Contents

Details

Analytics

Home

All Products

Docs

Source Cooperative is a Radiant Earth project

invasives · Carl Boettiger · Source Cooperative | Source Cooperative

invasives

Name: invasives
Creator: Carl Boettiger

Contents

root

Details

Visibility: Public
Owner: Carl Boettiger
Created: 9 Nov 2026
Last Updated: 9 Nov 2026

Analytics

28 days

Downloads

Data served

0 B

Countries

28-day downloads

13 Oct 2026

9 Nov 2026

Analytics

README

public-invasives

Invasive-species habitat-suitability surfaces. Canonical STAC: https://s3-west.nrp-nautilus.io/public-invasives/stac-collection.json

Currently one collection: inhabit-v4-2024 — USGS Invasive Species Habitat Tool (INHABIT) v4.0, June 2024, for 12 fire-cycle-relevant invasive plants across the contiguous United States.

STAC: https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/stac-collection.json

Read this before you use it

These are modelled POTENTIAL habitat surfaces, not occurrence records. A high-suitability pixel says the environment could support the species, not that the species is there. Phrase every result as "suitable habitat for X" — never "X is present". What the data supports is an argument about invasion pressure and vulnerability.

Four more constraints that change answers, not just wording:


`-masked` is the default for roadless work	The `-masked` variants suppress suitability outside the MESS training envelope. Roadless country is disproportionately high-elevation, undersampled terrain — exactly where unrestricted extrapolation is least trustworthy. Retention varies ~3× by species (table below).
`fifth` (0.05) is the canonical threshold	Named up front so an analysis cannot pick the threshold that suits its conclusion. `first` (0.01, inclusive) and `tenth` (0.10, targeted) are the sensitivity band and land in phase 2.
Class `-1` is a class, not fill	It is the MESS extrapolation flag. NoData is `-128`. `-1` reaches 48% of a raster (red brome) — a sign test that drops negatives deletes half the map.
CONUS only	INHABIT v4 excludes Alaska, where the Tongass and Chugach are the largest roadless acreage in the National Forest System. Any national claim from this data is a CONUS claim.

Species

Species	Common name	ScienceBase child item
Bromus tectorum	cheatgrass	`66a14c62d34ec831f2c2b1cc`
Bromus rubens	red brome	`66a14559d34ef32deb7ee1a2`
Bromus japonicus	Japanese brome	`66a14239d34ef32deb7ede9c`
Bromus arvensis	field brome	`669fd2e7d34eef99d5abc5eb`
Taeniatherum caput-medusae	medusahead	`66bb7a6ad34e033882814282`
Ventenata dubia	ventenata	`66c3adebd34e03388287af28`
Aegilops cylindrica	jointed goatgrass	`669ee0dcd34eef99d5abbb3b`
Agropyron cristatum	crested wheatgrass	`669ee8e0d34eef99d5abbb50`
Salsola tragus	Russian thistle

Products

Nine products per species = 108 layers, each as a COG and an H3 res-10 hex:

Product	Reducer	Warp resampling	Value
`occurrence`	`mean`	bilinear	relative suitability 0–100 for occurrence, unrestricted
`abundance`	`mean`	bilinear	relative suitability 0–100 for abundance (≥5% cover), unrestricted
`high-abundance`	`mean`	bilinear	relative suitability 0–100 for high abundance (≥25% cover), unrestricted
`occurrence-masked`	`mean`	bilinear	relative suitability 0–100 for occurrence, MESS-restricted
`abundance-masked`	`mean`	bilinear	relative suitability 0–100 for abundance (≥5% cover), MESS-restricted
`high-abundance-masked`	`mean`	bilinear	relative suitability 0–100 for high abundance (≥25% cover), MESS-restricted

mean for the continuous surfaces because relative suitability is a normalized index, not a per-pixel amount — SUM is meaningless; aggregate with AVG or MAX. mode for the class rasters, and nearest-neighbour on their warp: bilinear on class codes invents codes with no upstream referent.

The -masked variants are the same three continuous ensembles restricted to the MESS training envelope. Use them for roadless work; the retention table below is how much surface that removes.

`integrated-binary-*` class codes

Labels are from the release's own documentation, not inferred: INHABIT_VersionHistory.txt ("categorical map combining unsuitable, occurrence suitable, abundance suitable, and high abundance suitable") and Jarnevich et al. 2024 ("…while highlighting any areas of environmental extrapolation"). The FGDC gives the domain -1 .. 3 but no labels.

Code	Class	Meaning
`-1`	Novel environmental conditions	MESS extrapolation flag: at least one predictor at this pixel is outside the range of values used to train the model, so no suitability rank is asserted here. NOT NoData (NoData is -128) and NOT a suitability level -- exclude it from suitability accounting rather than treating it as unsuitable.
`0`	Unsuitable	Below the fifth-percentile threshold for all three model groups.
`1`	Occurrence suitable	At or above the fifth-percentile occurrence threshold, but below the abundance threshold.
`2`	Abundance suitable	At or above the fifth-percentile abundance (>=5% cover) threshold, but below the high abundance threshold.
`3`	High abundance suitable	At or above the fifth-percentile high abundance (>=25% cover) threshold.

Ranks are EXCLUSIVE in the raster but NESTED in the source's own tabulations: occurrence-suitable ground is suitability_class >= 1 (classes 1+2+3), abundance-suitable is >= 2, high-abundance-suitable is = 3. Counting class 1 alone undercounts occurrence-suitable ground by every cell that also clears a higher threshold.

mode keeps only each cell's dominant class and discards the mix, so this hex cannot answer "how much area is high-abundance-suitable" without undercounting to plurality cells. At ~1.55 source pixels per res-10 cell the loss is small but one-directional. Per-class fractional coverage is not produced by this pipeline.

How much surface `-masked` removes

Data pixels retained by the MESS-restricted variant, as a percent of the plain product (measured — exact full-raster bincount over all 108 source rasters):

Species	occurrence	abundance	high abundance
Cenchrus ciliaris (buffelgrass)	56.5%	33.6%	33.6%
Bromus rubens (red brome)	51.5%	79.6%	79.6%
Taeniatherum caput-medusae (medusahead)	64.9%	69.2%	69.2%
Ventenata dubia (ventenata)	78.3%	78.6%	78.6%
Agropyron cristatum (crested wheatgrass)	87.1%	87.7%	87.4%
Salsola tragus (Russian thistle)	93.9%	90.6%	92.0%
Aegilops cylindrica (jointed goatgrass)	92.4%	100.0%	96.4%
Tamarix chinensis/ramosissima (tamarisk)	92.8%

Two things to carry from this table. For buffelgrass, -masked discards two thirds of CONUS — a masked tabulation for that species is a statement about a third of the country, and the masked-vs-plain choice moves the answer more than any threshold choice does. Red brome and medusahead are next. Cheatgrass — the species most of the fire argument rests on — is barely touched at 95%.

One anomaly, recorded rather than smoothed over: for Aegilops cylindrica abundance, the masked file's value histogram is identical to the plain product (100.0% retained; the files differ by 476 bytes of TIFF metadata). Its high-abundance counterpart is genuinely masked. So for that one species × product, -masked implies a safeguard that did nothing.

Road-bias audit — it fires, and it splits the species

gHM (Global Human Modification — built substantially from roads, and already in this catalog as global-human-modification) is a predictor in all 12 models. Restricted to the occurrence models, which is what a distance-to-road gradient actually leans on:

Species	gHM share, occurrence models	rank	Road-distance gradient
Bromus tectorum (cheatgrass)	2.0%	6/23	defensible
Taeniatherum caput-medusae (medusahead)	3.1%	10/26	defensible
Bromus rubens (red brome)	4.0%	8/22	defensible
Ventenata dubia (ventenata)	6.0%	7/26	caution
Bromus japonicus (Japanese brome)	6.3%	6/26	caution
Agropyron cristatum (crested wheatgrass)	7.2%	3/26	caution
Bromus arvensis (field brome)	8.0%	4/27	caution
Salsola tragus (Russian thistle)	12.9%	2/22	circular — do not report

Method: mean AUCdiff per predictor from each species' variableImportance.csv, pooled over the five SAHM algorithms and data splits, negatives clipped to zero, expressed as a share of the positive total. Full per-model-group breakdown in catalog/invasives/BUILD.md.

