openEO API (1.1.0)

Download OpenAPI specification:Download

openEO Consortium: openeo.psc@uni-muenster.de URL: https://openeo.org License: Apache 2.0

The openEO API specification for interoperable cloud-based processing of large Earth observation datasets.

API Principles

Language

In the specification the key words “MUST”, “MUST NOT”, “REQUIRED”, “SHALL”, “SHALL NOT”, “SHOULD”, “SHOULD NOT”, “RECOMMENDED”, “MAY”, and “OPTIONAL” in this document are to be interpreted as described in RFC 2119 and RFC 8174.

Casing

Unless otherwise stated the API works case sensitive.

All names SHOULD be written in snake case, i.e. words are separated with one underscore character (_) and no spaces, with all letters lower-cased. Example: hello_world. This applies particularly to endpoints and JSON property names. HTTP header fields follow their respective casing conventions, e.g. Content-Type or OpenEO-Costs, despite being case-insensitive according to RFC 7230.

HTTP / REST

This uses HTTP REST Level 2 for communication between client and back-end server.

Public APIs MUST be available via HTTPS only.

Endpoints are made use meaningful HTTP verbs (e.g. GET, POST, PUT, PATCH, DELETE) whenever technically possible. If there is a need to transfer big chunks of data for a GET requests to the back-end, POST requests MAY be used as a replacement as they support to send data via request body. Unless otherwise stated, PATCH requests are only defined to work on direct (first-level) children of the full JSON object. Therefore, changing a property on a deeper level of the full JSON object always requires to send the whole JSON object defined by the first-level property.

Naming of endpoints follow the REST principles. Therefore, endpoints are centered around resources. Resource identifiers MUST be named with a noun in plural form except for single actions that can not be modelled with the regular HTTP verbs. Single actions MUST be single endpoints with a single HTTP verb (POST is RECOMMENDED) and no other endpoints beneath it.

JSON

The API uses JSON for request and response bodies whenever feasible. Services use JSON as the default encoding. Other encodings can be requested using Content Negotiation. Clients and servers MUST NOT rely on the order in which properties appears in JSON. Collections usually don't include nested JSON objects if those information can be requested from the individual resources.

Web Linking

The API is designed in a way that to most entities (e.g. collections and processes) a set of links can be added. These can be alternate representations, e.g. data discovery via OGC WCS or OGC CSW, references to a license, references to actual raw data for downloading, detailed information about pre-processing and more. Clients should allow users to follow the links.

Whenever links are utilized in the API, the description explains which relation (rel property) types are commonly used. A list of standardized link relations types is provided by IANA and the API tries to align whenever feasible.

Some very common relation types - usually not mentioned explicitly in the description of links fields - are:

  1. self: which allows link to the location that the resource can be (permanently) found online.This is particularly useful when the data is data is made available offline, so that the downstream user knows where the data has come from.

  2. alternate: An alternative representation of the resource, may it be another metadata standard the data is available in or simply a human-readable version in HTML or PDF.

  3. about: A resource that is related or further explains the resource, e.g. a user guide.

Error Handling

The success of requests MUST be indicated using HTTP status codes according to RFC 7231.

If the API responds with a status code between 100 and 399 the back-end indicates that the request has been handled successfully.

In general an error is communicated with a status code between 400 and 599. Client errors are defined as a client passing invalid data to the service and the service correctly rejecting that data. Examples include invalid credentials, incorrect parameters, unknown versions, or similar. These are generally "4xx" HTTP error codes and are the result of a client passing incorrect or invalid data. Client errors do not contribute to overall API availability.

Server errors are defined as the server failing to correctly return in response to a valid client request. These are generally "5xx" HTTP error codes. Server errors do contribute to the overall API availability. Calls that fail due to rate limiting or quota failures MUST NOT count as server errors.

JSON error object

A JSON error object SHOULD be sent with all responses that have a status code between 400 and 599.

{
  "id": "936DA01F-9ABD-4D9D-80C7-02AF85C822A8",
  "code": "SampleError",
  "message": "A sample error message.",
  "url": "https://example.openeo.org/docs/errors/SampleError"
}

Sending code and message is REQUIRED.

  • A back-end MAY add a free-form id (unique identifier) to the error response to be able to log and track errors with further non-disclosable details.

  • The code is either one of the standardized textual openEO error codes or a proprietary error code.

  • The message explains the reason the server is rejecting the request. For "4xx" error codes the message explains how the client needs to modify the request.

    By default the message MUST be sent in English language. Content Negotiation is used to localize the error messages: If an Accept-Language header is sent by the client and a translation is available, the message should be translated accordingly and the Content-Language header must be present in the response. See "How to localize your API" for more information.

  • url is an OPTIONAL attribute and contains a link to a resource that is explaining the error and potential solutions in-depth.

Standardized status codes

The openEO API usually uses the following HTTP status codes for successful requests:

  • 200 OK: Indicates a successful request with a response body being sent.
  • 201 Created Indicates a successful request that successfully created a new resource. Sends a Location header to the newly created resource without a response body.
  • 202 Accepted Indicates a successful request that successfully queued the creation of a new resource, but it has not been created yet. The response is sent without a response body.
  • 204 No Content: Indicates a successful request without a response body being sent.

The openEO API has some commonly used HTTP status codes for failed requests:

  • 400 Bad Request: The back-end responds with this error code whenever the error has its origin on client side and no other HTTP status code in the 400 range is suitable.

  • 401 Unauthorized: The client did not provide any authentication details for a resource requiring authentication or the provided authentication details are not correct.

  • 403 Forbidden: The client did provided correct authentication details, but the privileges/permissions of the provided credentials do not allow to request the resource.

  • 404 Not Found: The resource specified by the path does not exist, i.e. one of the resources belonging to the specified identifiers are not available at the back-end. Note: Unsupported endpoints MUST use HTTP status code 501.

  • 500 Internal Server Error: The error has its origin on server side and no other status code in the 500 range is suitable.

If a HTTP status code in the 400 range is returned, the client SHOULD NOT repeat the request without modifications. For HTTP status code in the 500 range, the client MAY repeat the same request later.

All HTTP status codes defined in RFC 7231 in the 400 and 500 ranges can be used as openEO error code in addition to the most used status codes mentioned here. Responding with openEO error codes 400 and 500 SHOULD be avoided in favor of any more specific standardized or proprietary openEO error code.

Temporal data

Date, time, intervals and durations are formatted based on ISO 8601 or its profile RFC 3339 whenever there is an appropriate encoding available in the standard. All temporal data are specified based on the Gregorian calendar.

Authentication

The openEO API offers two forms of authentication by default:

  • OpenID Connect (recommended) at GET /credentials/oidc
  • Basic at GET /credentials/basic

After authentication with any of the methods listed above, the tokens obtained during the authentication workflows can be sent to protected endpoints in subsequent requests.

Further authentication methods MAY be added by back-ends.

Bearer

Security Scheme Type: HTTP
HTTP Authorization Scheme: bearer
Bearer format: The Bearer Token MUST consist of the authentication method, a provider ID (if available) and the token itself. All separated by a forward slash `/`. Examples (replace `TOKEN` with the actual access token): (1) Basic authentication (no provider ID available): `basic//TOKEN` (2) OpenID Connect (provider ID is `ms`): `oidc/ms/TOKEN`. For OpenID Connect, the provider ID corresponds to the value specified for `id` for each provider in `GET /credentials/oidc`.

Basic

Security Scheme Type: HTTP
HTTP Authorization Scheme: basic

Cross-Origin Resource Sharing (CORS)

Cross-origin resource sharing (CORS) is a mechanism that allows restricted resources [...] on a web page to be requested from another domain outside the domain from which the first resource was served. [...] CORS defines a way in which a browser and server can interact to determine whether or not it is safe to allow the cross-origin request. It allows for more freedom and functionality than purely same-origin requests, but is more secure than simply allowing all cross-origin requests.

Source: https://en.wikipedia.org/wiki/Cross-origin_resource_sharing

openEO-based back-ends are usually hosted on a different domain / host than the client that is requesting data from the back-end. Therefore most requests to the back-end are blocked by all modern browsers. This leads to the problem that the JavaScript library and any browser-based application can't access back-ends. Therefore, all back-end providers SHOULD support CORS to enable browser-based applications to access back-ends. CORS is a recommendation of the W3C organization. The following chapters will explain how back-end providers can implement CORS support.

Tip: Most servers can send the required headers and the responses to the OPTIONS requests automatically for all endpoints. Otherwise you may also use a proxy server to add the headers and OPTIONS responses.

CORS headers

The following headers MUST be included with every response:

Name Description Example
Access-Control-Allow-Origin Allowed origin for the request, including protocol, host and port or * for all origins. It is RECOMMENDED to return the value * to allow requests from browser-based implementations such as the Web Editor. *
Access-Control-Expose-Headers Some endpoints require to send additional HTTP response headers such as OpenEO-Identifier and Location. To make these headers available to browser-based clients, they MUST be white-listed with this CORS header. The following HTTP headers are white-listed by browsers and MUST NOT be included: Cache-Control, Content-Language, Content-Length, Content-Type, Expires, Last-Modified and Pragma. At least the following headers MUST be listed in this version of the openEO API: Link, Location, OpenEO-Costs and OpenEO-Identifier. Link, Location, OpenEO-Costs, OpenEO-Identifier

Example request and response

Request:

POST /api/v1/jobs HTTP/1.1
Host: openeo.cloudprovider.com
Origin: https://client.org:8080
Authorization: Bearer basic//ZXhhbXBsZTpleGFtcGxl

Response:

HTTP/1.1 201 Created
Access-Control-Allow-Origin: *
Access-Control-Expose-Headers: Location, OpenEO-Identifier, OpenEO-Costs, Link
Content-Type: application/json
Location: https://openeo.cloudprovider.com/api/v1/jobs/abc123
OpenEO-Identifier: abc123

OPTIONS method

All endpoints must respond to the OPTIONS HTTP method. This is a response for the preflight requests made by web browsers before sending the actual request (e.g. POST /jobs). It needs to respond with a status code of 204 and no response body. In addition to the HTTP headers shown in the table above, the following HTTP headers MUST be included with every response to an OPTIONS request:

Name Description Example
Access-Control-Allow-Headers Comma-separated list of HTTP headers allowed to be sent with the actual (non-preflight) request. MUST contain at least Authorization if any kind of authorization is implemented by the back-end. Authorization, Content-Type
Access-Control-Allow-Methods Comma-separated list of HTTP methods allowed to be requested. Back-ends MUST list all implemented HTTP methods for the endpoint. OPTIONS, GET, POST, PATCH, PUT, DELETE
Content-Type SHOULD return the content type delivered by the request that the permission is requested for. application/json

Example request and response

Request:

OPTIONS /api/v1/jobs HTTP/1.1
Host: openeo.cloudprovider.com
Origin: https://client.org:8080
Access-Control-Request-Method: POST 
Access-Control-Request-Headers: Authorization, Content-Type

Note that the Access-Control-Request-* headers are automatically attached to the requests by the browsers.

