Skip to content

Workflows and jobs

Apply analytics on geospatial data by using workflows with data blocks and processing blocks.

Step 1. Create and populate a workflow

A workflow is a sequence of data blocks and processing blocks. It defines an order for operations.

A workflow starts with a data block and is usually followed by one or more processing blocks.

A workflow graph

  1. Create a workflow and populate it with blocks, for example:
    project = up42.initialize_project()
    workflow = project.create_workflow(name="My workflow with Processing from Storage")
      "Processing from Storage"
  2. If needed, search for compatible blocks:
    The search outputs a dataframe with blocks that can be added to your workflow, for example:
      "Sharpening Filter"

Step 2. Retrieve an input schema

Retrieve input parameters of the first block in a workflow:

For example, Processing from Storage requires asset_ids as input parameters:
  "up42-processing-from-storage:1": {
    "asset_ids": {
      "type": "array",
      "default": "None",
      "required": true

Step 3. Configure an AOI

If you use storage data as a data source, skip this step.

# geometry = up42.read_vector_file("data/aoi_washington.geojson")
# geometry = up42.get_example_aoi()
geometry = {
    "type": "Polygon",
    "coordinates": (
          (13.375966, 52.515068),
          (13.375966, 52.516639),
          (13.378314, 52.516639),
          (13.378314, 52.515068),
          (13.375966, 52.515068),

Step 4. Adjust JSON parameters

Define how data retrieval and analysis will be performed during a job run, for example:

input_parameters = workflow.construct_parameters(
input_parameters["esa-s2-l2a-gtiff-visual:1"].update({"max_cloud_cover": 5})

Step 5. Make a test query

A test query doesn't consume credits. It searches for available data based on your parameters and provides JSON metadata and low resolution quicklooks of available images.

workflow.test_job(input_parameters, track_status=True)

Step 6. Get a cost estimate

Get a cost estimation before running a live job:


Step 7. Run a job

job = workflow.run_job(input_parameters, track_status=True)

Step 8. Get results

You can download and visualize the results: