural Earth and Overture.
Analysis-Ready Data
Iβll download ICMMβs Excel file for this dataset.
$ wget -O global-mining-dataset.xlsx \
'https://www.icmm.com/website/data/2025/global-mining-dataset.xlsx?cb=117612'
Iβll use DuckDB to clean up the values and produce both a geometry field and a bounding box for each feature in this dataset. This will make working with this dataset remotely, such as from AWS S3, much easier.
$ ~/duckdb
CREATE OR REPLACE TABLE icmm AS
SELECT * EXCLUDE(Latitude,
Longitude,
"Primary Commodity",
"Secondary Commodity",
"Other Commodities"),
ST_POINT(Longitude, Latitude) AS geometry,
{'xmin': ST_XMIN(ST_EXTENT(geometry)),
'ymin': ST_YMIN(ST_EXTENT(geometry)),
'xmax': ST_XMAX(ST_EXTENT(geometry)),
'ymax': ST_YMAX(ST_EXTENT(geometry))} AS bbox,
TRIM("Primary Commodity") AS "Primary Commodity",
TRIM("Secondary Commodity") AS "Secondary Commodity"
FROM READ_XLSX('global-mining-dataset.xlsx',
sheet='External',
ignore_errors=true);
This following produced a ZStandard-compressed, spatially-sorted Parquet file of ICMMβs dataset.
COPY (
SELECT * EXCLUDE (geometry),
ST_ASWKB(geometry) geometry
FROM icmm
WHERE ST_X(ST_CENTROID(geometry)) IS NOT NULL
ORDER BY HILBERT_ENCODE([ST_Y(ST_CENTROID(geometry)),
ST_X(ST_CENTROID(geometry))]::double[2])
) TO 'icmm.parquet' (
FORMAT 'PARQUET',
CODEC 'ZSTD',
COMPRESSION_LEVEL 22,
ROW_GROUP_SIZE 15000);
Data Fluency
The above Parquet file is 508 KB and contains 8,508 rows. Below is an example record.
$ echo "SELECT * EXCLUDE(bbox),
bbox::JSON bbox
FROM READ_PARQUET('icmm.parquet')
WHERE ICMMID = 'ICMM08147'" \
| ~/duckdb -json \
| jq -S .
[
{
"Asset Type": "Mine",
"Confidence Factor": "High",
"Country or Region": "Canada",
"Group Names": "Diavik;Diavik",
"ICMMID": "ICMM08147",
"Mine Name": "Diavik",
"Primary Commodity": "diamond",
"Secondary Commodity": null,
"bbox": {
"xmax": -110.2743898,
"xmin": -110.2743898,
"ymax": 64.49236014,
"ymin": 64.49236014
},
"geometry": "POINT (-110.2743898 64.49236014)"
}
]
Below are the field names, data types, percentages of NULLs per column, number of unique values and minimum and maximum values for each column.
SELECT column_name,
column_type,
null_percentage,
approx_unique,
min,
max
FROM (SUMMARIZE
FROM icmm)
WHERE column_name != 'geometry'
AND column_name != 'bbox'
ORDER BY 1;
βββββββββββββββββββββββ¬ββββββββββββββ¬ββββββββββββββββββ¬ββββββββββββββββ¬βββββββββββββββββββββββββββββββββββββββββββββββ¬βββββββββββββββββββββ
β column_name β column_type β null_percentage β approx_unique β min β max β
β varchar β varchar β decimal(9,2) β int64 β varchar β varchar β
βββββββββββββββββββββββΌββββββββββββββΌββββββββββββββββββΌββββββββββββββββΌβββββββββββββββββββββββββββββββββββββββββββββββΌβββββββββββββββββββββ€
β Asset Type β VARCHAR β 0.00 β 17 β Mine β Steel Plant β
β Confidence Factor β VARCHAR β 0.00 β 3 β High β Very Low β
β Country or Region β VARCHAR β 0.00 β 129 β Afghanistan β Zimbabwe β
β Group Names β VARCHAR β 50.25 β 3639 β # 5 Mine;No. 5 Coal Mine (KY);Grapevine Mine β ΓΕSA Coal Mine;CSA β
β ICMMID β VARCHAR β 0.00 β 7920 β ICMM00026 β ICMM21432 β
β Mine Name β VARCHAR β 0.00 β 8754 β Bobrek Piekary Coal Mine β ΓΕSA Coal Mine β
β Primary Commodity β VARCHAR β 0.00 β 43 β alumina β zircon β
β Secondary Commodity β VARCHAR β 72.57 β 60 β alumina β zircon β
βββββββββββββββββββββββ΄ββββββββββββββ΄ββββββββββββββββββ΄ββββββββββββββββ΄βββββββββββββββββββββββββββββββββββββββββββββββ΄βββββββββββββββββββββ
Commodities
Below is the number of assets per primary community type broken down by confidence factor.
