This page describes [Google Cloud Storage](https://cloud.google.com/storage)
(GCS) buckets to which Custom Satellite Embeddings will be delivered. Imagery is delivered
as [Cloud Optimized Geotiffs](https://cogeo.org/) (COGs).

## COG-backed Earth Engine assets

COGs in Cloud Storage can be loaded into
[Google Earth Engine](https://earthengine.google.com/)
for visualization and analysis. They can be loaded directly using
[ee.Image.loadGeoTIFF](https://developers.google.com/earth-engine/apidocs/ee-image-loadgeotiff.md.txt).
Or you can create Earth Engine `Image` assets backed by COGs in Cloud Storage.
The advantage of creating COG-backed assets is that indexes on spatial and
metadata properties make collections of COG-backed assets performant. See
[the Earth Engine COG guide](https://developers.google.com/earth-engine/Earth_Engine_asset_from_cloud_geotiff)
for information on creating COG-backed Earth Engine assets.

## Bucket Details

The bucket in Google Cloud Storage to which COGs are delivered is owned and
managed by the customer. The delivery location, including the complete path
(e.g. `gs://your-bucket/your-folder`), is provided at the time of order.

## Folder structure

Datasets are delivered with this folder structure:

```bash
gs://your-bucket/your-folder/DATASET-NAME/START-DATE_END-DATE/UTM_ZONE/IMAGE-NAME/*
```

Both `START_DATE` and `END_DATE` have date format `yyyyMMdd`. There may be one
or more `START-DATE_END-DATE` folders depending on the temporal cadence of your
imagery. There may be one or more `IMAGE-NAME` folders per
`START-DATE_END-DATE` folder depending on whether your imagery is stored in
tiles. When a request spans more than five UTM zones but has a total area of
less than 250,000 square kilometers, `UTM_ZONE` is replaced by `global`.

Each `IMAGE-NAME` folder contains:

- `data.tiff`: Cloud-Optimized GeoTIFF file containing the image data.
- `image_manifest.json`: Image metadata not stored in the GeoTIFF header.

See
[the COGs README](https://developers.google.com/earth-engine/guides/custom_embedding_samples_on_gcs_readme.md.txt)
for more details on the COGs and the directory to which the COGs are delivered.