Response:

HTTP/1.1 204 No Content
Access-Control-Allow-Origin: *
Access-Control-Allow-Methods: OPTIONS, GET, POST, PATCH, PUT, DELETE
Access-Control-Allow-Headers: Authorization, Content-Type
Access-Control-Expose-Headers: Location, OpenEO-Identifier, OpenEO-Costs, Link
Content-Type: application/json

Processes

A process is an operation that performs a specific task on a set of parameters and returns a result. An example is computing a statistical operation, such as mean or median, on selected EO data. A process is similar to a function or method in programming languages. In openEO, processes are used to build a chain of processes (process graph), which can be applied to EO data to derive your own findings from the data.

A pre-defined process is a process provided by the back-end. There is a set of predefined processes by openEO to improve interoperability between back-ends. Back-ends SHOULD follow these specifications whenever possible. Not all processes need to be implemented by all back-ends. See the process reference for pre-defined processes.

A user-defined process is a process defined by the user. It can directly be part of another process graph or be stored as custom process on a back-end. Internally it is a process graph with optional additional metadata.

A process graph chains specific process calls from the set of pre-defined and user-defined processes together. A process graph itself can be stored as a (user-defined) process again. Similarly to scripts in the context of programming, process graphs organize and automate the execution of one or more processes that could alternatively be executed individually. In a process graph, processes need to be specific, i.e. concrete values or "placeholders" for input parameters need to be specified. These values can be scalars, arrays, objects, references to parameters or previous computations or other process graphs.

Defining Processes

Back-ends and users MAY define new proprietary processes for their domain.

Back-end providers MUST follow the schema for predefined processes as in GET /processes to define new processes. This includes:

  • Choosing a intuitive process id, consisting of only letters (a-z), numbers and underscores. It MUST be unique across the pre-defined processes.
  • Defining the parameters and their exact (JSON) schemes.
  • Specifying the return value of a process also with a (JSON) schema.
  • Providing examples or compliance tests.
  • Trying to make the process universally usable so that other back-end providers or openEO can adopt it.

Users MUST follow the schema for user-defined processes as in GET /process_graphs/{process_graph_id} to define new processes. This includes:

  • Choosing a intuitive name as process id, consisting of only letters (a-z), numbers and underscores. It MUST be unique across the user-defined processes.
  • Defining the algorithm as a process graph.
  • Optionally, specifying the additional metadata for processes.

If new process are potentially useful for other back-ends the openEO consortium is happily accepting pull requests to include them in the list of pre-defined processes.

Schemas

Each process parameter and the return values of a process define a schema that the value MUST comply to. The schemas are based on JSON Schema draft-07.

Three custom keywords have been defined:

  • subtype for more fine-grained data-types than JSON Schema supports.
  • parameters to specify the parameters of a process graph (object with subtype process-graph).
  • returns to describe the return value of a process graph (object with subtype process-graph).

Subtypes

JSON Schema allows to specify only a small set of native data types (string, boolean, number, integer, array, object, null). To support more fine grained data types, a custom JSON Schema keyword has been defined: subtype. It works similarly as the JSON Schema keyword format and standardizes a number of openEO related data types that extend the native data types, for example: bounding-box (object with at least west, south, east and north properties), date-time (string representation of date and time following RFC 3339), raster-cube (the type of data cubes), etc. The subtypes should be re-used in process schema definitions whenever suitable.

If a general data type such as string or number is used in a schema, all subtypes with the same parent data type can be passed, too. Clients should offer make passing subtypes as easy as passing a general data type. For example, a parameter accepting strings must also allow passing a string with subtype date and thus clients should encourage this by also providing a date-picker.

A list of predefined subtypes is available as JSON Schema in openeo-processes.

Process Graphs

As defined above, a process graph is a chain of processes with explicit values for their parameters. Technically, a process graph is defined to be a graph of connected processes with exactly one node returning the final result:

<ProcessGraph> := {
  "<ProcessNodeIdentifier>": <ProcessNode>,
  ...
}

<ProcessNodeIdentifier> is a unique key within the process graph that is used to reference (the return value of) this process in arguments of other processes. The identifier is unique only strictly within itself, excluding any parent and child process graphs. Process node identifiers are also strictly scoped and can not be referenced from child or parent process graphs. Circular references are not allowed.

Note: We provide a non-binding JSON Schema for basic process graph validation.

Processes (Process Nodes)

A single node in a process graph (i.e. a specific instance of a process) is defined as follows:

<ProcessNode> := {
  "process_id": <string>,
  "namespace": <string> / null,
  "description": <string>,
  "arguments": <Arguments>,
  "result": true / false
}

A process node MUST always contain key-value-pairs named process_id and arguments. It MAY contain a description.

One of the nodes in a map of processes (the final one) MUST have the result flag set to true, all the other nodes can omit it as the default value is false. Having such a node is important as multiple end nodes are possible, but in most use cases it is important to exactly specify the return value to be used by other processes. Each child process graph must also specify a result node similar to the "main" process graph.

process_id MUST be a valid process ID in the namespace given. Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

Arguments

A process can have an arbitrary number of arguments. Their name and value are specified in the process specification as an object of key-value pairs:

<Arguments> := {
  "<ParameterName>": <string|number|boolean|null|array|object|ResultReference|UserDefinedProcess|ParameterReference>
}

Notes:

  • The specified data types are the native data types supported by JSON, except for ResultReference, UserDefinedProcess and ParameterReference.

  • Objects are not allowed to have keys with the following reserved names:

    • from_node, except for objects of type ResultReference
    • process_graph, except for objects of type UserDefinedProcess
    • from_parameter, except for objects of type ParameterReference
  • Arrays and objects can also contain a ResultReference, a UserDefinedProcess or a ParameterReference. So back-ends must fully traverse the process graphs, including all children.

Accessing results of other process nodes

A value of type <ResultReference> is an object with a key from_node and a <ProcessNodeIdentifier> as corresponding value:

<ResultReference> := {
  "from_node": "<ProcessNodeIdentifier>"
}

This tells the back-end that the process expects the result (i.e. the return value) from another process node to be passed as argument. The <ProcessNodeIdentifier> is strictly scoped and can only reference nodes from within the same process graph, not child or parent process graphs.

User-defined process

A user-defined process in a process graph is a child process graph, to be evaluated as part of another process.

Example: You want to calculate the absolute value of each pixel in a data cube. This can be achieved in openEO by executing the apply process and pass it a user-defined process as the "operator" to apply to each pixel. In this simple example, the "child" process graph defining the user-defined process consists of a single process absolute, but it can be arbitrary complex in general.

A <UserDefinedProcess> argument MUST at least consist of an object with a key process_graph. Optionally, it can also be described with the same additional properties available for pre-defined processes such as an id, parameters, return values etc. When embedded in a process graph, these additional properties of a user-defined process are usually not used, except for validation purposes.

<UserDefinedProcess> := {
  "process_graph": <ProcessGraph>,
  ...
}

Accessing process parameters

A "parent" process that works with a user-defined process can make so called process graph parameters available to the "child" logic. Processes in the "child" process graph can access these parameters by passing a ParameterReference object as argument. It is an object with key from_parameter specifying the name of the process graph parameter:

<ParameterReference> := {
  "from_parameter": "<ParameterReferenceName>"
}

The parameter names made available for <ParameterReferenceName> are defined and passed to the process graph by one of the parent entities. The parent could be a process (such as apply or reduce_dimension) or something else that executes a process graph (a secondary web service for example). If the parent is a process, the parameter are defined in the parameters property of the corresponding JSON Schema.

In case of the example given above, the parameter process in the process apply defines two process graph parameters: x (the value of each pixel that will be processed) and context (additional data passed through from the user). The process absolute expects an argument with the same name x. The process graph for the example would look as follows:

{
  "process_id": "apply",
  "arguments": {
    "data": {"from_node": "loadcollection1"}
    "process": {
      "process_graph": {
        "abs1": {
          "process_id": "absolute",
          "arguments": {
            "x": {"from_parameter": "x"}
          },
          "result": true
        }
      }
    }
  }
}

loadcollection1 would be a result from another process, which is not part of this example.

Important: <ParameterReferenceName> is less strictly scoped than <ProcessNodeIdentifier>. <ParameterReferenceName> can be any parameter from the process graph or any of its parents.

The value for the parameter MUST be resolved as follows:

  1. In general the most specific parameter value is used. This means the parameter value is resolved starting from the current scope and then checking each parent for a suitable parameter value until a parameter values is found or the "root" process graph has been reached.
  2. In case a parameter value is not available, the most unspecific default value from the process graph parameter definitions are used. For example, if default values are available for the root process graph and all children, the default value from the root process graph is used.
  3. If no default values are available either, the error ProcessParameterMissing must be thrown.

Full example for an EVI computation

Deriving minimum EVI (Enhanced Vegetation Index) measurements over pixel time series of Sentinel 2 imagery. The main process graph in blue, child process graphs in yellow:

Graph with processing instructions

The process graph for the algorithm: pg-evi-example.json

Data Processing

Processes can run in three different ways:

  1. Results can be pre-computed by creating a batch job. They are submitted to the back-end's processing system, but will remain inactive until explicitly put into the processing queue. They will run only once and store results after execution. Results can be downloaded. Batch jobs are typically time consuming and user interaction is not possible although log files are generated for them. This is the only mode that allows to get an estimate about time, volume and costs beforehand.

  2. A more dynamic way of processing and accessing data is to create a secondary web service. They allow web-based access using different protocols such as OGC WMS, OGC WCS, OGC API - Features or XYZ tiles. Some protocols such as the OGC WMS or XYZ tiles allow users to change the viewing extent or level of detail (zoom level). Therefore, computations often run on demand so that the requested data is calculated during the request. Back-ends should make sure to cache processed data to avoid additional/high costs and reduce waiting times for the user.