.maxrows 100
PIVOT READ_PARQUET('icmm.parquet')
ON "Confidence Factor"
USING COUNT(*)
GROUP BY "Primary Commodity"
ORDER BY "Primary Commodity";
ββββββββββββββββββββββββββ¬ββββββββ¬βββββββββββ¬βββββββββββ
β Primary Commodity β High β Moderate β Very Low β
β varchar β int64 β int64 β int64 β
ββββββββββββββββββββββββββΌββββββββΌβββββββββββΌβββββββββββ€
β alumina β 64 β 5 β 0 β
β aluminium β 91 β 9 β 0 β
β antimony β 0 β 1 β 5 β
β barium β 0 β 0 β 134 β
β bauxite β 17 β 5 β 0 β
β borates β 0 β 1 β 0 β
β boron β 0 β 0 β 2 β
β chromite β 2 β 27 β 4 β
β chromium β 3 β 9 β 46 β
β coal β 508 β 729 β 0 β
β cobalt β 3 β 0 β 12 β
β copper β 336 β 212 β 353 β
β diamond β 5 β 0 β 0 β
β ferrochrome β 0 β 2 β 0 β
β ferromanganese β 3 β 2 β 0 β
β ferronickel β 8 β 1 β 0 β
β ferroniobium β 0 β 1 β 0 β
β ferrosilicon manganese β 0 β 1 β 0 β
β fluorspar β 0 β 1 β 0 β
β gold β 379 β 261 β 237 β
β heavy mineral sands β 3 β 0 β 0 β
β ilmenite β 1 β 3 β 0 β
β iron ore β 299 β 312 β 0 β
β lanthanides β 0 β 1 β 0 β
β lead β 31 β 46 β 33 β
β lithium β 34 β 7 β 24 β
β manganese β 5 β 1 β 0 β
β mercury β 0 β 0 β 2 β
β metallurgical coal β 366 β 525 β 0 β
β molybdenum β 2 β 1 β 0 β
β nickel β 52 β 25 β 71 β
β niobium β 1 β 0 β 0 β
β palladium β 1 β 0 β 0 β
β phosphate β 2 β 5 β 0 β
β platinum β 24 β 2 β 59 β
β potash β 1 β 0 β 0 β
β silver β 49 β 18 β 179 β
β steel β 8 β 75 β 0 β
β thermal coal β 675 β 1382 β 0 β
β tin β 1 β 20 β 277 β
β titanium β 1 β 0 β 0 β
β tungsten β 0 β 11 β 94 β
β uranium β 1 β 3 β 3 β
β vanadium β 1 β 0 β 0 β
β zinc β 183 β 18 β 89 β
β zircon β 0 β 1 β 0 β
ββββββββββββββββββββββββββ΄ββββββββ΄βββββββββββ΄βββββββββββ€
β 46 rows 4 columns β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Below are the most common primary and secondary commodity pairs.