Consequence. Cheatgrass is the cleanest of the 12 and has no KDE-background occurrence model at all — its only occurrence model uses the bias-mitigated target-background design. Medusahead and red brome are also clean. Tamarisk is not: at 17.1% (rank 3/25) a tamarisk road-distance gradient is substantially the model reproducing its own predictor. Report tamarisk, Russian olive, jointed goatgrass, buffelgrass and Russian thistle as suitability-only.

This does not invalidate the suitability surfaces — human modification is a legitimate predictor of invasion. It invalidates only the inference "suitability rises near roads, therefore roads drive invasion" for the high-gHM species.

Grid


Source	Albers Conical Equal Area, NAD83 / GRS 1980, 98.4693338923368 m (FGDC `absres`; the papers round to "90 m"), 44319 × 30671
Published COG	EPSG:4326, 0.001096100359°, 57745 × 25711, origin (−128.386308874497, 51.268044444672) — identical for all 48 layers, asserted at build time
H3	native 10, parents 9, 8, 0 (≈1.55 source pixels per res-10 cell). `h8` is the catalog's universal join key

Query examples

DuckDB — suitability is an index, so AVG never SUM

MapLibre GL JS (COG via titiler)

Raster layers have no PMTiles; render the COG through a titiler raster-tile endpoint. Use a categorical colormap for integrated-binary-* — a continuous ramp misreads the -1 extrapolation flag as a suitability level.

Not held, deliberately

INHABIT v3.0 (doi:10.5066/P9V54H5K) — superseded, and v4 changed the ensemble from model agreement to continuous relative suitability, so v3 and v4 pixel values are not comparable.
The 2026-02-11 integrated management summaries (doi:10.5066/P14G2CHI) — pre-summarized to 118,433 administrative units, none of which is the inventoried-roadless-area boundary.
INHABIT Global V1 (doi:10.5066/P13AJ46S) — could close the Alaska gap later, but different grid and different models; not pixel-comparable with v4 CONUS.
The other 247 v4 species — a follow-up, not a silent expansion.

Licence

USGS data release; FGDC useconst and accconst are both "None" → US federal work, public domain.

Citation

Jarnevich, C.S., Engelstad, P., Williams, D.A., Shadwell, K.S., Reimer, C.J., Henderson, G.C., Prevey, J.S., and Pearse, I.S., 2024, INHABIT species potential distribution across the contiguous United States (ver. 4.0, June 2024): U.S. Geological Survey data release, https://doi.org/10.5066/P14HNEJF. Method: Jarnevich et al. 2024, NeoBiota 96:261-278, https://doi.org/10.3897/neobiota.96.134842.

README

public-invasives

Invasive-species habitat-suitability surfaces. Canonical STAC: https://s3-west.nrp-nautilus.io/public-invasives/stac-collection.json

Currently one collection: inhabit-v4-2024 — USGS Invasive Species Habitat Tool (INHABIT) v4.0, June 2024, for 12 fire-cycle-relevant invasive plants across the contiguous United States.

STAC: https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/stac-collection.json

Read this before you use it

Four more constraints that change answers, not just wording:


`-masked` is the default for roadless work	The `-masked` variants suppress suitability outside the MESS training envelope. Roadless country is disproportionately high-elevation, undersampled terrain — exactly where unrestricted extrapolation is least trustworthy. Retention varies ~3× by species (table below).
`fifth` (0.05) is the canonical threshold	Named up front so an analysis cannot pick the threshold that suits its conclusion. `first` (0.01, inclusive) and `tenth` (0.10, targeted) are the sensitivity band and land in phase 2.
Class `-1` is a class, not fill	It is the MESS extrapolation flag. NoData is `-128`. `-1` reaches 48% of a raster (red brome) — a sign test that drops negatives deletes half the map.
CONUS only	INHABIT v4 excludes Alaska, where the Tongass and Chugach are the largest roadless acreage in the National Forest System. Any national claim from this data is a CONUS claim.

