earthdaily Sentinel2C1L2A

rslearn.data_sources.earthdaily.Sentinel2C1L2A

Sentinel-2 L2A data on EarthDaily platform using the Collection 1 archive (sentinel-2-c1-l2a).

If you are migrating from Planetary Computer or want PC-compatible asset keys and band names, use rslearn.data_sources.earthdaily.Sentinel2L2A instead — it targets the sentinel-2-l2a collection and exposes the same band names as rslearn.data_sources.planetary_computer.Sentinel2.

See EarthDaily Setup for required dependency/credentials.

By default, this data source applies per-asset scale/offset values from STAC raster:bands metadata (apply_scale_offset: true) to convert raw pixel values into physical units using physical = raw * scale + offset. Set apply_scale_offset: false to keep raw values. The underlying COG pixels are stored as integer DN/sample values, not physical reflectance values. rslearn applies the scale/offset during read/materialization unless apply_scale_offset: false is configured. For Sentinel-2 spectral bands, this physical unit is reflectance (typically BOA reflectance for L2A products), e.g. raw 10000 with scale 0.0001 maps to 1.0.

apply_scale_offset is not Sentinel-2 processing-baseline harmonization. It decodes the C1 COG storage values into reflectance. The harmonize option used by Planetary Computer-style Sentinel-2 sources adjusts DN values across processing baselines; this C1 source does not expose a harmonize argument.

When apply_scale_offset: true, configure the target band_sets[].dtype as float32. rslearn will raise during initialization if a non-float dtype is configured through the layer context. Nodata is read from STAC raster:bands metadata and preserved during scale/offset application (for this collection, nodata is typically 0).

Note: EarthDaily may include a preview thumbnail asset; rslearn does not ingest/materialize it.

Collection Status

According to the Earth Search project, sentinel-2-c1-l2a is intended to eventually replace sentinel-2-l2a, and contains COG assets processed to at least baseline 5.0. As noted there (as of April 2024), ESA archive reprocessing to baseline 5.0 is still incomplete, with known gaps for Nov 2016 to Nov 2019 and for 2022, and no published completion date.

Source: https://github.com/Element84/earth-search

Configuration

{
  "class_path": "rslearn.data_sources.earthdaily.Sentinel2C1L2A",
  "init_args": {
    // Whether to apply STAC `raster:bands` scale/offset (default true). Set to false to
    // keep raw values.
    "apply_scale_offset": true,
    // Optional: EarthDaily Sentinel-2 STAC asset keys to fetch (default null).
    // If null and the layer config is available, assets are inferred from the layer's
    // requested band names.
    //
    // Note: this is different from the "Available bands" list below:
    // - "assets" uses EarthDaily STAC asset keys (e.g. "red", "nir", "visual", "scl").
    // - "band_sets[].bands" uses rslearn band names (e.g. "B04", "B08", "R", "scl").
    //
    // Example: ["red", "green", "blue", "nir", "swir16", "swir22", "visual", "scl"]
    "assets": null,
    // Optional: maximum cloud cover (%) to filter items at search time. If set,
    // injects an `eo:cloud_cover` upper bound into the STAC query.
    "cloud_cover_max": null,
    // Maximum number of STAC items to fetch per window before rslearn grouping/matching.
    "search_max_items": 500,
    // Optional ordering of items before grouping (useful with SpaceMode.COMPOSITE +
    // CompositingMethod.FIRST_VALID): "cloud_cover" (default), "datetime", or null.
    "sort_items_by": "cloud_cover",
    // Optional: STAC API `query` filter passed to searches.
    // Example: {"s2:product_type": {"eq": "S2MSI2A"}}
    "query": null,
    // Optional: STAC item property to sort by before grouping/matching (default null).
    // If set, it takes precedence over sort_items_by.
    "sort_by": null,
    // Whether to sort ascending when sort_by is set (default true).
    "sort_ascending": true,
    // Optional cache directory for cached item metadata.
    "cache_dir": null,
    // Timeout for HTTP asset downloads.
    "timeout": "10s",
    // Retry settings for EarthDaily API client requests (search/get item).
    "max_retries": 3,
    "retry_backoff_factor": 5.0
  }
}

Example layer snippet:

{
  "type": "raster",
  "band_sets": [{
    "bands": ["B02", "B03", "B04"],
    "dtype": "float32",
    // Optional: override nodata explicitly if needed.
    // "nodata_vals": [0, 0, 0]
  }],
  "data_source": {
    "class_path": "rslearn.data_sources.earthdaily.Sentinel2C1L2A",
    "init_args": {
      "apply_scale_offset": true
    }
  }
}

Available Bands

Available rslearn band names (select via band_sets[].bands; rslearn infers required EarthDaily assets when assets is null): - B01 - B02 - B03 - B04 - B05 - B06 - B07 - B08 - B09 - B11 - B12 - B8A - R, G, B (from the visual asset) - scl, aot, wvp

Common EarthDaily asset key to rslearn band name mapping: - coastal → B01 - blue → B02 - green → B03 - red → B04 - rededge1 → B05 - rededge2 → B06 - rededge3 → B07 - nir → B08 - nir08 → B8A - nir09 → B09 - swir16 → B11 - swir22 → B12 - visual → R, G, B - scl → scl - aot → aot - wvp → wvp