Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/terrafloww/rasteret/llms.txt

Use this file to discover all available pages before exploring further.

​
Overview

Rasteret includes a catalog of 15+ built-in datasets (Sentinel-2, Landsat, NAIP, etc.). You can:
  • Contribute new datasets to the public catalog
  • Register private datasets locally
  • Share custom DatasetDescriptor JSONs with your team
This guide covers the full workflow from descriptor creation to upstream contribution.

​
DatasetDescriptor Anatomy

From src/rasteret/catalog.py:50-186, a DatasetDescriptor captures: 
from rasteret.catalog import DatasetDescriptor

descriptor = DatasetDescriptor(
    # --- Identity ---
    id="acme/field-survey-2024",
    name="ACME Field Survey",
    description="Drone imagery from field surveys",
    
    # --- Access (pick one: STAC or GeoParquet) ---
    stac_api="https://api.acme.com/stac/v1",
    stac_collection="field-survey-2024",
    # OR
    geoparquet_uri="s3://acme-data/items.parquet",
    
    # --- Band mapping ---
    band_map={
        "R": "image",
        "G": "image",
        "B": "image",
        "NIR": "image",
    },
    band_index_map={"R": 0, "G": 1, "B": 2, "NIR": 3},
    separate_files=False,  # Single multi-band COG
    
    # --- Coverage metadata ---
    spatial_coverage="north-america",
    temporal_range=("2024-01-01", "2024-12-31"),
    requires_auth=False,
    license="CC-BY-4.0",
    license_url="https://creativecommons.org/licenses/by/4.0/",
    commercial_use=True,
    
    # --- Cloud configuration (optional) ---
    cloud_config={
        "provider": "aws",
        "requester_pays": False,
        "region": "us-west-2",
        "url_patterns": {},
    },
    
    # --- Examples for docs/tests ---
    example_bbox=(-122.5, 37.7, -122.3, 37.9),
    example_date_range=("2024-06-01", "2024-07-01"),
)

​
Prerequisites for Contributing

Before opening a PR to add a dataset, verify:

​
1. Data Source is Reachable

The STAC API or GeoParquet URI must be publicly accessible (or gated behind standard OAuth/SAS). 
# For STAC APIs
curl "https://api.example.com/stac/v1/collections/my-collection"

# For GeoParquet
AWS_NO_SIGN_REQUEST=YES aws s3 ls s3://bucket/items.parquet

​
2. Band Map Points to Parseable COGs

Assets must be tiled GeoTIFFs with TileOffsets and TileByteCounts: 
import rasteret
from rasteret.catalog import DatasetDescriptor, DatasetRegistry

# Register descriptor
DatasetRegistry.register(descriptor)

# Attempt a build
collection = rasteret.build(
    "acme/field-survey-2024",
    name="test",
    bbox=descriptor.example_bbox,
    date_range=descriptor.example_date_range,
)

print(f"✓ Build succeeded: {collection.dataset.count_rows()} rows")
From CONTRIBUTING.md:33-39, enrichment must succeed end-to-end.

​
3. License Metadata is Verified

Confirm:
  • license matches the STAC collection’s license field (typically an SPDX ID like "CC-BY-4.0" or "proprietary")
  • license_url points to the full license text (from STAC’s rel=license link)
  • commercial_use is False if the license prohibits commercial use (e.g., CC-BY-NC)

​
4. Example Works End-to-End

import rasteret

collection = rasteret.build(
    "acme/field-survey-2024",
    name="test",
    bbox=descriptor.example_bbox,
    date_range=descriptor.example_date_range,
)

# Fetch a sample
array = collection.get_numpy(
    geometries=descriptor.example_bbox,
    bands=["R", "G", "B"],
)

assert array.shape[0] > 0, "No data returned"
print(f"✓ Read succeeded: {array.shape}")

​
Adding a Dataset to the Catalog

​
Step 1: Create the Descriptor

Edit src/rasteret/catalog.py and add your descriptor after the existing entries: 
# At the end of the built-in descriptors section (~line 720)
DatasetRegistry.register(
    DatasetDescriptor(
        id="acme/field-survey-2024",
        name="ACME Field Survey",
        description="Drone imagery from field surveys, 10cm resolution",
        stac_api="https://api.acme.com/stac/v1",
        stac_collection="field-survey-2024",
        band_map={
            "R": "image",
            "G": "image",
            "B": "image",
            "NIR": "image",
        },
        band_index_map={"R": 0, "G": 1, "B": 2, "NIR": 3},
        separate_files=False,
        spatial_coverage="north-america",
        temporal_range=("2024-01-01", "2024-12-31"),
        license="CC-BY-4.0",
        license_url="https://creativecommons.org/licenses/by/4.0/",
        example_bbox=(-122.5, 37.7, -122.3, 37.9),
        example_date_range=("2024-06-01", "2024-07-01"),
    )
)

​
Step 2: Run Tests

uv run pytest -k test_catalog
Ensure the new descriptor appears in DatasetRegistry.list().

