Ocean Baseline Pack

Why

Terminal expansions, marinas, outfalls, dredging and coastal protection works all change a shoreline. Once the excavator arrives, what the beach looked like before is gone, and arguments about damage, restoration targets or compensation come down to whoever kept a record. The satellite part of this pack can be recomputed later; the human surveys and photos in it cannot be re-collected.

What's in a pack

• Every shoreline cell in your area (~180 m squares on the waterline), scored from a Sentinel-2 composite and ranked: CSV + GeoJSON for QGIS/ArcGIS.
• Every BeachBook human survey inside the area: full answers (sand, rocks, logs, oysters, clams, garbage, access ...), aggregates and the original photos. Contributors anonymised.
• Provenance: exact Sentinel-2 scene IDs and dates, model file hashes, survey export timestamp.
• Maps of the cells over aerial imagery with survey sites marked.
• A dated README and a SHA-256 manifest with one pack hash that fixes every file. Change one byte anywhere and the hash no longer matches.

Read this before buying: what each number is worth

Same honesty rule as Territory View. Only one output is a calibrated model, and even it is weak.

Field	Grade	What it can be used for
sand	MODEL	Calibrated on 318 surveyed beaches, cross-validated R² ≈ 0.20. Good for ranking where the sandy shore is, not a measured sand percentage.
rock_prob / sandiness	SCREENING-GRADE	Classifier, CV-AUC ≈ 0.69. Points at where to send a person. Not a substrate measurement.
oyster_prob	SCREENING-GRADE	Tidal presence/absence classifier, CV-AUC ≈ 0.73 on 179 beaches. Habitat likelihood only. Not harvest-safety guidance and says nothing about DFO closures.
BeachBook surveys	Observations	What people recorded on the beach. Older records came over in a Sept 2025 database migration, so their saved date is the migration date; camera dates in photo filenames are given as hints.

Built structures (ramps, docks, breakwaters) are not masked and can score as "sandy"; it shows on the map. Sentinel-2 is 10–20 m: nothing here resolves individual boulders or shell beds. Where your area has no BeachBook surveys, the pack says so; we don't invent them.

Free sample: Sidney / Tsehum Harbour / Swartz Bay, BC

18 km² around the Swartz Bay ferry terminal, Tsehum Harbour marinas and the Sidney waterfront. 108 shoreline cells, 12 surveyed beaches, 22 survey photos. Imagery: 8 Sentinel-2 scenes, Aug–Sep 2024. Packed 30 Sep 2026.

Download sample pack (ZIP, 12.6 MB)  Read the README first

Sample pack hash (photos downscaled to 1600 px for download size; the originals' hashes are inside):

0bcfa0592706334fe1467d02df0795e0df22316a5601fec733866d71f20cfc02

Price

Per project area up to ~20 km² of BC coast, delivered within a week. Full-resolution survey photos, both maps, and the pack hash emailed to you separately on delivery day so the date is on record outside our servers. Larger areas or before/after re-runs of a pack you already hold: ask.

Order a baseline pack

Send the area (a map screenshot, a bounding box, or a project name) and when works are expected to start. Contains modified Copernicus Sentinel data (ESA). Survey data: BeachBook community.

Linear project instead of a shoreline? See Major-project corridors below.

Major-project corridors

Pipelines, transmission lines, rail and highway twinning run hundreds of kilometres and go through review in stages. Once a project is designated or fast-tracked, the pre-construction condition of the land along the route matters to everyone involved: the proponent, landowners, Indigenous rights holders, regulators and the project's critics. A corridor pack is a neutral, dated record of that "before". We don't take a position on any project, and the pack reads the same whoever orders it.

What's in a corridor pack

• The route cut into 1 km segments, each a strip along the centreline (±500 m by default, or the width you need), numbered by kilometre post (KP). CSV + GeoJSON.
• Every summer back to 2019, for every segment: Sentinel-2 peak-season vegetation (NDVI) and moisture (NDMI), with cloud, snow and shadow masked per pixel and every scene ID recorded.
• A segment-by-segment screen: each kilometre's latest season compared with its own history, using the same fixed-threshold rules as our CCS surface-watch work for carbon-storage sites. Statuses: within own range, watch, screen, or not enough clear imagery.
• Ground Boots field records along the strip, where we have them. Ground-check visits to flagged segments can be added.
• Coastal ends (landfalls, marine terminals) can be packed as an Ocean Baseline Pack, with the BeachBook surveys above.
• The same dated README, provenance and SHA-256 manifest with one pack hash as the shoreline packs.

What it is and isn't

Output	Grade	What it can be used for
NDVI / NDMI per segment per year	Measurement	Band arithmetic on Sentinel-2 surface reflectance. No model runs anywhere in a corridor pack. It is a 10–20 m satellite measure, taken as a median over each segment.
Watch / Screen flags	SCREENING PROXY	A kilometre that sits outside its own multi-year range. Harvest, clearing, fire, insects, drought and road work all cause this. A flag means "send a person here first". It is never a finding about a project.
Ground Boots records	Observations	Geotagged, timestamped field notes and photos. Where there are none, the pack says so.

Free sample: Lake Louise to Field, Kicking Horse Pass (27.4 km)

An illustrative corridor along the existing Trans-Canada Highway and rail line over Kicking Horse Pass. We chose it because it is an existing linear corridor with Ground Boots field records already on file. It is not the route of any proposed project. 28 segments, 119 Sentinel-2 tile-scenes from summers 2019–2026, 5 Ground Boots records in the strip. Packed 2 October 2026.

Download sample corridor pack (ZIP, 2.9 MB)  Read the README first

Sample pack hash (public version: Ground Boots notes, photos and times removed):

1d6c75378e05305e18abc3e0bf375fb1d99f085f9dd26a85a7c6c51e587229d9

Corridor price

Priced by centreline kilometre, with a 25 km minimum ($1,000). Kilometres are how routes are described in filings, and our 1 km segments match that one-to-one, so a 180 km route is $7,200 whether the proponent splits it into three spreads or twelve. Standard strip ±500 m. Wider strips, multiple route alternatives and annual re-runs during construction (to compare against the sealed "before"): ask. Delivered within two weeks for routes up to 500 km. The pack hash is emailed separately on delivery day so the date is on record outside our servers.

Ask about a corridor pack

Send the route (KML, shapefile, GeoJSON, or a map with the route drawn on) and the width you need. Contains modified Copernicus Sentinel data (ESA). Basemap: Esri World Imagery.