Species

Species	Common name	ScienceBase child item
Bromus tectorum	cheatgrass	`66a14c62d34ec831f2c2b1cc`
Bromus rubens	red brome	`66a14559d34ef32deb7ee1a2`
Bromus japonicus	Japanese brome	`66a14239d34ef32deb7ede9c`
Bromus arvensis	field brome	`669fd2e7d34eef99d5abc5eb`
Taeniatherum caput-medusae	medusahead	`66bb7a6ad34e033882814282`
Ventenata dubia	ventenata	`66c3adebd34e03388287af28`
Aegilops cylindrica	jointed goatgrass	`669ee0dcd34eef99d5abbb3b`
Agropyron cristatum	crested wheatgrass	`669ee8e0d34eef99d5abbb50`
Salsola tragus	Russian thistle

Products

Nine products per species = 108 layers, each as a COG and an H3 res-10 hex:

Product	Reducer	Warp resampling	Value
`occurrence`	`mean`	bilinear	relative suitability 0–100 for occurrence, unrestricted
`abundance`	`mean`	bilinear	relative suitability 0–100 for abundance (≥5% cover), unrestricted
`high-abundance`	`mean`	bilinear	relative suitability 0–100 for high abundance (≥25% cover), unrestricted
`occurrence-masked`	`mean`	bilinear	relative suitability 0–100 for occurrence, MESS-restricted
`abundance-masked`	`mean`	bilinear	relative suitability 0–100 for abundance (≥5% cover), MESS-restricted
`high-abundance-masked`	`mean`	bilinear	relative suitability 0–100 for high abundance (≥25% cover), MESS-restricted

The -masked variants are the same three continuous ensembles restricted to the MESS training envelope. Use them for roadless work; the retention table below is how much surface that removes.

`integrated-binary-*` class codes

Code	Class	Meaning
`-1`	Novel environmental conditions	MESS extrapolation flag: at least one predictor at this pixel is outside the range of values used to train the model, so no suitability rank is asserted here. NOT NoData (NoData is -128) and NOT a suitability level -- exclude it from suitability accounting rather than treating it as unsuitable.
`0`	Unsuitable	Below the fifth-percentile threshold for all three model groups.
`1`	Occurrence suitable	At or above the fifth-percentile occurrence threshold, but below the abundance threshold.
`2`	Abundance suitable	At or above the fifth-percentile abundance (>=5% cover) threshold, but below the high abundance threshold.
`3`	High abundance suitable	At or above the fifth-percentile high abundance (>=25% cover) threshold.

How much surface `-masked` removes

Data pixels retained by the MESS-restricted variant, as a percent of the plain product (measured — exact full-raster bincount over all 108 source rasters):

Species	occurrence	abundance	high abundance
Cenchrus ciliaris (buffelgrass)	56.5%	33.6%	33.6%
Bromus rubens (red brome)	51.5%	79.6%	79.6%
Taeniatherum caput-medusae (medusahead)	64.9%	69.2%	69.2%
Ventenata dubia (ventenata)	78.3%	78.6%	78.6%
Agropyron cristatum (crested wheatgrass)	87.1%	87.7%	87.4%
Salsola tragus (Russian thistle)	93.9%	90.6%	92.0%
Aegilops cylindrica (jointed goatgrass)	92.4%	100.0%	96.4%
Tamarix chinensis/ramosissima (tamarisk)	92.8%

Road-bias audit — it fires, and it splits the species

Species	gHM share, occurrence models	rank	Road-distance gradient
Bromus tectorum (cheatgrass)	2.0%	6/23	defensible
Taeniatherum caput-medusae (medusahead)	3.1%	10/26	defensible
Bromus rubens (red brome)	4.0%	8/22	defensible
Ventenata dubia (ventenata)	6.0%	7/26	caution
Bromus japonicus (Japanese brome)	6.3%	6/26	caution
Agropyron cristatum (crested wheatgrass)	7.2%	3/26	caution
Bromus arvensis (field brome)	8.0%	4/27	caution
Salsola tragus (Russian thistle)	12.9%	2/22	circular — do not report