  3. Processes can also be executed on-demand (i.e. synchronously). Results are delivered with the request itself and no job is created. Only lightweight computations, for example previews, should be executed using this approach as timeouts are to be expected for long-polling HTTP requests.

Validation

Process graph validation is a quite complex task. There's a JSON schema for basic process graph validation. It checks the general structure of a process graph, but only checking against the schema is not fully validating a process graph. Note that this JSON Schema is probably good enough for a first version, but should be revised and improved for production. There are further steps to do:

  1. Validate whether there's exactly one result: true per process graph.
  2. Check whether the process names that are referenced in the field process_id are actually available in the corresponding namespace.
  3. Validate all arguments for each process against the JSON schemas that are specified in the corresponding process specifications.
  4. Check whether the values specified for from_node have a corresponding node in the same process graph.
  5. Validate whether the return value and the arguments requesting a return value with from_node are compatible.
  6. Check the content of arrays and objects. These could include parameter and result references (from_node, from_parameter etc.).

Execution

To process the process graph on the back-end you need to go through all nodes/processes in the list and set for each node to which node it passes data and from which it expects data. In another iteration the back-end can find all start nodes for processing by checking for zero dependencies.

You can now start and execute the start nodes (in parallel, if possible). Results can be passed to the nodes that were identified beforehand. For each node that depends on multiple inputs you need to check whether all dependencies have already finished and only execute once the last dependency is ready.

Please be aware that the result node (result set to true) is not necessarily the last node that is executed. The author of the process graph may choose to set a non-end node to the result node!

Capabilities

General information about the API implementation and other supported capabilities at the back-end.

Information about the back-end

Lists general information about the back-end, including which version and endpoints of the openEO API are supported. May also include billing information.

Authorizations:

Responses

Response Schema: application/json
api_version
required
string
Value: "1.1.0"

Version number of the openEO specification this back-end implements.

backend_version
required
string

Version number of the back-end implementation. Every change on back-end side MUST cause a change of the version number.

stac_version
required
string (stac_version) ^(0\.9.\d+|1\.\d+.\d+)

The version of the STAC specification, which MAY not be equal to the STAC API version. Supports versions 0.9.x and 1.x.x.

type
string
Value: "Catalog"

For STAC versions >= 1.0.0-rc.1 this field is required.

id
required
string

Identifier for the service. This field originates from STAC and is used as unique identifier for the STAC catalog available at /collections.

title
required
string

The name of the service.

description
required
string <commonmark>

A description of the service, which allows the service provider to introduce the user to its service. CommonMark 0.29 syntax MAY be used for rich text representation.

production
boolean (production)
Default: false

Specifies whether the implementation is ready to be used in production use (true) or not (false). Clients SHOULD only connect to non-production implementations if the user explicitly confirmed to use a non-production implementation. This flag is part of GET /.well-known/openeo and GET /. It MUST be used consistently in both endpoints.

required
Array of objects (Endpoint)

Lists all supported endpoints. Supported are all endpoints, which are implemented, return a 2XX or 3XX HTTP status code and are fully compatible to the API specification. An entry for this endpoint (path / with method GET) SHOULD NOT be listed.

object (Billing)

Billing related data, e.g. the currency used or available plans to process jobs. This property MUST be specified if the back-end uses any billing related API functionalities, e.g. budgeting or estimates. The absence of this property doesn't mean the back-end is necessarily free to use for all. Providers may choose to bill users outside of the API, e.g. with a monthly fee that is not depending on individual API interactions.

required
Array of objects (Link)

Links related to this service, e.g. the homepage of the service provider or the terms of service.

It is highly RECOMMENDED to provide links with the following rel (relation) types:

  1. version-history: A link back to the Well-Known URL (see /.well-known/openeo) to allow clients to work on the most recent version.

  2. terms-of-service: A link to the terms of service. If a back-end provides a link to the terms of service, the clients MUST provide a way to read the terms of service and only connect to the back-end after the user agreed to them. The user interface MUST be designed in a way that the terms of service are not agreed to by default, i.e. the user MUST explicitly agree to them.

  3. privacy-policy: A link to the privacy policy (GDPR). If a back-end provides a link to a privacy policy, the clients MUST provide a way to read the privacy policy and only connect to the back-end after the user agreed to them. The user interface MUST be designed in a way that the privacy policy is not agreed to by default, i.e. the user MUST explicitly agree to them.

  4. service-desc or service-doc: A link to the API definition. Use service-desc for machine-readable API definition and service-doc for human-readable API definition. Required if full OGC API compatibility is desired.

  5. conformance: A link to the Conformance declaration (see /conformance). Required if full OGC API compatibility is desired.

  6. data: A link to the collections (see /collections). Required if full OGC API compatibility is desired.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "api_version": "1.1.0",
  • "backend_version": "1.1.2",
  • "stac_version": "string",
  • "type": "Catalog",
  • "id": "cool-eo-cloud",
  • "title": "Cool EO Cloud",
  • "description": "This service is provided to you by [Cool EO Cloud Corp.](http://cool-eo-cloud-corp.com). It implements the full openEO API and allows to process a range of 999 EO data sets, including \n\n* Sentinel 1/2/3 and 5\n* Landsat 7/8\n\nA free plan is available to test the service. For further information please contact our customer service at [support@cool-eo-cloud-corp.com](mailto:support@cool-eo-cloud-corp.com).",
  • "production": false,
  • "endpoints": [
    ],
  • "billing": {},
  • "links": []
}

Supported openEO versions

Lists all implemented openEO versions supported by the service provider. This endpoint is the Well-Known URI (see RFC 5785) for openEO.

This allows a client to easily identify the most recent openEO implementation it supports. By default, a client SHOULD connect to the most recent production-ready version it supports. If not available, the most recent supported version of all versions SHOULD be connected to. Clients MAY let users choose to connect to versions that are not production-ready or outdated. The most recent version is determined by comparing the version numbers according to rules from Semantic Versioning, especially §11. Any pair of API versions in this list MUST NOT be equal according to Semantic Versioning.

The Well-Known URI is the entry point for clients and users, so make sure it is permanent and easy to use and remember. Clients MUST NOT require the well-known path (./well-known/openeo) in the URL that is specified by a user to connect to the back-end. A client MUST request https://example.com/.well-known/openeo if a user tries to connect to https://example.com. If the request to the well-known URI fails, the client SHOULD try to request the capabilities at / from https://example.com.

This URI MUST NOT be versioned as the other endpoints. If your API is available at https://example.com/api/v1.0, the Well-Known URI SHOULD be located at https://example.com/.well-known/openeo and the URI users connect to SHOULD be https://example.com.

Clients MAY get additional information (e.g. title or description) about a back-end from the most recent version that has the production flag set to true.

Authorizations:

Responses

Response Schema: application/json
required
Array of objects (API Instance)

Response samples

Content type
application/json
{}

Supported file formats

Lists supported input and output file formats. Input file formats specify which file a back-end can read from. Output file formats specify which file a back-end can write to.

The response to this request is an object listing all available input and output file formats separately with their parameters and additional data. This endpoint does not include the supported secondary web services.

Note: Format names and parameters MUST be fully aligned with the GDAL codes if available, see GDAL Raster Formats and OGR Vector Formats. It is OPTIONAL to support all output format parameters supported by GDAL. Some file formats not available through GDAL may be defined centrally for openEO. Custom file formats or parameters MAY be defined.

The format descriptions must describe how the file formats relate to data cubes. Input file formats must describe how the files have to be structured to be transformed into data cubes. Output file formats must describe how the data cubes are stored at the back-end and how the resulting file structure looks like.

Back-ends MUST NOT support aliases, for example it is not allowed to support geotiff instead of gtiff. Nevertheless, openEO Clients MAY translate user input input for convenience (e.g. translate geotiff to gtiff). Also, for a better user experience the back-end can specify a title.

Format names MUST be accepted in a case insensitive manner throughout the API.

Authorizations:
Bearer

Responses

Response Schema: application/json
required
object (Input File Formats)

Map of supported input file formats, i.e. file formats a back-end can read from. The property keys are the file format names that are used by clients and users, for example in process graphs.

required
object (Output File Formats)

Map of supported output file formats, i.e. file formats a back-end can write to. The property keys are the file format names that are used by clients and users, for example in process graphs.

Response samples

Content type
application/json
{
  • "output": {
    },
  • "input": {
    }
}

OGC specifications this API conforms to

Lists all conformance classes specified in OGC standards that the server conforms to. This endpoint is only required if full compliance with OGC API standards is desired. Therefore, openEO back-ends MAY implement it for compatibility with OGC API clients and openEO clients don't need to request it.

Responses

Response Schema: application/json
conformsTo
required
Array of strings <uri>

Response samples

Content type
application/json

Supported UDF runtimes

Lists the supported runtimes for user-defined functions (UDFs), which includes either the programming languages including version numbers and available libraries including version numbers or docker containers.

Authorizations:
Bearer

Responses

Response Schema: application/json
additional property
object

Response samples

Content type
application/json
{}

Supported secondary web service protocols

Lists supported secondary web service protocols such as OGC WMS, OGC WCS, OGC API - Features or XYZ tiles. The response is an object of all available secondary web service protocols with their supported configuration settings and expected process parameters.

  • The configuration settings for the service SHOULD be defined upon creation of a service and the service will be set up accordingly.
  • The process parameters SHOULD be referenced (with a from_parameter reference) in the user-defined process that is used to compute web service results. The appropriate arguments MUST be provided to the user-defined process, usually at runtime from the context of the web service, For example, a map service such as a WMS would need to inject the spatial extent into the user-defined process so that the back-end can compute the corresponding tile correctly.

To improve interoperability between back-ends common names for the services SHOULD be used, e.g. the abbreviations used in the official OGC Schema Repository for the respective services.

Service names MUST be accepted in a case insensitive manner throughout the API.

Authorizations:
Bearer

Responses

Response Schema: application/json
additional property
object (Service Type)

Response samples

Content type
application/json
{
  • "WMS": {
    },
  • "OGCAPI-FEATURES": {}
}

Account Management

The following endpoints handle user profiles, accounting and authentication. See also Authentication. In general, the openEO API only defines a minimum subset of user management and accounting functionality. It allows to

  • authenticate and authorize a user, which may include user registration with OpenID Connect,
  • handle storage space limits (disk quota),
  • manage billing, which includes to
    • query the credit a user has available,
    • estimate costs for certain operations (data processing and downloading),
    • get information about produced costs,
    • limit costs of certain operations.

Therefore, the API leaves some aspects open that have to be handled by the back-ends separately, including

  • credential recovery, e.g. retrieving a forgotten password
  • user data management, e.g. changing the users payment details or email address
  • payments, i.e. topping up credits for pre-paid services or paying for post-paid services
  • other accounting related tasks, e.g. creating invoices,
  • user registration (except for user registration with OpenID Connect).

OpenID Connect authentication

Lists the supported OpenID Connect providers (OP). OpenID Connect Providers MUST support OpenID Connect Discovery.

It is highly RECOMMENDED to implement OpenID Connect for public services in favor of Basic authentication.

openEO clients MUST use the access token as part of the Bearer token for authorization in subsequent API calls (see also the information about Bearer tokens in this document). Clients MUST NOT use the id token or the authorization code. The access token provided by an OpenID Connect Provider does not necessarily provide information about the issuer (i.e. the OpenID Connect provider) and therefore a prefix MUST be added to the Bearer Token sent in subsequent API calls to protected endpoints. The Bearer Token sent to protected endpoints MUST consist of the authentication method (here oidc), the provider ID and the access token itself. All separated by a forward slash /. The provider ID corresponds to the value specified for id for each provider in the response body of this endpoint. The header in subsequent API calls for a provider with id ms would look as follows: Authorization: Bearer oidc/ms/TOKEN (replace TOKEN with the actual access token received from the OpenID Connect Provider).

Back-ends MAY request user information (including Claims) from the OpenID Connect Userinfo endpoint using the access token (without the prefix described above). Therefore, both openEO client and openEO back-end are relying parties (clients) to the OpenID Connect Provider.

Authorizations:

Responses

Response Schema: application/json
required
Array of objects (OpenID Connect Provider) non-empty

The first provider in this list is the default provider for authentication. Clients can either pre-select or directly use the default provider for authentication if the user doesn't specify a specific value.

Response samples

Content type
application/json
{
  • "providers": [
    ]
}

HTTP Basic authentication

Checks the credentials provided through HTTP Basic Authentication according to RFC 7617 and returns an access token for valid credentials.

The credentials (username and password) MUST be sent in the HTTP header Authorization with type Basic and the Base64 encoded string consisting of username and password separated by a double colon :. The header would look as follows for username user and password pw: Authorization: Basic dXNlcjpwdw==.

The access token has to be used in the Bearer token for authorization in subsequent API calls (see also the information about Bearer tokens in this document). The access token returned by this request MUST NOT be provided with basic// prefix, but the Bearer Token sent in subsequent API calls to protected endpoints MUST be prefixed with basic//. The header in subsequent API calls would look as follows: Authorization: Bearer basic//TOKEN (replace TOKEN with the actual access token).

It is RECOMMENDED to implement this authentication method for non-public services only.

Authorizations:
Basic

Responses

Response Schema: application/json
access_token
required
string

The access token (without basic// prefix) to be used in the Bearer token for authorization in subsequent API calls.

Response samples

Content type
application/json
{
  • "access_token": "b34ba2bdf9ac9ee1"
}

Information about the authenticated user

Lists information about the authenticated user, e.g. the user id. The endpoint MAY return the disk quota available to the user. The endpoint MAY also return links related to user management and the user profile, e.g. where payments are handled or the user profile could be edited. For back-ends that involve accounting, this service MAY also return the currently available money or credits in the currency the back-end is working with. This endpoint MAY be extended to fulfil the specification of the OpenID Connect UserInfo Endpoint.

Authorizations:
Bearer

Responses

Response Schema: application/json
user_id
required
string (user_id) ^[\w\-\.~]+$

Unique identifier of the user. MUST match the specified pattern. This is usually a randomly generated internal identifier from the provider not meant for displaying purposes.

name
string

The user name, a human-friendly displayable name. Could be the user's real name or a nickname.

default_plan
string or null

Name of the plan the user has subscribed to.

Overrides the default plan of the back-end, but back-ends MAY also allow overriding this plan for each individual processing request (e.g. job or service) with the corresponding plan property.

object or null (User Storage)

Information about the storage space available to the user.

budget
number or null (budget)
Default: null

Maximum amount of costs the request is allowed to produce. The value MUST be specified in the currency of the back-end. No limits apply, if the value is null or the back-end has no currency set in GET /.

Array of objects (Link)

Links related to the user profile, e.g. where payments are handled or the user profile could be edited.

It is RECOMMENDED to provide links with the following rel (relation) types:

  1. payment: A page where users can recharge their user account with money or credits.

  2. edit-form: Points to a page where the user can edit his user profile.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{}

EO Data Discovery

These endpoints allow to list the collections that are available at the back-end and can be used as data cubes for data processing.

STAC

For data discovery of Earth Observation Collections at the back-ends, openEO strives for compatibility with the specifications SpatioTemporal Asset Catalog (STAC) and OGC API - Features - Part 1: Core as far as possible. Implementing the data discovery endpoints of openEO should also produce valid STAC API 0.9.0 and OGC API - Features 1.0 responses, including (partial) compatibility with their APIs.

The data discovery endpoints GET /collections and GET /collections/{collection_id} are compatible with OGC API - Features and STAC. Both specifications define additional endpoints that need to be implemented to be fully compatible. The additional endpoints can easily be integrated into an openEO API implementation. A rough list of actions for compatibility is available below, but please refer to their specifications to find out the full details.

Important: STAC specification and STAC API are different specifications and have different version numbers after version 0.9.0. The openEO API only implements STAC API version 0.9.0, which allows to serve all STAC specification versions in the range of 0.9.x and 1.x.x (see the stac_version property).

Content Extensions

STAC has several extensions that can be used to better describe your data. Clients and server are not required to implement all of them, so be aware that some clients may not be able to read all your metadata.

Some commonly used extensions that are relevant for datasets exposed through the openEO API are:

Provide data for download

If you'd like to provide your data for download in addition to offering the cloud processing service, you can implement the full STAC API. Therefore you can implement the endpoints GET /collections/{collection_id}/items and GET /collections/{collection_id}/items/{feature_id} to support retrieval of individual items. To benefit from the STAC ecosystem and allow searching for items you can also implement POST /search and GET /search. Further information can be found in the STAC API repository.

Basic metadata for all datasets

Lists available collections with at least the required information.

It is strongly RECOMMENDED to keep the response size small by omitting larger optional values from the objects in collections (e.g. the summaries and cube:dimensions properties). To get the full metadata for a collection clients MUST request GET /collections/{collection_id}.

This endpoint is compatible with STAC API 0.9.0 and OGC API - Features. STAC API features / extensions and STAC extensions can be implemented in addition to what is documented here.

Authorizations:
Bearer
query Parameters
limit
integer >= 1
Example: limit=10

This parameter enables pagination for the endpoint and specifies the maximum number of elements that arrays in the top-level object (e.g. collections, processes, batch jobs, secondary services, log entries, etc.) are allowed to contain. The links array MUST NOT be paginated like the resources, but instead contain links related to the paginated resources or the pagination itself (e.g. a link to the next page). If the parameter is not provided or empty, all elements are returned.

Pagination is OPTIONAL: back-ends or clients may not support it. Therefore it MUST be implemented in a way that clients not supporting pagination get all resources regardless. Back-ends not supporting pagination MUST return all resources.

If the response is paginated, the links array MUST be used to communicate the links for browsing the pagination with pre-defined rel types. See the links array schema for supported rel types. Backend implementations can, unless specified otherwise, use all kind of pagination techniques, depending on what is supported best by their infrastructure: page-based, offset-based, token-based or something else. The clients SHOULD use whatever is specified in the links with the corresponding rel types.

Responses

Response Schema: application/json
required
Array of objects (Collection)
required
Array of objects (links_pagination)

Links related to this list of resources, for example links for pagination or alternative formats such as a human-readable HTML version. The links array MUST NOT be paginated.

If pagination is implemented, the following rel (relation) types apply:

  1. next (REQUIRED): A link to the next page, except on the last page.

  2. prev (OPTIONAL): A link to the previous page, except on the first page.

  3. first (OPTIONAL): A link to the first page, except on the first page.

  4. last (OPTIONAL): A link to the last page, except on the last page.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "collections": [
    ],
  • "links": []
}

Full metadata for a specific dataset

Lists all information about a specific collection specified by the identifier collection_id.

This endpoint is compatible with STAC API 0.9.0 and OGC API - Features. STAC API features / extensions and STAC extensions can be implemented in addition to what is documented here.

Authorizations:
Bearer
path Parameters
collection_id
required
string (collection_id) ^[\w\-\.~\/]+$
Example: Sentinel-2A

Collection identifier

Responses

Response Schema: application/json
stac_version
required
string (stac_version) ^(0\.9.\d+|1\.\d+.\d+)

The version of the STAC specification, which MAY not be equal to the STAC API version. Supports versions 0.9.x and 1.x.x.

Array of Reference to a JSON Schema (string) or Reference to a core extension (STAC < 1.0.0-rc.1 only) (string) (stac_extensions) unique

A list of implemented STAC extensions. The list contains URLs to the JSON Schema files it can be validated against. For STAC < 1.0.0-rc.1 shortcuts such as sar can be used instead of the schema URL.

type
string
Value: "Collection"

For STAC versions >= 1.0.0-rc.1 this field is required.

id
required
string (collection_id) ^[\w\-\.~\/]+$

A unique identifier for the collection, which MUST match the specified pattern.

title
string

A short descriptive one-line title for the collection.

description
required
string <commonmark>

Detailed multi-line description to explain the collection.

CommonMark 0.29 syntax MAY be used for rich text representation.

keywords
Array of strings

List of keywords describing the collection.

version
string

Version of the collection.

This property REQUIRES to add version (STAC < 1.0.0-rc.1) or https://stac-extensions.github.io/version/v1.0.0/schema.json (STAC >= 1.0.0-rc.1) to the list of stac_extensions.

deprecated
boolean
Default: false

Specifies that the collection is deprecated with the potential to be removed. It should be transitioned out of usage as soon as possible and users should refrain from using it in new projects.

A link with relation type latest-version SHOULD be added to the links and MUST refer to the collection that can be used instead.

This property REQUIRES to add version (STAC < 1.0.0-rc.1) or https://stac-extensions.github.io/version/v1.0.0/schema.json (STAC >= 1.0.0-rc.1) to the list of stac_extensions.

license
required
string (stac_license)

License(s) of the data as a SPDX License identifier. Alternatively, use proprietary if the license is not on the SPDX license list or various if multiple licenses apply. In these two cases links to the license texts SHOULD be added, see the license link relation type.

Non-SPDX licenses SHOULD add a link to the license text with the license relation in the links section. The license text MUST NOT be provided as a value of this field. If there is no public license URL available, it is RECOMMENDED to host the license text and link to it.

Array of objects (stac_providers)

A list of providers, which MAY include all organizations capturing or processing the data or the hosting provider. Providers SHOULD be listed in chronological order with the most recent provider being the last element of the list.

required
object (Collection Extent)

The extent of the data in the collection. Additional members MAY be added to represent other extents, for example, thermal or pressure ranges.

The first item in the array always describes the overall extent of the data. All subsequent items describe more preciseextents, e.g. to identify clusters of data. Clients only interested in the overall extent will only need to access the first item in each array.

required
Array of objects (Link)

Links related to this collection. Could reference to licensing information, other meta data formats with additional information or a preview image. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. root and parent: URL to the data discovery endpoint at /collections.
  2. license: A link to the license(s) SHOULD be specified if the license field is set to proprietary or various.
  3. example: Links to examples of processes that use this collection.
  4. latest-version: If a collection has been marked as deprecated, a link SHOULD point to the latest version of the collection. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  5. alternate: An alternative representation of the collection. For example, this could be the collection available through another catalog service such as OGC CSW, a human-readable HTML version or a metadata document following another standard such as ISO 19115 or DCAT. For additional relation types see also the lists of common relation types in openEO and the STAC specification for Collections.
required
object (STAC Collection Cube Dimensions)

Uniquely named dimensions of the data cube.

The keys of the object are the dimension names. For interoperability, it is RECOMMENDED to use the following dimension names if there is only a single dimension with the specified criteria:

  • x for the dimension of type spatial with the axis set to x
  • y for the dimension of type spatial with the axis set to y
  • z for the dimension of type spatial with the axis set to z
  • t for the dimension of type temporal
  • bands for dimensions of type bands

This property REQUIRES to add datacube to the list of stac_extensions.

required
object (STAC Summaries (Collection Properties))

Collection properties from STAC extensions (e.g. EO, SAR, Satellite or Scientific) or even custom extensions.

Summaries are either a unique set of all available values or statistics. Statistics by default only specify the range (minimum and maximum values), but can optionally be accompanied by additional statistical values. The range can specify the potential range of values, but it is recommended to be as precise as possible. The set of values MUST contain at least one element and it is strongly RECOMMENDED to list all values. It is recommended to list as many properties as reasonable so that consumers get a full overview of the Collection. Properties that are covered by the Collection specification (e.g. providers and license) SHOULD NOT be repeated in the summaries.

Potential fields for the summaries can be found here:

  • STAC Common Metadata: A list of commonly used fields throughout all domains
  • Content Extensions: Domain-specific fields for domains such as EO, SAR and point clouds.
  • Custom Properties: It is generally allowed to add custom fields.
object (Assets)

Dictionary of asset objects for data that can be downloaded, each with a unique key. The keys MAY be used by clients as file names.

Implementing this property REQUIRES to add collection-assets to the list of stac_extensions in STAC < 1.0.0-rc.1.

Response samples

Content type
application/json
{
  • "stac_version": "1.0.0-rc.2",
  • "type": "Collection",
  • "id": "Sentinel-2",
  • "title": "Sentinel-2 MSI L2A",
  • "description": "Sentinel-2A is a wide-swath, high-resolution, multi-spectral imaging mission supporting Copernicus Land Monitoring studies.",
  • "license": "proprietary",
  • "keywords": [
    ],
  • "providers": [],
  • "extent": {
    },
  • "links": [],
  • "cube:dimensions": {
    },
  • "summaries": {
    },
  • "assets": {}
}

Process Discovery

These endpoints allow to list the predefined processes that are available at the back-end. To list user-defined processes see 'User-Defined Processes'.

Supported predefined processes

Lists all predefined processes and returns detailed process descriptions, including parameters and return values.

Authorizations:
Bearer
query Parameters
limit
integer >= 1
Example: limit=10

This parameter enables pagination for the endpoint and specifies the maximum number of elements that arrays in the top-level object (e.g. collections, processes, batch jobs, secondary services, log entries, etc.) are allowed to contain. The links array MUST NOT be paginated like the resources, but instead contain links related to the paginated resources or the pagination itself (e.g. a link to the next page). If the parameter is not provided or empty, all elements are returned.

Pagination is OPTIONAL: back-ends or clients may not support it. Therefore it MUST be implemented in a way that clients not supporting pagination get all resources regardless. Back-ends not supporting pagination MUST return all resources.

If the response is paginated, the links array MUST be used to communicate the links for browsing the pagination with pre-defined rel types. See the links array schema for supported rel types. Backend implementations can, unless specified otherwise, use all kind of pagination techniques, depending on what is supported best by their infrastructure: page-based, offset-based, token-based or something else. The clients SHOULD use whatever is specified in the links with the corresponding rel types.

Responses

Response Schema: application/json
required
Array of objects (Pre-Defined Process)
required
Array of objects (links_pagination)

Links related to this list of resources, for example links for pagination or alternative formats such as a human-readable HTML version. The links array MUST NOT be paginated.

If pagination is implemented, the following rel (relation) types apply:

  1. next (REQUIRED): A link to the next page, except on the last page.

  2. prev (OPTIONAL): A link to the previous page, except on the first page.

  3. first (OPTIONAL): A link to the first page, except on the first page.

  4. last (OPTIONAL): A link to the last page, except on the last page.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "processes": [
    ],
  • "links": []
}

List all user-defined processes

Lists all user-defined processes (process graphs) of the authenticated user that are stored at the back-end.

It is strongly RECOMMENDED to keep the response size small by omitting larger optional values from the objects in processes (e.g. the exceptions, examples and links properties). To get the full metadata for a user-defined process clients MUST request GET /process_graphs/{process_graph_id}.

Authorizations:
Bearer
query Parameters
limit
integer >= 1
Example: limit=10

This parameter enables pagination for the endpoint and specifies the maximum number of elements that arrays in the top-level object (e.g. collections, processes, batch jobs, secondary services, log entries, etc.) are allowed to contain. The links array MUST NOT be paginated like the resources, but instead contain links related to the paginated resources or the pagination itself (e.g. a link to the next page). If the parameter is not provided or empty, all elements are returned.

Pagination is OPTIONAL: back-ends or clients may not support it. Therefore it MUST be implemented in a way that clients not supporting pagination get all resources regardless. Back-ends not supporting pagination MUST return all resources.

If the response is paginated, the links array MUST be used to communicate the links for browsing the pagination with pre-defined rel types. See the links array schema for supported rel types. Backend implementations can, unless specified otherwise, use all kind of pagination techniques, depending on what is supported best by their infrastructure: page-based, offset-based, token-based or something else. The clients SHOULD use whatever is specified in the links with the corresponding rel types.

Responses

Response Schema: application/json
required
Array of objects (User-defined Process Metadata)

Array of user-defined processes

required
Array of objects (links_pagination)

Links related to this list of resources, for example links for pagination or alternative formats such as a human-readable HTML version. The links array MUST NOT be paginated.

If pagination is implemented, the following rel (relation) types apply:

  1. next (REQUIRED): A link to the next page, except on the last page.

  2. prev (OPTIONAL): A link to the previous page, except on the first page.

  3. first (OPTIONAL): A link to the first page, except on the first page.

  4. last (OPTIONAL): A link to the last page, except on the last page.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "processes": [
    ],
  • "links": [ ]
}

Full metadata for a user-defined process

Lists all information about a user-defined process, including its process graph.

Authorizations:
Bearer
path Parameters
process_graph_id
required
string (process_id) ^\w+$
Example: ndvi

Unique identifier for a user-defined process.

Responses

Response Schema: application/json
summary
string or null

A short summary of what the process does.

description
string or null <commonmark>

Detailed description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation. In addition to the CommonMark syntax, clients can convert process IDs that are formatted as in the following example into links instead of code blocks: ``process_id()``

Array of objects or null (Process Parameter)

A list of parameters.

The order in the array corresponds to the parameter order to be used in clients that don't support named parameters.

Note: Specifying an empty array is different from (if allowed) null or the property being absent. An empty array means the process has no parameters. null / property absent means that the parameters are unknown as the user has not specified them. There could still be parameters in the process graph, if one is specified.

object or null (Process Return Value)

Description of the data that is returned by this process.

id
required
string (process_id) ^\w+$

The identifier for the process. It MUST be unique across its namespace (e.g. pre-defined processes or user-defined processes).

Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

categories
Array of strings (process_categories)

A list of categories.

deprecated
boolean (deprecated)
Default: false

Declares that the specified entity is deprecated with the potential to be removed in any of the next versions. It should be transitioned out of usage as soon as possible and users should refrain from using it in new implementations.

experimental
boolean (experimental)
Default: false

Declares that the specified entity is experimental, which means that it is likely to change or may produce unpredictable behaviour. Users should refrain from using it in production, but still feel encouraged to try it out and give feedback.

object (Process Exceptions)

Declares exceptions (errors) that might occur during execution of this process. This list is just for informative purposes and may be incomplete. This list MUST only contain exceptions that stop the execution of a process and MUST NOT contain warnings, notices or debugging messages. It is meant to primarily contain errors that have been caused by the user. It is RECOMMENDED that exceptions are referred to and explained in process or parameter descriptions.

The keys define the error code and MUST match the following pattern: ^\w+$

This schema follows the schema of the general openEO error list (see errors.json).

Array of objects (Process Example)

Examples, may be used for unit tests.

