Export Formats

The openEO API offers various options for exporting results in different formats. It could differ depending on the backend capabilities and the type of data being exported.

An openEO workflow remains a process graph until it is executed. A detailed explanation of how to execute openEO jobs can be found in the Job Execution section. On this page, we focus on exploring the available export formats and saving results in different formats.

Common choices are GeoTIFF (GTiff) for raster images, NetCDF (netCDF) or Zarr for multidimensional scientific data, PNG for images, and JSON, CSV, GeoJSON, or Parquet for tables and vector results.

To explore the available export formats, you can use the list_file_formats() method provided by the openEO client.

Python

R

library(openeo)
con <- connect("your_backend_url")
formats <- con$list_file_formats()
print(names(formats$output))

JavaScript

const con = await openeo.connect("your_backend_url");
const formats = await con.listFileFormats();
console.log(Object.keys(formats.output));

The result identifies the format names accepted by save_result and may list format-specific options. A workflow can fail if the selected format cannot represent the result, for example when exporting a multidimensional raster to an image format.

Export a raster data cube

GeoTIFF is a common choice for a spatial raster result that you want to open in desktop GIS software. For a time series or a data cube with several dimensions, NetCDF or Zarr is often a better fit because it preserves multidimensional data. The following examples add a GeoTIFF result node to an existing cube and run a batch job.

Python

result = cube.save_result(format="GTiff")
job = result.create_job(title="GeoTIFF export")
job.start_and_wait()
job.get_results().download_files("./output")

R

result <- p$save_result(data = cube, format = "GTiff")
job <- create_job(graph = result, title = "GeoTIFF export")
start_job(job = job)

# Download after the job has finished.
download_results(job = job, folder = "./output")

JavaScript

const result = builder.save_result(cube, "GTiff");
const job = await con.createJob(result, "GeoTIFF export");
await job.startJob();

// Obtain the result download URLs after the job has finished.
const results = await job.listResults();
console.log(results);

For a very small result, the Python client also supports a direct download:

cube.download("./output/result.tif", format="GTiff")

Use batch jobs for larger areas, long time ranges, or workflows that may take more than a few minutes.

Export tables and vector data

Processes such as aggregate_spatial usually produce a table-like or vector result rather than a raster. JSON is a portable choice for inspecting a small result. CSV is useful for tabular analysis, while GeoJSON and Parquet are common vector-data formats when supported by the backend.

Python

result = aggregated_cube.save_result(format="JSON")
job = result.create_job(title="JSON export")
job.start_and_wait()
job.get_results().download_files("./output")

R

result <- p$save_result(data = aggregated_cube, format = "JSON")
job <- create_job(graph = result, title = "JSON export")
start_job(job = job)
download_results(job = job, folder = "./output")

JavaScript

const result = builder.save_result(aggregatedCube, "JSON");
const job = await con.createJob(result, "JSON export");
await job.startJob();

Replace aggregated_cube or aggregatedCube with the variable holding your result. Use a format listed by the backend; not every backend provides every format for every type of result.

Use format options carefully

Some formats accept options, such as image rendering settings or compression. Options are backend-specific, so always consult list_file_formats() and the backend documentation. The following PNG example assigns cube bands to red, green, and blue channels. It only works when the selected backend supports these options and the cube contains suitable bands.

Python

result = rgb_cube.save_result(
        format="PNG",
        options={"red": "B04", "green": "B03", "blue": "B02"},
)

R

result <- p$save_result(
    data = rgb_cube,
    format = "PNG",
    options = list(red = "B04", green = "B03", blue = "B02")
)

JavaScript

const result = builder.save_result(rgbCube, "PNG", {
 red: "B04",
 green: "B03",
 blue: "B02"
});

PNG is for a display-ready image. It is generally not appropriate when you need the original measurement values for later analysis.

Export a process graph

Exporting a process graph as JSON lets you inspect, archive, share, or reuse a workflow in another compatible openEO tool. Export the final graph object, such as the result returned by save_result, rather than an internal graph representation.

Python

from pathlib import Path

process_json = result.to_json()
Path("process-graph.json").write_text(process_json, encoding="utf-8")

R

library(jsonlite)

process_json <- toJSON(as(result, "Process"), auto_unbox = TRUE, pretty = TRUE)
writeLines(process_json, "process-graph.json")

JavaScript

const processJson = JSON.stringify(result.toJSON(), null, 2);
console.log(processJson);