STAC-like GeoParquet Generator

This Python script helps you generate STAC-like items in a GeoParquet file from a list of geospatial files (.fgb, .geojson, .tif/.cog).

It supports:

• Automatic extraction of datetime from filenames.
• Mandatory style link for STAC Browser or web clients.
• Inferring asset types from file extensions, with an optional override.
• Optional bounding box for geometry.
• Validation of existing GeoParquet files to check structure and required fields.

---

Features

• Input formats: FlatGeobuf (.fgb), GeoJSON (.geojson), Cloud Optimized GeoTIFF (.tif, .tiff, .cog)
• Auto-merge assets by day (optional)
• Asset key reindexing (asset_0, asset_1, …) to avoid nulls
• Validation of required fields: id, datetime, geometry, bbox, assets, links
• STAC-compatible structure

---

Requirements

Python 3.11+ and the following libraries:

numpy==1.24.4
pandas==2.0.0
geopandas==1.0.1
shapely==2.0.7
pyarrow==12.0.0

Install in a clean environment:

python -m venv myenv
source myenv/bin/activate  # Windows: myenv\Scripts\activate
pip install --upgrade pip
pip install -r requirements.txt

---

Usage

Generate GeoParquet from file list

python generate_geoparquet.py \
  --style-url "https://raw.githubusercontent.com/gtif-cerulean/assets/main/style.json" \
  --base-url "https://your-bucket.example.com/daily" \
  file1.fgb file2.geojson file3.tif \
  -o items.parquet

Note: --style-url needs to be provided in order for eodash to visulize the data

Generate GeoParquet from CSV

CSV must contain at least a path column. Optionally, include date.

python generate_geoparquet.py \
  --style-url "https://raw.githubusercontent.com/gtif-cerulean/assets/main/style.json" \
  --base-url "https://your-bucket.example.com/daily" \
  --csv input_files.csv \
  -o items.parquet

Generate GeoParquet from JSON

JSON should be a list of objects with a path key:

python generate_geoparquet.py \
  --style-url "https://raw.githubusercontent.com/gtif-cerulean/assets/main/style.json" \
  --base-url "https://your-bucket.example.com/daily" \
  --json input_files.json \
  -o items.parquet

Optional: Specify bounding box

python generate_geoparquet.py \
  --bbox -10 40 10 60 \
  --style-url "https://raw.githubusercontent.com/gtif-cerulean/assets/main/style.json" \
  --base-url "https://your-bucket.example.com/daily" \
  file1.fgb file2.geojson

Optional: Override asset type

python generate_geoparquet.py \
  --asset-type "image/tiff; application=geotiff; profile=cloud-optimized" \
  --style-url "https://raw.githubusercontent.com/gtif-cerulean/assets/main/style.json" \
  --base-url "https://your-bucket.example.com/daily" \
  daily_20250106.cog

---

Validate Existing GeoParquet

Check if the GeoParquet file contains all required fields and correct structure:

python generate_geoparquet.py --validate items.parquet

Output will indicate missing columns or malformed assets/links.

---

Date Extraction

The script attempts to parse datetime from filenames automatically. Supported formats include:

• YYYYMMDD (e.g., 20250106)
• YYYY-MM-DD (e.g., 2025-01-06)
• YYYYMMDDHHMM (e.g., 202501061230)
• YYYY-MM-DDTHHMM or YYYY-MM-DDTHH:MM:SS (e.g., 2025-01-06T12:30:45)

If no date is found, it falls back to current UTC time.

---

Notes

• Each input file becomes a STAC item with asset_0. Multiple files per day can be merged manually or via workflow.
• Asset keys are reindexed to avoid null fields when saving to GeoParquet.
• Style link are required in order for eodash to visulize the assets correctly.