Array of objects (Link)

Links related to this process, e.g. additional external documentation. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. latest-version: If a process has been marked as deprecated, a link SHOULD point to the preferred version of the process. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  2. example: Links to examples of other processes that use this process.
  3. cite-as: For all DOIs associated with the process, the respective DOI links SHOULD be added. For additional relation types see also the lists of common relation types in openEO.
required
object (Process Graph)

A process graph defines a graph-like structure as a connected set of executable processes. Each key is a unique identifier (node ID) that is used to refer to the process in the graph.

Response samples

Content type
application/json

A user-defined process that computes the EVI.

{
  • "id": "evi",
  • "summary": "Enhanced Vegetation Index",
  • "description": "Computes the Enhanced Vegetation Index (EVI). It is computed with the following formula: `2.5 * (NIR - RED) / (1 + NIR + 6*RED + -7.5*BLUE)`.",
  • "parameters": [
    ],
  • "returns": {
    },
  • "process_graph": {
    }
}

User-Defined Processes

These endpoints allow to store and manage user-defined processes with their process graphs at the back-end.

Validate a user-defined process (graph)

Validates a user-defined process without executing it. A user-defined process is considered valid unless the errors array in the response contains at least one error.

Checks whether the process graph is schematically correct and the processes are supported by the back-end. It MUST also check the arguments against the schema, but checking whether the arguments are adequate in the context of data is OPTIONAL. For example, a non-existing band name may get rejected only by a few back-ends. The validation MUST NOT throw an error for unresolvable process parameters.

Back-ends MUST validate the process graph. Validating the corresponding metadata is OPTIONAL.

Errors that usually occur during processing MAY NOT get reported, e.g. if a referenced file is accessible at the time of execution.

Back-ends can either report all errors at once or stop the validation once they found the first error.

Please note that a validation always returns with HTTP status code 200. Error codes in the 4xx and 5xx ranges MUST be returned only when the general validation request is invalid (e.g. server is busy or properties in the request body are missing), but never if an error was found during validation of the user-defined process (e.g. an unsupported process).

Authorizations:
Bearer
Request Body schema: application/json

Specifies the user-defined process to be validated.

id
string or null^\w+$

The identifier for the process. It MUST be unique across its namespace (e.g. pre-defined processes or user-defined processes).

Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

summary
string or null

A short summary of what the process does.

description
string or null <commonmark>

Detailed description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation. In addition to the CommonMark syntax, clients can convert process IDs that are formatted as in the following example into links instead of code blocks: ``process_id()``

Array of objects or null (Process Parameter)

A list of parameters.

The order in the array corresponds to the parameter order to be used in clients that don't support named parameters.

Note: Specifying an empty array is different from (if allowed) null or the property being absent. An empty array means the process has no parameters. null / property absent means that the parameters are unknown as the user has not specified them. There could still be parameters in the process graph, if one is specified.

object or null (Process Return Value)

Description of the data that is returned by this process.

categories
Array of strings (process_categories)

A list of categories.

deprecated
boolean (deprecated)
Default: false

Declares that the specified entity is deprecated with the potential to be removed in any of the next versions. It should be transitioned out of usage as soon as possible and users should refrain from using it in new implementations.

experimental
boolean (experimental)
Default: false

Declares that the specified entity is experimental, which means that it is likely to change or may produce unpredictable behaviour. Users should refrain from using it in production, but still feel encouraged to try it out and give feedback.

object (Process Exceptions)

Declares exceptions (errors) that might occur during execution of this process. This list is just for informative purposes and may be incomplete. This list MUST only contain exceptions that stop the execution of a process and MUST NOT contain warnings, notices or debugging messages. It is meant to primarily contain errors that have been caused by the user. It is RECOMMENDED that exceptions are referred to and explained in process or parameter descriptions.

The keys define the error code and MUST match the following pattern: ^\w+$

This schema follows the schema of the general openEO error list (see errors.json).

Array of objects (Process Example)

Examples, may be used for unit tests.

Array of objects (Link)

Links related to this process, e.g. additional external documentation. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. latest-version: If a process has been marked as deprecated, a link SHOULD point to the preferred version of the process. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  2. example: Links to examples of other processes that use this process.
  3. cite-as: For all DOIs associated with the process, the respective DOI links SHOULD be added. For additional relation types see also the lists of common relation types in openEO.
required
object (Process Graph)

A process graph defines a graph-like structure as a connected set of executable processes. Each key is a unique identifier (node ID) that is used to refer to the process in the graph.

Responses

Response Schema: application/json
required
Array of objects (General Error)

A list of validation errors.

Request samples

Content type
application/json

A user-defined process that computes the EVI.

{
  • "id": "evi",
  • "summary": "Enhanced Vegetation Index",
  • "description": "Computes the Enhanced Vegetation Index (EVI). It is computed with the following formula: `2.5 * (NIR - RED) / (1 + NIR + 6*RED + -7.5*BLUE)`.",
  • "parameters": [
    ],
  • "returns": {
    },
  • "process_graph": {
    }
}

Response samples

Content type
application/json
{}

List all user-defined processes

Lists all user-defined processes (process graphs) of the authenticated user that are stored at the back-end.

It is strongly RECOMMENDED to keep the response size small by omitting larger optional values from the objects in processes (e.g. the exceptions, examples and links properties). To get the full metadata for a user-defined process clients MUST request GET /process_graphs/{process_graph_id}.

Authorizations:
Bearer
query Parameters
limit
integer >= 1
Example: limit=10

This parameter enables pagination for the endpoint and specifies the maximum number of elements that arrays in the top-level object (e.g. collections, processes, batch jobs, secondary services, log entries, etc.) are allowed to contain. The links array MUST NOT be paginated like the resources, but instead contain links related to the paginated resources or the pagination itself (e.g. a link to the next page). If the parameter is not provided or empty, all elements are returned.

Pagination is OPTIONAL: back-ends or clients may not support it. Therefore it MUST be implemented in a way that clients not supporting pagination get all resources regardless. Back-ends not supporting pagination MUST return all resources.

If the response is paginated, the links array MUST be used to communicate the links for browsing the pagination with pre-defined rel types. See the links array schema for supported rel types. Backend implementations can, unless specified otherwise, use all kind of pagination techniques, depending on what is supported best by their infrastructure: page-based, offset-based, token-based or something else. The clients SHOULD use whatever is specified in the links with the corresponding rel types.

Responses

Response Schema: application/json
required
Array of objects (User-defined Process Metadata)

Array of user-defined processes

required
Array of objects (links_pagination)

Links related to this list of resources, for example links for pagination or alternative formats such as a human-readable HTML version. The links array MUST NOT be paginated.

If pagination is implemented, the following rel (relation) types apply:

  1. next (REQUIRED): A link to the next page, except on the last page.

  2. prev (OPTIONAL): A link to the previous page, except on the first page.

  3. first (OPTIONAL): A link to the first page, except on the first page.

  4. last (OPTIONAL): A link to the last page, except on the last page.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "processes": [
    ],
  • "links": [ ]
}

Full metadata for a user-defined process

Lists all information about a user-defined process, including its process graph.

Authorizations:
Bearer
path Parameters
process_graph_id
required
string (process_id) ^\w+$
Example: ndvi

Unique identifier for a user-defined process.

Responses

Response Schema: application/json
summary
string or null

A short summary of what the process does.

description
string or null <commonmark>

Detailed description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation. In addition to the CommonMark syntax, clients can convert process IDs that are formatted as in the following example into links instead of code blocks: ``process_id()``

Array of objects or null (Process Parameter)

A list of parameters.

The order in the array corresponds to the parameter order to be used in clients that don't support named parameters.

Note: Specifying an empty array is different from (if allowed) null or the property being absent. An empty array means the process has no parameters. null / property absent means that the parameters are unknown as the user has not specified them. There could still be parameters in the process graph, if one is specified.

object or null (Process Return Value)

Description of the data that is returned by this process.

id
required
string (process_id) ^\w+$

The identifier for the process. It MUST be unique across its namespace (e.g. pre-defined processes or user-defined processes).

Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

categories
Array of strings (process_categories)

A list of categories.

deprecated
boolean (deprecated)
Default: false

Declares that the specified entity is deprecated with the potential to be removed in any of the next versions. It should be transitioned out of usage as soon as possible and users should refrain from using it in new implementations.

experimental
boolean (experimental)
Default: false

Declares that the specified entity is experimental, which means that it is likely to change or may produce unpredictable behaviour. Users should refrain from using it in production, but still feel encouraged to try it out and give feedback.

object (Process Exceptions)

Declares exceptions (errors) that might occur during execution of this process. This list is just for informative purposes and may be incomplete. This list MUST only contain exceptions that stop the execution of a process and MUST NOT contain warnings, notices or debugging messages. It is meant to primarily contain errors that have been caused by the user. It is RECOMMENDED that exceptions are referred to and explained in process or parameter descriptions.

The keys define the error code and MUST match the following pattern: ^\w+$

This schema follows the schema of the general openEO error list (see errors.json).

Array of objects (Process Example)

Examples, may be used for unit tests.

Array of objects (Link)

Links related to this process, e.g. additional external documentation. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. latest-version: If a process has been marked as deprecated, a link SHOULD point to the preferred version of the process. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  2. example: Links to examples of other processes that use this process.
  3. cite-as: For all DOIs associated with the process, the respective DOI links SHOULD be added. For additional relation types see also the lists of common relation types in openEO.
required
object (Process Graph)

A process graph defines a graph-like structure as a connected set of executable processes. Each key is a unique identifier (node ID) that is used to refer to the process in the graph.

Response samples

Content type
application/json

A user-defined process that computes the EVI.

{
  • "id": "evi",
  • "summary": "Enhanced Vegetation Index",
  • "description": "Computes the Enhanced Vegetation Index (EVI). It is computed with the following formula: `2.5 * (NIR - RED) / (1 + NIR + 6*RED + -7.5*BLUE)`.",
  • "parameters": [
    ],
  • "returns": {
    },
  • "process_graph": {
    }
}

Store a user-defined process

Stores a provided user-defined process with process graph that can be reused in other processes. If a process with the specified process_graph_id exists, the process is fully replaced. The id can't be changed for stored user-defined processes.

The id MUST be unique for the authenticated user, including all pre-defined processes by the back-end.

Partially updating user-defined processes is not supported.

To simplify exchanging user-defined processes, the property id can be part of the request body. If the values don't match, the value for id gets replaced with the value from the process_graph_id parameter in the path.

