Working with processes¶

In openEO, a process is an operation that performs a specific task on a set of parameters and returns a result. For example: one wishes to apply a statistical operation such as mean or median on selected EO data. A process is similar to a function or method in common programming languages and likewise, multiple processes can be combined or chained together into new, more complex operations.

A pre-defined process is a process provided by a given backend. These are often the processes defined centrally by openEO, including but not limited to common mathematical (sum, divide, sqrt, …), logical (and, if, …), statistical (mean, max, …), image processing (mask, apply_kernel, …) operations. Backends are expected to support most of these. Backends are also free to pre-define additional processes that are not (yet) centrally defined by openEO.

As noted above, processes can be combined into a larger, reusable unit and stored on the backend as a user-defined process. Note: don’t confuse user-defined processes with user-defined functions (UDF’s) in openEO, which is a mechanism to inject actual source code (for example Python code) into a process for fine-grained processing.

How processes are combined into a larger unit is internally represented by a so-called process graph. It describes how the inputs and outputs of process graphs should be linked together. A user of the Python client should normally not worry about the details of a process graph structure, as most of these aspects are hidden behind regular Python functions, classes and methods.

Using common pre-defined processes¶

The listing of pre-defined processes provided by a backend can be inspected with list_processes(). For example, to get a list of the process names (process ids):

>>> process_ids = [process["id"] for process in connection.list_processes()]
>>> print(process_ids[:16])
['arccos', 'arcosh', 'power', 'last', 'subtract', 'not', 'cosh', 'artanh',
'is_valid', 'first', 'median', 'eq', 'absolute', 'arctan2', 'divide','is_nan']

More information about the processes, like a description or expected parameters, can be queried like that, but it is often easier to look them up on the official openEO process documentation

Most of the important pre-defined processes are covered directly by methods of the DataCube class. For example, to apply the filter_temporal process to a data cube:

cube = cube.filter_temporal("2020-02-20", "2020-06-06")

Being regular Python methods, you get usual function call features you’re accustomed to: default values, keyword arguments, kwargs usage, … For example, to use a bounding box dictionary with kwargs-expansion:

bbox = {
    "west": 5.05, "south": 51.20, "east": 5.10, "north": 51.23,
    "crs": "EPSG:4326"
}
cube = cube.filter_bbox(**bbox)

Note that some methods try to be more flexible and convenient to use than how the official process definition prescribes. For example, the filter_temporal process expects an extent array with 2 items (the start and end date), but you can call the corresponding client method in multiple equivalent ways:

cube.filter_temporal("2019-07-01", "2019-08-01")
cube.filter_temporal(["2019-07-01", "2019-08-01"])
cube.filter_temporal(extent=["2019-07-01", "2019-08-01"])
cube.filter_temporal(start_date="2019-07-01", end_date="2019-08-01"])

Generic API for adding processes¶

An openEO backend may offer processes that are not part of the core API, or the client may not (yet) have a corresponding method for a process that you wish to use. In that case, you can fall back to a more generic API that allows you to add processes directly.

Basics¶

To add a simple process to the graph, use the process() method on a DataCube. You have to specify the process id and arguments (as a single dictionary or through keyword arguments **kwargs). It will return a new DataCube with the new process appended to the internal process graph.

A very simple example using the mean process and a literal list in an arguments dictionary:

arguments= {
    "data": [1, 3, -1]
}
res = cube.process("mean", arguments)

or equivalently, leveraging keyword arguments:

res = cube.process("mean", data=[1, 3, -1])

Passing data cube arguments¶

The example above is a bit convoluted however in the sense that you start from a given data cube cube, you add a mean process that works on a given data array, while completely ignoring the original cube. In reality you typically want to apply the process on the cube. This is possible by passing a data cube object directly as argument, for example with the ndvi process that at least expects a data cube as data argument

res = cube.process("ndvi", data=cube)