​
Step 3: Add Documentation

Create an example page in docs/examples/ showing how to use the dataset: 
---
title: "ACME Field Survey"
description: "High-resolution drone imagery for crop monitoring"
---

## Overview

The ACME Field Survey dataset provides 10cm resolution RGB+NIR imagery...

## Quick Start

```python
import rasteret

collection = rasteret.build(
    "acme/field-survey-2024",
    name="field_survey",
    bbox=(-122.5, 37.7, -122.3, 37.9),
    date_range=("2024-06-01", "2024-07-01"),
)


### Step 4: Open a Pull Request

From `CONTRIBUTING.md:26-32`, before submitting:

1. Run `uv run ruff check src/` and `uv run ruff format --check src/`
2. Run `uv run pytest -q` - all tests should pass
3. Sign your commits with `git commit -s` (DCO)

Include in your PR description:
- Link to the data provider's documentation
- Confirmation that the license permits inclusion
- Output from the example query showing successful reads

## Registering Local Datasets

For private or experimental datasets, use `register_local()`:

```python
import rasteret
from pathlib import Path

# Build and persist a collection
collection = rasteret.build_from_stac(
    name="my_private_collection",
    stac_api="https://private-api.example.com/stac/v1",
    collection="my-collection",
    bbox=(77.5, 12.9, 77.7, 13.1),
    date_range=("2024-01-01", "2024-06-30"),
)

collection.export("private_collection")

# Register it as a local dataset
from rasteret.catalog import DatasetDescriptor

descriptor = rasteret.register_local(
    dataset_id="local/my-private-collection",
    path="private_collection",
    name="My Private Collection",
    description="Internal field campaign data",
    persist=True,  # Save to ~/.rasteret/datasets.local.json
)

print(f"Registered: {descriptor.id}")
Now it’s available via rasteret.build(): 
# In a new session
import rasteret

collection = rasteret.build(
    "local/my-private-collection",
    name="reloaded",
)

​
Sharing Custom Descriptors

Export a descriptor as JSON and share with your team: 
from rasteret.catalog import export_local_descriptor

export_local_descriptor(
    dataset_id="local/my-private-collection",
    output_path="shared_descriptor.json",
)
Team members can load it: 
import json
from rasteret.catalog import DatasetDescriptor, DatasetRegistry

with open("shared_descriptor.json") as f:
    descriptor_dict = json.load(f)

descriptor = DatasetDescriptor(**descriptor_dict)
DatasetRegistry.register(descriptor)

# Now usable with rasteret.build()

​
GeoParquet-Backed Descriptors

For datasets distributed as GeoParquet instead of STAC: 
from rasteret.catalog import DatasetDescriptor, DatasetRegistry

descriptor = DatasetDescriptor(
    id="maxar/opendata",
    name="Maxar Open Data",
    description="Post-disaster satellite imagery",
    geoparquet_uri="s3://us-west-2.opendata.source.coop/maxar/maxar-opendata/maxar-opendata.parquet",
    column_map={
        "id": "id",
        "date": "datetime",
        "geom": "geometry",
        "tif_url": "assets",
    },
    href_column="tif_url",
    band_index_map={"R": 0, "G": 1, "B": 2},
    bbox_columns={
        "minx": "bbox_minx",
        "miny": "bbox_miny",
        "maxx": "bbox_maxx",
        "maxy": "bbox_maxy",
    },
    band_map={"R": "image", "G": "image", "B": "image"},
    separate_files=False,
    spatial_coverage="global",
    requires_auth=False,
    license="CC-BY-NC-4.0",
    license_url="https://creativecommons.org/licenses/by-nc/4.0/",
    commercial_use=False,
)

DatasetRegistry.register(descriptor)
From src/rasteret/catalog.py:86-101, the normalisation layer constructs assets from href_column and band_index_map.

​
Best Practices

Prefix dataset IDs with the provider or organization: earthsearch/sentinel-2-l2a, pc/sentinel-2-l2a, local/my-data. This avoids collisions when multiple providers host the same collection.
Always include example_bbox and example_date_range that are known to return data. These are used in live tests and documentation.
Run rasteret.build() followed by get_numpy() with your example parameters. If enrichment or reads fail, the dataset isn’t ready for upstreaming.

Build docs developers (and LLMs) love

Get started for freeTalk to us