JRC Global Surface Water Download for Openclaw

1. Install Python dependencies required by Openclaw Skills for this workflow:

pip install requests>=2.28.0 tqdm

2. Or install from the project requirements file if available:

pip install -r scripts/requirements.txt

3. Optional: install Google Earth Engine tooling if you want the GEE access path:

pip install geemap earthengine-api

4. Authenticate and initialize Earth Engine if you plan to use GEE-based access:

import ee
ee.Initialize()

5. Verify the skill is available and inspect supported layers:

python scripts/global_surface_water.py list-layers
python scripts/global_surface_water.py info

6. Run a test download with a small bounding box to confirm output and file permissions:

python scripts/global_surface_water.py download \
  --layer occurrence \
  --bbox 116.0 39.5 116.8 40.2 \
  --output ./water/beijing_occurrence.tif

7. For large regions, keep requests conservative and plan disk usage before downloading multiple tiles with Openclaw Skills.

The skill organizes data around a fixed JRC raster taxonomy and tile-based delivery model.

Layer taxonomy

• occurrence: water detection percentage, 0–100%
• change: gain/loss classification, values 0–2
• seasonality: number of months with water, 1–12
• recurrence: recurrence frequency, 1–11
• transition: categorical water-class transitions, values 1–11
• extent: maximum observed water extent, binary
• max_extent: maximum extent with extra quality processing

Output organization

• Single-layer runs typically produce one .tif per request.
• --layer all produces a directory of layer outputs rather than a single file.
• Large-area jobs are automatically decomposed into intersecting tiles, each often ~300–600 MB.
• For downstream analysis, outputs can be mosaicked, reprojected, or imported into PostGIS.

Metadata and interpretation notes

• change values: 0 no change, 1 new water, 2 lost water.
• transition values encode permanent/seasonal/ephemeral shifts.
• recurrence values map to seasonal through permanent frequency bands.
• Openclaw Skills keeps the naming consistent with the JRC catalog so raster layers can be used directly in GIS and remote sensing pipelines.