Note that you have have to specify cube twice here: a first time to call the method and a second time as argument. Moreover, it requires you to define a Python variable for the data cube, which is annoying if you want to use a chained expressions. To solve these issues, you can use the THIS constant as symbolic reference to the “current” cube:

from openeo.rest.datacube import THIS

res = (
    cube
        .process("filter_bands", data=THIS)
        .process("mask", data=THIS, mask=mask)
        .process("ndvi", data=THIS)
)

Data cube from process¶

There is a convenience function datacube_from_process() to directly create a DataCube from a single process using the Connection:

cube = connection.datacube_from_process("mean", data=[1, 3, -1])

Working with User-Defined Processes (UDP)¶

The openEO API specification allow users to define their own user-defined processes, expressed in terms of other existing pre-defined or other user-defined processes, and to store them on the backend so they can easily be reused.

To store a user-defined process, you have to express it as a process graph. Where you expect input (e.g. a data cube from preceding processes), you have to reference a parameter of your user-defined process with {"from_parameter": "parameter_name"}. For example:

blur = {
    "applykernel1": {
        "process_id": "apply_kernel",
        "arguments": {
            "data": {"from_parameter": "data"},
            "kernel": [[1, 1, 1], [1, 2, 1], [1, 1, 1]],
            "factor": 0.1,
        },
        "result": True,
    },
}
connection.save_user_defined_process("blur", blur)

This user-defined process can now be applied to a data cube as follows:

res = cube.process("blur", arguments={"data": THIS})

Process parameters in user-defined processes¶

To keep things well-documented, it is recommended to properly list the parameters used in your user-defined process, as Parameter instances. This also allows to specify default values. For example, iterating on the “blur” example:

from openeo.api.process import Parameter

blur = {
    "applykernel1": {
        "process_id": "apply_kernel",
        "arguments": {
            "data": {"from_parameter": "data"},
            "kernel": [[1, 1, 1], [1, 2, 1], [1, 1, 1]],
            "factor": {"from_parameter": "scale"},
        },
        "result": True,
    },
}
connection.save_user_defined_process("blur", blur, parameters=[
    Parameter(
        name="data", description="A data cube",
        schema={"type": "object", "subtype": "raster-cube"}
    ),
    Parameter(
        name="scale", description="Kernel multiplication factor",
        schema="number", default=0.1
    ),
])

Because the “raster-cube” parameter is so common, there is a helper function raster_cube() to easily create such a parameter. Also, you can specify the parameters as dictionaries if that would be more convenient. The parameter listing of the example above could be written like this:

parameters=[
    Parameter.raster_cube(name="data"),
    {
        "name": "scale", "description": "Kernel multiplication factor",
        "schema": "number", "default": 0.1
    }
]

Processes with “callbacks”¶

Some openEO processes expect some kind of sub-process to be invoked on a subset or slice of the datacube. For example:

process apply requires a transformation that will be applied to each pixel in the cube (separately)
process reduce_dimension requires an aggregation function to convert an array of pixel values (along a given dimension) to a single value
process apply_neighborhood requires a function to transform a small “neighborhood” cube to another

These transformation functions are usually called “callbacks” because instead of being called explicitly by the user, they are called by their “parent” process (the apply, reduce_dimension and apply_neighborhood in the examples)

The openEO Python Client Library currently provides a couple of functions that expect a callback, including: openeo.rest.datacube.DataCube.apply(), openeo.rest.datacube.DataCube.apply_dimension(), openeo.rest.datacube.DataCube.apply_neighborhood(), openeo.rest.datacube.DataCube.merge_cubes(), openeo.rest.datacube.DataCube.reduce_dimension(), and openeo.rest.datacube.DataCube.load_collection(). These functions support several ways to specify the desired callback.

Callback as string¶

The easiest way is passing a process name as a string, for example:

# Take the absolute value of each pixel
cube.apply("absolute")

# Reduce a cube along the temporal dimension by taking the maximum value
cube.reduce_dimension("max", dimension="t")

This approach is only possible if the desired transformation is available as a single process. If not, use one of the methods below.