Grid


Source	Albers Conical Equal Area, NAD83 / GRS 1980, 98.4693338923368 m (FGDC `absres`; the papers round to "90 m"), 44319 × 30671
Published COG	EPSG:4326, 0.001096100359°, 57745 × 25711, origin (−128.386308874497, 51.268044444672) — identical for all 48 layers, asserted at build time
H3	native 10, parents 9, 8, 0 (≈1.55 source pixels per res-10 cell). `h8` is the catalog's universal join key

Query examples

DuckDB — suitability is an index, so AVG never SUM

MapLibre GL JS (COG via titiler)

Not held, deliberately

INHABIT v3.0 (doi:10.5066/P9V54H5K) — superseded, and v4 changed the ensemble from model agreement to continuous relative suitability, so v3 and v4 pixel values are not comparable.
The 2026-02-11 integrated management summaries (doi:10.5066/P14G2CHI) — pre-summarized to 118,433 administrative units, none of which is the inventoried-roadless-area boundary.
INHABIT Global V1 (doi:10.5066/P13AJ46S) — could close the Alaska gap later, but different grid and different models; not pixel-comparable with v4 CONUS.
The other 247 v4 species — a follow-up, not a silent expansion.

Licence

USGS data release; FGDC useconst and accconst are both "None" → US federal work, public domain.

Citation

1INSTALL httpfs; LOAD httpfs;
2INSTALL h3 FROM community; LOAD h3;
3
4-- Mean cheatgrass occurrence suitability per res-8 cell
5SELECT h8, AVG(suitability) AS mean_suitability, COUNT(*) AS n_res10_cells
6FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked/hex/h0=*/data_0.parquet')
7GROUP BY h8;

1INSTALL httpfs; LOAD httpfs;
2INSTALL h3 FROM community; LOAD h3;
3
4-- Mean cheatgrass occurrence suitability per res-8 cell
5SELECT h8, AVG(suitability) AS mean_suitability, COUNT(*) AS n_res10_cells
6FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked/hex/h0=*/data_0.parquet')
7GROUP BY h8;

1-- Integrated class distribution. `-1` is the extrapolation flag, NOT unsuitable:
2-- exclude it explicitly rather than letting it fall in with class 0.
3SELECT suitability_class, COUNT(*) AS cells
4FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth/hex/h0=*/data_0.parquet')
5GROUP BY suitability_class ORDER BY suitability_class;
6
7-- Occurrence-suitable ground is `>= 1` (classes 1+2+3), not `= 1`.
8SELECT COUNT(*) AS occurrence_suitable_cells
9FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth/hex/h0=*/data_0.parquet')
10WHERE suitability_class >= 1;

1-- Integrated class distribution. `-1` is the extrapolation flag, NOT unsuitable:
2-- exclude it explicitly rather than letting it fall in with class 0.
3SELECT suitability_class, COUNT(*) AS cells
4FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth/hex/h0=*/data_0.parquet')
5GROUP BY suitability_class ORDER BY suitability_class;
6
7-- Occurrence-suitable ground is `>= 1` (classes 1+2+3), not `= 1`.
8SELECT COUNT(*) AS occurrence_suitable_cells
9FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth/hex/h0=*/data_0.parquet')
10WHERE suitability_class >= 1;

1-- Join two species at h8 (both are native res 10, so h8 is present in both).
2-- Both species below have PUBLISHED hex; species with only a COG cannot be joined this way.
3SELECT a.h8, AVG(a.suitability) AS cheatgrass, AVG(b.suitability) AS red_brome
4FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked/hex/h0=*/data_0.parquet') a
5JOIN read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_rubens/occurrence-masked/hex/h0=*/data_0.parquet') b
6  USING (h10)
7GROUP BY a.h8;