Authorizations:
Bearer
path Parameters
process_graph_id
required
string (process_id) ^\w+$
Example: ndvi

Unique identifier for a user-defined process.

Request Body schema: application/json

Specifies the process graph with its meta data.

id
string or null^\w+$

The identifier for the process. It MUST be unique across its namespace (e.g. pre-defined processes or user-defined processes).

Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

summary
string or null

A short summary of what the process does.

description
string or null <commonmark>

Detailed description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation. In addition to the CommonMark syntax, clients can convert process IDs that are formatted as in the following example into links instead of code blocks: ``process_id()``

Array of objects or null (Process Parameter)

A list of parameters.

The order in the array corresponds to the parameter order to be used in clients that don't support named parameters.

Note: Specifying an empty array is different from (if allowed) null or the property being absent. An empty array means the process has no parameters. null / property absent means that the parameters are unknown as the user has not specified them. There could still be parameters in the process graph, if one is specified.

object or null (Process Return Value)

Description of the data that is returned by this process.

categories
Array of strings (process_categories)

A list of categories.

deprecated
boolean (deprecated)
Default: false

Declares that the specified entity is deprecated with the potential to be removed in any of the next versions. It should be transitioned out of usage as soon as possible and users should refrain from using it in new implementations.

experimental
boolean (experimental)
Default: false

Declares that the specified entity is experimental, which means that it is likely to change or may produce unpredictable behaviour. Users should refrain from using it in production, but still feel encouraged to try it out and give feedback.

object (Process Exceptions)

Declares exceptions (errors) that might occur during execution of this process. This list is just for informative purposes and may be incomplete. This list MUST only contain exceptions that stop the execution of a process and MUST NOT contain warnings, notices or debugging messages. It is meant to primarily contain errors that have been caused by the user. It is RECOMMENDED that exceptions are referred to and explained in process or parameter descriptions.

The keys define the error code and MUST match the following pattern: ^\w+$

This schema follows the schema of the general openEO error list (see errors.json).

Array of objects (Process Example)

Examples, may be used for unit tests.

Array of objects (Link)

Links related to this process, e.g. additional external documentation. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. latest-version: If a process has been marked as deprecated, a link SHOULD point to the preferred version of the process. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  2. example: Links to examples of other processes that use this process.
  3. cite-as: For all DOIs associated with the process, the respective DOI links SHOULD be added. For additional relation types see also the lists of common relation types in openEO.
required
object (Process Graph)

A process graph defines a graph-like structure as a connected set of executable processes. Each key is a unique identifier (node ID) that is used to refer to the process in the graph.

Responses

Request samples

Content type
application/json

A user-defined process that computes the EVI.

{
  • "id": "evi",
  • "summary": "Enhanced Vegetation Index",
  • "description": "Computes the Enhanced Vegetation Index (EVI). It is computed with the following formula: `2.5 * (NIR - RED) / (1 + NIR + 6*RED + -7.5*BLUE)`.",
  • "parameters": [
    ],
  • "returns": {
    },
  • "process_graph": {
    }
}

Response samples

Content type
application/json
{}

Delete a user-defined process

Deletes the data related to this user-defined process, including its process graph.

Does NOT delete jobs or services that reference this user-defined process.

Authorizations:
Bearer
path Parameters
process_graph_id
required
string (process_id) ^\w+$
Example: ndvi

Unique identifier for a user-defined process.

Responses

Response samples

Content type
application/json
{}

Data Processing

Organizes and manages data processing on the back-end, either as synchronous on-demand computation or batch jobs.

Supported file formats

Lists supported input and output file formats. Input file formats specify which file a back-end can read from. Output file formats specify which file a back-end can write to.

The response to this request is an object listing all available input and output file formats separately with their parameters and additional data. This endpoint does not include the supported secondary web services.

Note: Format names and parameters MUST be fully aligned with the GDAL codes if available, see GDAL Raster Formats and OGR Vector Formats. It is OPTIONAL to support all output format parameters supported by GDAL. Some file formats not available through GDAL may be defined centrally for openEO. Custom file formats or parameters MAY be defined.

The format descriptions must describe how the file formats relate to data cubes. Input file formats must describe how the files have to be structured to be transformed into data cubes. Output file formats must describe how the data cubes are stored at the back-end and how the resulting file structure looks like.

Back-ends MUST NOT support aliases, for example it is not allowed to support geotiff instead of gtiff. Nevertheless, openEO Clients MAY translate user input input for convenience (e.g. translate geotiff to gtiff). Also, for a better user experience the back-end can specify a title.

Format names MUST be accepted in a case insensitive manner throughout the API.

Authorizations:
Bearer

Responses

Response Schema: application/json
required
object (Input File Formats)

Map of supported input file formats, i.e. file formats a back-end can read from. The property keys are the file format names that are used by clients and users, for example in process graphs.

required
object (Output File Formats)

Map of supported output file formats, i.e. file formats a back-end can write to. The property keys are the file format names that are used by clients and users, for example in process graphs.

Response samples

Content type
application/json
{
  • "output": {
    },
  • "input": {
    }
}

Validate a user-defined process (graph)

Validates a user-defined process without executing it. A user-defined process is considered valid unless the errors array in the response contains at least one error.

Checks whether the process graph is schematically correct and the processes are supported by the back-end. It MUST also check the arguments against the schema, but checking whether the arguments are adequate in the context of data is OPTIONAL. For example, a non-existing band name may get rejected only by a few back-ends. The validation MUST NOT throw an error for unresolvable process parameters.

Back-ends MUST validate the process graph. Validating the corresponding metadata is OPTIONAL.

Errors that usually occur during processing MAY NOT get reported, e.g. if a referenced file is accessible at the time of execution.

Back-ends can either report all errors at once or stop the validation once they found the first error.

Please note that a validation always returns with HTTP status code 200. Error codes in the 4xx and 5xx ranges MUST be returned only when the general validation request is invalid (e.g. server is busy or properties in the request body are missing), but never if an error was found during validation of the user-defined process (e.g. an unsupported process).

Authorizations:
Bearer
Request Body schema: application/json

Specifies the user-defined process to be validated.

id
string or null^\w+$

The identifier for the process. It MUST be unique across its namespace (e.g. pre-defined processes or user-defined processes).

Clients SHOULD warn the user if a user-defined process is added with the same identifier as one of the pre-defined process.

summary
string or null

A short summary of what the process does.

description
string or null <commonmark>

Detailed description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation. In addition to the CommonMark syntax, clients can convert process IDs that are formatted as in the following example into links instead of code blocks: ``process_id()``

Array of objects or null (Process Parameter)

A list of parameters.

The order in the array corresponds to the parameter order to be used in clients that don't support named parameters.

Note: Specifying an empty array is different from (if allowed) null or the property being absent. An empty array means the process has no parameters. null / property absent means that the parameters are unknown as the user has not specified them. There could still be parameters in the process graph, if one is specified.

object or null (Process Return Value)

Description of the data that is returned by this process.

categories
Array of strings (process_categories)

A list of categories.

deprecated
boolean (deprecated)
Default: false

Declares that the specified entity is deprecated with the potential to be removed in any of the next versions. It should be transitioned out of usage as soon as possible and users should refrain from using it in new implementations.

experimental
boolean (experimental)
Default: false

Declares that the specified entity is experimental, which means that it is likely to change or may produce unpredictable behaviour. Users should refrain from using it in production, but still feel encouraged to try it out and give feedback.

object (Process Exceptions)

Declares exceptions (errors) that might occur during execution of this process. This list is just for informative purposes and may be incomplete. This list MUST only contain exceptions that stop the execution of a process and MUST NOT contain warnings, notices or debugging messages. It is meant to primarily contain errors that have been caused by the user. It is RECOMMENDED that exceptions are referred to and explained in process or parameter descriptions.

The keys define the error code and MUST match the following pattern: ^\w+$

This schema follows the schema of the general openEO error list (see errors.json).

Array of objects (Process Example)

Examples, may be used for unit tests.

Array of objects (Link)

Links related to this process, e.g. additional external documentation. It is RECOMMENDED to provide links with the following rel (relation) types:

  1. latest-version: If a process has been marked as deprecated, a link SHOULD point to the preferred version of the process. The relation types predecessor-version (link to older version) and successor-version (link to newer version) can also be used to show the relation between versions.
  2. example: Links to examples of other processes that use this process.
  3. cite-as: For all DOIs associated with the process, the respective DOI links SHOULD be added. For additional relation types see also the lists of common relation types in openEO.
required
object (Process Graph)

A process graph defines a graph-like structure as a connected set of executable processes. Each key is a unique identifier (node ID) that is used to refer to the process in the graph.

Responses

Response Schema: application/json
required
Array of objects (General Error)

A list of validation errors.

Request samples

Content type
application/json

A user-defined process that computes the EVI.

{
  • "id": "evi",
  • "summary": "Enhanced Vegetation Index",
  • "description": "Computes the Enhanced Vegetation Index (EVI). It is computed with the following formula: `2.5 * (NIR - RED) / (1 + NIR + 6*RED + -7.5*BLUE)`.",
  • "parameters": [
    ],
  • "returns": {
    },
  • "process_graph": {
    }
}

Response samples

Content type
application/json
{}

Process and download data synchronously

Executes a user-defined process directly (synchronously) and the result will be downloaded in the format specified in the process graph. This endpoint can be used to generate small previews or test user-defined processes before starting a batch job. Timeouts on either client- or server-side are to be expected for complex computations. Back-ends MAY send the openEO error ProcessGraphComplexity immediately if the computation is expected to time out. Otherwise requests MAY time-out after a certain amount of time by sending openEO error RequestTimeout. A header named OpenEO-Costs MAY be sent with all responses, which MUST include the costs for processing and downloading the data. Additionally, a link to a log file MAY be sent in the header.

Authorizations:
Bearer
Request Body schema: application/json

Specifies the job details, e.g. the user-defined process and billing details.

required
object (Process Graph with metadata)

A process graph, optionally enriched with process metadata.

budget
number or null (budget)
Default: null

Maximum amount of costs the request is allowed to produce. The value MUST be specified in the currency of the back-end. No limits apply, if the value is null or the back-end has no currency set in GET /.

plan
string or null (billing_plan_null_default)
Default: null

The billing plan to process and charge the job or service with.

