Migration

Due to a few breaking changes we've made between the old Batch Processing API and Batch Processing V2, we've prepared a migration guide on what has changed between the two APIs.

1. Removal of collections support

With Batch Processing V2, we have standardized how to leverage Sentinel Hub collections - please see ingesting results into BYOC for the new streamlined approach.

This allowed us to simplify and remove the need to create or reference collections directly when using Batch Processing V2 and to remove the endpoints below:

The following fields concerning this functionality have also been removed from request/response schemas:

output.collectionId
output.createCollection

2. Removal of tile status monitoring via REST endpoints

We've removed the ability to monitor the status of your tiles via the provided REST endpoints. This has been instead supplemented by a mechanism that periodically uploads an SQLite database to the root folder of your specified bucket. More information about this functionality can be found here.

The following endpoints have been removed:

3. Changes to the `output` and `zarrOutput` properties

We've removed the zarrOutput property and have instead merged it with the output property. This means that the output property now supports both GeoTIFF and Zarr output. To specify which output format you want, you can use the type property within the output property, alongside all the other properties that were previously used for Zarr and GeoTiff output.

An example of how to specify Zarr output:

{
  "output": {
    "type": "zarr",
    "delivery": {
      "s3": {
        "url": "s3://<your-bucket>/<requestId>",
        "iamRoleARN": "<your-IAM-role-ARN>"
      }
    },
    "group": {
      "zarr_format": 2
    },
    "arrayParameters": {
      "dtype": "<u2",
      "order": "C",
      "chunks": [
        1,
        1000,
        1000
      ],
      "fill_value": 0
    },
    "arrayOverrides": {
      "maxNDVI": {
        "dtype": "<f4",
        "fill_value": "NaN"
      }
    }
  }
}

and an example of how to specify a GeoTiff output:

{
  "output": {
    "type": "raster",
    "delivery": {
      "s3": {
        "url": "s3://<your-bucket>",
        "iamRoleARN": "<your-IAM-role-ARN>"
      }
    },
    "cogOutput": true,
    "cogParameters": {
      "resamplingAlgorithm": "gauss"
    }
  }
}

4. Introduction of the `input` property

We've added a new property called input. This property is used in place of the tilingGrid property that has been since removed. The input property is used to specify the input format, which can be either a tiling grid or a GeoPackage. An example of how to specify a tiling grid input is as follows:

{
  "input": {
    "type" : "tiling-grid",
    "id": 0,
    "resolution": 60.0
  }
}

and an example of how to specify a GeoPackage input:

{
  "input": {
    "type" : "geopackage",
    "features": {
      "s3": {
        "url": "s3://<my-bucket>/<path-to-geopackage>",
        "iamRoleARN": "<your-IAM-role-ARN>"
      }
    }
  }
}

5. Templating and S3 paths

The path or path template of your request can now only be configured via the output.delivery.s3.url property for both GeoTiff and Zarr output.

This means all other properties for specifying the path have been removed, namely:

bucketName
output.defaultTilePath
zarrOutput.path

Example for specifying S3 paths for output:

{
  "output": {
    "delivery": {
      "s3": {
        "url": "s3://<your-bucket>/<path>",
        "iamRoleARN": "<your-IAM-role-ARN>"
      }
    }
   }
}

6. Removal of several additional properties from request & response schemas

From the /process endpoint, we've removed the following properties from the request schema:

tilingGridId
resolution

The following fields have been removed from all response schemas:

valueEstimate
tileCount
tileWidthPx
tileHeightPx
userId
tilingGridId
resolution

Instead of using bucket policies, Batch Processing V2 now supports two alternative approaches for accessing AWS buckets, similarly to Batch Statistical API. These approaches are detailed in the AssumeRole Workflow and AWS Access Key & Secret Key Workflow sections of.

Action Required

When migrating to Batch Processing V2, ensure that you update your AWS bucket access mechanisms to align with one of the approaches mentioned above.