1-- Join two species at h8 (both are native res 10, so h8 is present in both).
2-- Both species below have PUBLISHED hex; species with only a COG cannot be joined this way.
3SELECT a.h8, AVG(a.suitability) AS cheatgrass, AVG(b.suitability) AS red_brome
4FROM read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked/hex/h0=*/data_0.parquet') a
5JOIN read_parquet('https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_rubens/occurrence-masked/hex/h0=*/data_0.parquet') b
6  USING (h10)
7GROUP BY a.h8;

1// Continuous suitability, 0-100
2map.addSource("cheatgrass-occ", {
3  type: "raster",
4  tiles: [
5    "https://titiler.../cog/tiles/{z}/{x}/{y}.png" +
6    "?url=https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked.tif" +
7    "&rescale=0,100&colormap_name=inferno"
8  ],
9  tileSize: 256,
10});
11map.addLayer({ id: "cheatgrass-occ", type: "raster", source: "cheatgrass-occ" });
12
13// Integrated class map - discrete colours from the STAC classification:classes color_hint
14const cmap = encodeURIComponent(JSON.stringify({
15  "-1": [123, 104, 166, 255],  // novel environmental conditions (MESS extrapolation)
16  "0":  [242, 242, 242, 255],  // unsuitable
17  "1":  [254, 217, 118, 255],  // occurrence suitable
18  "2":  [253, 141,  60, 255],  // abundance suitable
19  "3":  [189,   0,  38, 255],  // high abundance suitable
20}));
21map.addSource("cheatgrass-class", {
22  type: "raster",
23  tiles: [
24    "https://titiler.../cog/tiles/{z}/{x}/{y}.png" +
25    "?url=https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth.tif" +
26    "&colormap=" + cmap
27  ],
28  tileSize: 256,
29});
30map.addLayer({ id: "cheatgrass-class", type: "raster", source: "cheatgrass-class" });

1// Continuous suitability, 0-100
2map.addSource("cheatgrass-occ", {
3  type: "raster",
4  tiles: [
5    "https://titiler.../cog/tiles/{z}/{x}/{y}.png" +
6    "?url=https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/occurrence-masked.tif" +
7    "&rescale=0,100&colormap_name=inferno"
8  ],
9  tileSize: 256,
10});
11map.addLayer({ id: "cheatgrass-occ", type: "raster", source: "cheatgrass-occ" });
12
13// Integrated class map - discrete colours from the STAC classification:classes color_hint
14const cmap = encodeURIComponent(JSON.stringify({
15  "-1": [123, 104, 166, 255],  // novel environmental conditions (MESS extrapolation)
16  "0":  [242, 242, 242, 255],  // unsuitable
17  "1":  [254, 217, 118, 255],  // occurrence suitable
18  "2":  [253, 141,  60, 255],  // abundance suitable
19  "3":  [189,   0,  38, 255],  // high abundance suitable
20}));
21map.addSource("cheatgrass-class", {
22  type: "raster",
23  tiles: [
24    "https://titiler.../cog/tiles/{z}/{x}/{y}.png" +
25    "?url=https://s3-west.nrp-nautilus.io/public-invasives/inhabit-v4-2024/bromus_tectorum/integrated-binary-fifth.tif" +
26    "&colormap=" + cmap
27  ],
28  tileSize: 256,
29});
30map.addLayer({ id: "cheatgrass-class", type: "raster", source: "cheatgrass-class" });

invasives

invasives

public-invasives

Read this before you use it

Species

Products

integrated-binary-* class codes

How much surface -masked removes

Road-bias audit — it fires, and it splits the species

Grid

Query examples

DuckDB — suitability is an index, so AVG never SUM

MapLibre GL JS (COG via titiler)

Not held, deliberately

Licence

Citation

public-invasives

Read this before you use it

Species

Products

integrated-binary-* class codes

How much surface -masked removes

Road-bias audit — it fires, and it splits the species

Grid

Query examples

DuckDB — suitability is an index, so AVG never SUM

MapLibre GL JS (COG via titiler)

Not held, deliberately

Licence

Citation

`integrated-binary-*` class codes

How much surface `-masked` removes

`integrated-binary-*` class codes

How much surface `-masked` removes