Disaster Data Releases from Planet Labs PBC

# Planet Crisis Response — Venezuela Earthquake (2026) (llms.txt)

> High-resolution Planet imagery of the earthquake that struck Venezuela's central coast on
> 24 June 2026, causing extensive damage around Caracas, La Guaira, Catia La Mar, Puerto
> Cabello, Valencia, Ocumare de la Costa and Yumare. Post-event 50 cm SkySat and Pelican
> collects (26–28 June 2026) are paired with a pre-event Q1-2026 Planet quarterly basemap for
> before/after comparison. Part of the Planet Crisis Response Program
> (https://www.planet.com/disasterdata/). Imagery © Planet Labs PBC, CC-BY-NC-4.0
> (non-commercial).

This file describes the event for machines and AI assistants. It is one event in the
Disaster Data umbrella catalog (https://data.source.coop/planet/disasterdata). Built with
Portolan (https://portolan-sdi.org).

Base URL: https://data.source.coop/planet/disasterdata/venezuela-earthquake-2026-06-24
Root catalog: https://data.source.coop/planet/disasterdata/venezuela-earthquake-2026-06-24/catalog.json
Browse: https://browser.portolan-sdi.org/#/external/data.source.coop/planet/disasterdata/venezuela-earthquake-2026-06-24/catalog.json
Human landing page: https://source.coop/planet/disasterdata/venezuela-earthquake-2026-06-24

## Structure

- **post-event/** — a single STAC Collection of 50 cm SkySat + Pelican collects acquired
  26–28 June 2026 over the affected coast. 17 scene Items, each tagged with `location` /
  `location_slug` properties for the 7 towns: caracas, la-guaira, catia-la-mar,
  puerto-cabello, valencia, independencia-ocumare, yumare. Items live under
  `post-event/<location>/items/<scene>/`.
- **pre-event/** — the `quarterly-mosaic` Collection: Planet's Q1 2026 global quarterly
  basemap clipped to the area (~4.8 m RGB), as 40 tile Items plus two merged whole-area COGs
  (`mosaic`, a JPEG-compressed display COG, and `mosaic-lossless`). Pre-earthquake, for
  before/after comparison.

STAC-GeoParquet indexes (one row per scene): `post-event/items.parquet`,
`pre-event/items.parquet`.

## Items & assets

- One Item per scene; Item id = the Planet scene id (e.g. `20260627_112621_ssc2_u0001`).
- Post-event asset keys (subset present per scene):
  - `visual` — true-colour RGB COG (uint8), 50 cm, display-ready
  - `analytic` — pansharpened 4-band BGRN COG (uint16), 50 cm, for analysis
  - `udm2` — Usable Data Mask 2 (clear / cloud / shadow / haze + confidence)
  - `udm` — Unusable Data Mask (single band)
  - `thumbnail` — small PNG preview
- Item properties include `datetime`, `gsd` (0.5), `eo:cloud_cover`, view/sun angles,
  `constellation` (skysat | pelican), `platform`, and `proj:code`.
- Asset hrefs are absolute public https URLs. (Interim: the COGs are currently served from
  Planet's original delivery repo `planet/venezuela-earthquake-2026-06-24`; use the exact
  href from each Item / the parquet — do not construct paths by hand.)

## Data access

Find post-event scenes over a town and get their COG URLs (only the parquet is read):

```sql
INSTALL spatial; LOAD spatial; INSTALL httpfs; LOAD httpfs;
SELECT id, datetime, properties['eo:cloud_cover'] AS cloud,
       assets['visual']['href'] AS visual_cog
FROM read_parquet(
  'https://data.source.coop/planet/disasterdata/venezuela-earthquake-2026-06-24/post-event/items.parquet')
WHERE properties['location_slug'] = 'caracas'
ORDER BY cloud;
```

Open a COG window without downloading the whole file (href from the query above):

```python
import rasterio
from rasterio.windows import from_bounds
with rasterio.open(visual_cog) as ds:            # /vsicurl streams only the needed bytes
    rgb = ds.read(window=from_bounds(*bbox, ds.transform))
```

Before/after: pull a `pre-event` mosaic tile (or the merged `mosaic` COG) covering the same
bbox and compare against the matching post-event `visual` scene.

## Explore with AI

Point Claude Code (https://claude.com/claude-code) or another agent at this llms.txt and ask
it to, e.g., "map the least-cloudy post-event SkySat scene over La Guaira" or "build a
pre/post swipe map for Caracas."

## Notes & license

- Non-commercial license: CC-BY-NC-4.0. Attribution: © Planet Labs PBC — Crisis Response Program.
- Contact: crisis-response@planet.com
- Pelican `analytic` assets carry `bands` by identity (blue/green/red/nir) but no
  sensor-specific centre wavelengths.
- Coordinates are lon/lat WGS84 (EPSG:4326) in STAC metadata; each COG carries its own
  projected CRS (`proj:code`).