SELECT "Primary Commodity",
"Secondary Commodity",
COUNT(*)
FROM READ_PARQUET('icmm.parquet')
WHERE "Secondary Commodity" IS NOT NULL
GROUP BY 1, 2
ORDER BY 3 DESC
LIMIT 50;
ββββββββββββββββββββββ¬ββββββββββββββββββββββ¬βββββββββββββββ
β Primary Commodity β Secondary Commodity β count_star() β
β varchar β varchar β int64 β
ββββββββββββββββββββββΌββββββββββββββββββββββΌβββββββββββββββ€
β gold β silver β 398 β
β metallurgical coal β thermal coal β 396 β
β copper β gold β 361 β
β zinc β lead β 115 β
β copper β silver β 53 β
β lead β zinc β 52 β
β gold β copper β 48 β
β copper β cobalt β 44 β
β copper β lead β 40 β
β nickel β cobalt β 40 β
β copper β molybdenum β 40 β
β silver β gold β 36 β
β nickel β copper β 35 β
β copper β zinc β 28 β
β gold β platinum β 28 β
β gold β lead β 27 β
β copper β nickel β 27 β
β silver β lead β 23 β
β lead β silver β 22 β
β zinc β copper β 21 β
β tin β tungsten β 20 β
β chromite β chromium β 20 β
β zinc β silver β 20 β
β platinum β palladium β 19 β
β alumina β aluminium β 19 β
β iron ore β manganese β 17 β
β silver β zinc β 13 β
β lithium β tantalum β 11 β
β tin β zinc β 11 β
β lead β copper β 10 β
β silver β copper β 10 β
β gold β palladium β 9 β
β steel β iron β 8 β
β gold β antimony β 8 β
β ferronickel β nickel β 8 β
β copper β platinum β 8 β
β alumina β bauxite β 8 β
β gold β uranium β 7 β
β zinc β cadmium β 7 β
β iron ore β copper β 6 β
β iron ore β gold β 6 β
β gold β zinc β 6 β
β gold β arsenic β 6 β
β lithium β potash β 5 β
β tin β silver β 5 β
β chromium β chromite β 5 β
β gold β iron ore β 5 β
β gold β nickel β 5 β
β zinc β gold β 5 β
β antimony β gold β 4 β
ββββββββββββββββββββββ΄ββββββββββββββββββββββ΄βββββββββββββββ€
β 50 rows 3 columns β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Assets
This is the number of asset types in this dataset broken down by confidence factor. Some assets are the combination of two or more types.
PIVOT READ_PARQUET('icmm.parquet')
ON "Confidence Factor"
USING COUNT(*)
GROUP BY "Asset Type"
ORDER BY "Asset Type";
ββββββββββββββββββββββββββ¬ββββββββ¬βββββββββββ¬βββββββββββ
β Asset Type β High β Moderate β Very Low β
β varchar β int64 β int64 β int64 β
ββββββββββββββββββββββββββΌββββββββΌβββββββββββΌβββββββββββ€
β Mine β 2702 β 3518 β 1576 β
β Mine;Plant β 68 β 48 β 7 β
β Mine;Refinery β 13 β 5 β 0 β
β Mine;Refinery;Plant β 0 β 1 β 0 β
β Mine;Smelter β 22 β 21 β 1 β
β Mine;Smelter;Plant β 3 β 3 β 0 β
β Mine;Smelter;Refinery β 18 β 11 β 0 β
β Plant β 9 β 11 β 31 β
β Refinery β 59 β 16 β 1 β
β Refinery; Steel Plant β 1 β 0 β 0 β
β Refinery;Plant β 3 β 1 β 1 β
β Smelter β 143 β 64 β 7 β
β Smelter;Plant β 3 β 1 β 0 β
β Smelter;Refinery β 106 β 3 β 0 β
β Smelter;Refinery;Plant β 5 β 2 β 0 β
β Smelter;Steel Plant β 2 β 0 β 0 β
β Steel Plant β 3 β 18 β 0 β
ββββββββββββββββββββββββββ΄ββββββββ΄βββββββββββ΄βββββββββββ€
β 17 rows 4 columns β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
Countries
Below are the most and least represented countries in this dataset.
.maxrows 25
SELECT COUNT(*),
"Country or Region"
FROM READ_PARQUET('icmm.parquet')
GROUP BY 2
ORDER BY 1 DESC;
ββββββββββββββββ¬ββββββββββββββββββββββ
β count_star() β Country or Region β
β int64 β varchar β
ββββββββββββββββΌββββββββββββββββββββββ€
β 1839 β China β
β 1627 β United States β
β 588 β Australia β
β 473 β Indonesia β
β 424 β India β
β 328 β Russia β
β 272 β South Africa β
β 252 β Brazil β
β 206 β Peru β
β 197 β Canada β
β 146 β Bolivia β
β 144 β Chile β
β 136 β Mexico β
β Β· β Β· β
β Β· β Β· β
β Β· β Β· β
β 2 β Fiji β
β 1 β Rwanda β
β 1 β Trinidad and Tobago β
β 1 β Gabon β
β 1 β Bangladesh β
β 1 β Bahrain β
β 1 β Nepal β
β 1 β Luxembourg β
β 1 β Taiwan β
β 1 β Niger β
β 1 β Qatar β
β 1 β Greenland β
ββββββββββββββββ΄ββββββββββββββββββββββ€
β 129 rows (25 shown) 2 columns β
ββββββββββββββββββββββββββββββββββββββ
Iβll generate a heatmap of the asset locations in this dataset.
CREATE OR REPLACE TABLE h3_3_stats AS
SELECT H3_LATLNG_TO_CELL(
bbox.ymin,
bbox.xmin, 3) AS h3_3,
COUNT(*) num_buildings
FROM READ_PARQUET('icmm.parquet')
WHERE bbox.xmin BETWEEN -178.25 AND 178.25
GROUP BY 1;
COPY (
SELECT ST_ASWKB(H3_CELL_TO_BOUNDARY_WKT(h3_3)::geometry) geometry,
num_buildings
FROM h3_3_stats
) TO 'h3_3_stats.gpkg'
WITH (FORMAT GDAL,
DRIVER 'GPKG',
LAYER_CREATION_OPTIONS 'WRITE_BBOX=YES');
Diavik Diamond Mine
Iβll download OpenStreetMapβs (OSM) data for Canadaβs Diavik Diamond Mine. This is using the tiles2columns utility I built last year following OSMβs addition of vector-based tiles to their service.
$ mkdir -p ~/diavik
$ cd ~/diavik
$ python3 ~/tiles2columns/main.py \
centroid \
-110.29299 64.48937 \
--distance 0.1
The above extracted 1.2 MB worth of GeoPackage (GPKG) files. Theyβre grouped by feature type.
$ du -hsc *
96K buildings.gpkg
96K dam_polygons.gpkg
124K land.gpkg
96K public_transport.gpkg
96K street_labels.gpkg
128K streets.gpkg
148K water_lines.gpkg
404K water_polygons.gpkg
1.2M total
Umbra are a satellite manufacturer and constellation operator. They produce Synthetic Aperture Radar (SAR) satellites that can see through clouds and capture imagery of the earth day or night.
They operate an open feed of some of the imagery they capture. One of the locations they capture regular imagery from is Diavik Diamond Mine.
Iβve placed four images they captured on August 26th and September 5th, 17th and 26th on top of one another. Each image has a slightly different footprint so the four images in combination cover a wider area than any one image. Given how mines can change from day to day Iβve constrained this overlapping of images to a few weeks.
The image below has Esriβs World Imagery basemap tinted blue. Tinted red is Umbraβs imagery and the vector data on top is from OSM. Iβve used EPSG:3978 for the projection as this is in Northern Canada.
Thank you for taking the time to read this post. I offer both consulting and hands-on development services to clients in North America and Europe. If youβd like to discuss how my offerings can help your business please contact me via LinkedIn.