Billing plans MUST be accepted in a case insensitive manner. If null is given implicitly (through the default value) or explicitly by the client, the server MUST store the default plan and persist it regardless of any changes to the default billing plan in the future. If given, the default plan specific to the user has higher priority than the default plan of the back-end.

The list of plans and the default plan for the back-end can be retrieved from GET /. The default plan for the user can be retrieved from GET /me. Billing plans not on the list of available plans MUST be rejected with openEO error BillingPlanInvalid.

Responses

Response Headers
Content-Type
string

The appropriate media (MIME) type for the requested output format MUST be sent, if the response contains a single file.

To send multiple files at once it is RECOMMENDED to use the tar file format (media type: application/x-tar).

To mimic the results of batch jobs, it is RECOMMENDED that

  1. clients extract the tar file directly after receiving it so that users can directly work on the contained files and
  2. back-ends add STAC Items and/or Collections to the tar file so that users can make sense of the files.
OpenEO-Costs
number or null (money)
Default: null
Example: 12.98

MAY include the costs for processing and downloading the data.

Link
string^<[^>]+>;\s?rel="monitor"
Example: "<https://example.openeo.org/api/logs/258489231>; rel=\"monitor\""

The header MAY indicate a link to a log file generated by the request. If provided, the link MUST be serialized according to RFC 8288 and MUST use the relation type monitor. The link MUST follow the specifications for the links GET /jobs/{job_id}/logs and GET /services/{service_id}/logs, except that is MUST NOT accept any parameters (limit/offset). Therefore, the link MUST be accessible with HTTP GET, MUST be secured using a Bearer token and MUST follow the corresponding request body schema.

Request samples

Content type
application/json
{
  • "process": {
    },
  • "budget": 100,
  • "plan": "free"
}

Response samples

Content type
application/json
{}

List all batch jobs

Lists all batch jobs submitted by a user.

It is strongly RECOMMENDED to keep the response size small by omitting all optional non-scalar values from objects in jobs (i.e. the process property). To get the full metadata for a job clients MUST request GET /jobs/{job_id}.

Authorizations:
Bearer
query Parameters
limit
integer >= 1
Example: limit=10

This parameter enables pagination for the endpoint and specifies the maximum number of elements that arrays in the top-level object (e.g. collections, processes, batch jobs, secondary services, log entries, etc.) are allowed to contain. The links array MUST NOT be paginated like the resources, but instead contain links related to the paginated resources or the pagination itself (e.g. a link to the next page). If the parameter is not provided or empty, all elements are returned.

Pagination is OPTIONAL: back-ends or clients may not support it. Therefore it MUST be implemented in a way that clients not supporting pagination get all resources regardless. Back-ends not supporting pagination MUST return all resources.

If the response is paginated, the links array MUST be used to communicate the links for browsing the pagination with pre-defined rel types. See the links array schema for supported rel types. Backend implementations can, unless specified otherwise, use all kind of pagination techniques, depending on what is supported best by their infrastructure: page-based, offset-based, token-based or something else. The clients SHOULD use whatever is specified in the links with the corresponding rel types.

Responses

Response Schema: application/json
required
Array of objects (Batch Job)
required
Array of objects (links_pagination)

Links related to this list of resources, for example links for pagination or alternative formats such as a human-readable HTML version. The links array MUST NOT be paginated.

If pagination is implemented, the following rel (relation) types apply:

  1. next (REQUIRED): A link to the next page, except on the last page.

  2. prev (OPTIONAL): A link to the previous page, except on the first page.

  3. first (OPTIONAL): A link to the first page, except on the first page.

  4. last (OPTIONAL): A link to the last page, except on the last page.

For additional relation types see also the lists of common relation types in openEO.

Response samples

Content type
application/json
{
  • "jobs": [
    ],
  • "links": []
}

Create a new batch job

Creates a new batch processing task (job) from one or more (chained) processes at the back-end.

Processing the data doesn't start yet. The job status gets initialized as created by default.

Authorizations:
Bearer
Request Body schema: application/json

Specifies the job details, e.g. the user-defined process and billing details.

title
string or null (eo_title)

A short description to easily distinguish entities.

description
string or null <commonmark> (eo_description)

Detailed multi-line description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation.

required
object (Process Graph with metadata)

A process graph, optionally enriched with process metadata.

plan
string or null (billing_plan_null_default)
Default: null

The billing plan to process and charge the job or service with.

Billing plans MUST be accepted in a case insensitive manner. If null is given implicitly (through the default value) or explicitly by the client, the server MUST store the default plan and persist it regardless of any changes to the default billing plan in the future. If given, the default plan specific to the user has higher priority than the default plan of the back-end.

The list of plans and the default plan for the back-end can be retrieved from GET /. The default plan for the user can be retrieved from GET /me. Billing plans not on the list of available plans MUST be rejected with openEO error BillingPlanInvalid.

budget
number or null (budget)
Default: null

Maximum amount of costs the request is allowed to produce. The value MUST be specified in the currency of the back-end. No limits apply, if the value is null or the back-end has no currency set in GET /.

Responses

Response Headers
Location
required
string <uri>
Example: "https://example.openeo.org/api/jobs/123"

Absolute URL to the newly created batch job.

The URL points to the metadata endpoint GET /jobs/{job_id} with the {job_id} being the unique identifier (ID) of the created batch job.

OpenEO-Identifier
required
string (job_id) ^[\w\-\.~]+$
Example: "a3cca2b2aa1e3b5b"

Unique identifier of the batch job, generated by the back-end during creation. MUST match the specified pattern.

Request samples

Content type
application/json
{
  • "title": "NDVI based on Sentinel 2",
  • "description": "Deriving minimum NDVI measurements over pixel time series of Sentinel 2",
  • "process": {
    },
  • "plan": "free",
  • "budget": 100
}

Response samples

Content type
application/json
{}

Modify a batch job

Modifies an existing job at the back-end, but maintains the identifier. Changes can be grouped in a single request.

Jobs can only be modified when the job is not queued or running. Otherwise requests to this endpoint MUST be rejected with openEO error JobLocked.

Authorizations:
Bearer
path Parameters
job_id
required
string (job_id) ^[\w\-\.~]+$
Example: a3cca2b2aa1e3b5b

Unique job identifier.

Request Body schema: application/json

Specifies the job details to update.

title
string or null (eo_title)

A short description to easily distinguish entities.

description
string or null <commonmark> (eo_description)

Detailed multi-line description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation.

object (Process Graph with metadata)

A process graph, optionally enriched with process metadata.

plan
string or null (billing_plan_null)

The billing plan to process and charge the job or service with.

Billing plans MUST be accepted in a case insensitive manner. If null is specified by the client explicitly, the server MUST store the default plan and persist it regardless of any changes to the default billing plan in the future. If given, the default plan specific to the user has higher priority than the default plan of the back-end.

The list of plans and the default plan for the back-end can be retrieved from GET /. The default plan for the user can be retrieved from GET /me. Billing plans not on the list of available plans MUST be rejected with openEO error BillingPlanInvalid.

budget
number or null (budget_update)

Maximum amount of costs the request is allowed to produce. The value MUST be specified in the currency of the back-end. No limits apply, if the value is null.

Responses

Request samples

Content type
application/json
{
  • "title": "NDVI based on Sentinel 2",
  • "description": "Deriving minimum NDVI measurements over pixel time series of Sentinel 2",
  • "process": {
    },
  • "plan": "free",
  • "budget": 100
}

Response samples

Content type
application/json
{}

Full metadata for a batch job

Lists all information about a submitted batch job.

Authorizations:
Bearer
path Parameters
job_id
required
string (job_id) ^[\w\-\.~]+$
Example: a3cca2b2aa1e3b5b

Unique job identifier.

Responses

Response Schema: application/json
id
required
string (job_id) ^[\w\-\.~]+$

Unique identifier of the batch job, generated by the back-end during creation. MUST match the specified pattern.

title
string or null (eo_title)

A short description to easily distinguish entities.

description
string or null <commonmark> (eo_description)

Detailed multi-line description to explain the entity.

CommonMark 0.29 syntax MAY be used for rich text representation.

required
object (Process Graph with metadata)

A process graph, optionally enriched with process metadata.

status
required
string
Default: "created"
Enum: "created" "queued" "running" "canceled" "finished" "error"

The current status of a batch job.

The following status changes can occur:

  • POST /jobs: The status is initialized as created.
  • POST /jobs/{job_id}/results: The status is set to queued, if processing doesn't start instantly.
    • Once the processing starts the status is set to running.
    • Once the data is available to download the status is set to finished.
    • Whenever an error occurs during processing, the status MUST be set to error.
  • DELETE /jobs/{job_id}/results: The status is set to canceled if the status was running beforehand and partial or preliminary results are available to be downloaded. Otherwise the status is set to created.
progress
number [ 0 .. 100 ]

Indicates the process of a running batch job in percent. Can also be set for a job which stopped due to an error or was canceled by the user. In this case, the value indicates the progress at which the job stopped. The Property may not be available for the status codes created and queued. Submitted and queued jobs only allow the value 0, finished jobs only allow the value 100.

created
required
string <date-time> (created)

Date and time of creation, formatted as a RFC 3339 date-time.

updated
string <date-time> (updated)

Date and time of the last status change, formatted as a RFC 3339 date-time.

plan
string (billing_plan)

The billing plan to process and charge the job or service with.

Billing plans MUST be accepted in a case insensitive manner.

The plans can be retrieved from GET /, but the value returned here may not be in the list of plans any longer.

costs
number or null (money)
Default: null

An amount of money or credits. The value MUST be specified in the currency the back-end is working with. The currency can be retrieved by calling GET /. If no currency is set, this field MUST be null.

budget
number or null (budget)
Default: null

Maximum amount of costs the request is allowed to produce. The value MUST be specified in the currency of the back-end. No limits apply, if the value is null or the back-end has no currency set in GET /.

object (Resource usage metrics)

Metrics about the resource usage of the batch job.

Back-ends are not expected to update the metrics while processing data, so the metrics can only be available after the job has been finished or has errored. For usage metrics during processing, metrics can better be added to the logs (e.g. GET /jobs/{job_id}/logs) with the same schema.

Response samples

Content type
application/json
{
  • "id": "a3cca2b2aa1e3b5b",
  • "title": "NDVI based on Sentinel 2",
  • "description": "Deriving minimum NDVI measurements over pixel time series of Sentinel 2",
  • "process": {