Also important is that the “signature” of the provided callback process should correspond properly with what the parent process expects. For example: apply requires a callback process that receives a number and returns one (like absolute or sqrt), while reduce_dimension requires a callback process that receives an array of numbers and returns a single number (like max or mean).

Callback as a callable¶

You can also specify the callback as a “callable”: a Python object that can be called (e.g. a function without parenthesis).

The openEO Python Client Library defines the official processes in the openeo.process.processes module, which can be used directly:

from openeo.processes import absolute, max

cube.apply(absolute)
cube.reduce_dimension(max, dimension="t")

You can also use lambda functions:

cube.apply(lambda x: x * 2 + 3)

or normal Python functions:

from openeo.processes import array_element

def my_bandmath(data):
    band1 = array_element(data, index=1)
    band1 = array_element(data, index=1)
    return band1 + 1.2 * band 2


cube.reduce_dimension(my_bandmath, dimension="bands")

The argument that is passed to these functions is an instance of openeo.processes.ProcessBuilder. This is a helper object with predefined methods for all standard processes, allowing to use an object oriented coding style to define the callback. For example:

from openeo.processes import ProcessBuilder

def avg(data: ProcessBuilder):
    return data.mean()

cube.reduce_dimension(avg, dimension="t")

These methods also return ProcessBuilder objects, which also allows writing callbacks in chained fashion:

cube.apply(lambda x: x.absolute().cos().add(y=1.23))

All this gives a lot of flexibility to define callbacks compactly in a desired coding style. The following examples result in the same callback:

from openeo.processes import ProcessBuilder, mean, cos, add

# Chained methods
cube.reduce_dimension(
    lambda data: data.mean().cos().add(y=1.23),
    dimension="t"
)

# Functions
cube.reduce_dimension(
    lambda data: add(x=cos(mean(data)), y=1.23),
    dimension="t"
)

# Mixing methods, functions and operators
cube.reduce_dimension(
    lambda data: cos(data.mean())) + 1.23,
    dimension="t"
)

Caveats¶

Specifying callbacks through Python functions (or lambdas) looks intuitive and straightforward, but it should be noted that not everything is allowed in these functions. You should just limit yourself to calling openeo.process.processes functions, openeo.processes.ProcessBuilder methods and basic math operators. Don’t call functions from other libraries like numpy or scipy. Don’t use Python control flow statements like if/else constructs or for loops.

The reason for this is that the openEO Python Client Library does not translate the function source code itself to an openEO process graph. Instead, when building the openEO process graph, it passes a special object to the function and keeps track of which openeo.process.processes functions were called to assemble the corresponding process graph. If you use control flow statements or use numpy functions for example, this procedure will incorrectly detect what you want to do in the callback.

Callback as `PGNode`¶

You can also pass a PGNode object as callback. This method is used internally and could be useful for more advanced use cases, but it requires more in-depth knowledge of the openEO API and openEO Python Client Library to construct correctly. Some examples:

from openeo.internal.graph_building import PGNode

cube.apply(PGNode(
    "add",
    x=PGNode(
        "cos",
        x=PGNode("absolute", x={"from_parameter": "x"})
    ),
    y=1.23
))

cube.reduce_dimension(
    reducer=PGNode("max", data={"from_parameter": "data"}),
    dimension="bands"
)

openEO

Table of Contents

Related Topics

Working with processes¶

Using common pre-defined processes¶

Generic API for adding processes¶

Basics¶

Passing data cube arguments¶

Data cube from process¶

Working with User-Defined Processes (UDP)¶

Process parameters in user-defined processes¶

Processes with “callbacks”¶

Callback as string¶

Callback as a callable¶

Caveats¶

Callback as `PGNode`¶

Working with processes¶

Using common pre-defined processes¶

Generic API for adding processes¶

Basics¶

Passing data cube arguments¶

Data cube from process¶

Working with User-Defined Processes (UDP)¶

Process parameters in user-defined processes¶

Processes with “callbacks”¶

Callback as string¶

Callback as a callable¶

Caveats¶

Callback as PGNode¶

Callback as `PGNode`¶