Creating a Model Workflow

A model workflow (the .yaml file that tells hydromt what todo) consists of three main sections:

1. modeltype

2. global

3. steps

The model type tells hydromt which model to use. In the case of using hydromt core, model is the only option but if you have plugins installed those will probably provide other options as well. You can discover which options you have installed with the command hydromt --models.

The global is where you provide any configuration that the model will need at initialization. This is where you for example, can (and must in the case of core) add components, and tell the model which spacial component should be used to figure out what the model region is.

In a workflow file, all the components that a model needs to have (appart from any default components your plugin has already provided) must be declared upfront. This is done with the global keyword (typcially placed at the start of the file)

it should look like this

global:
    components:
        grid:
            type: GridComponent
        config:
            type: ConfigComponent
        forcing:
            type: SpatialDatasetsComponent
            region_component: grid
    region_component: grid

here you can see that components should take a mapping where the keys are the name the component will have. These must then again take a mapping that specifies at least the type of component. The name of the component type should correspond to the python class name.

An additional point of note is that spacial components (such as forcing and grid) in the example above, can either define their own region (grid) or derive their region from another component (forcing). This can be done by specifying the region_component key, and should refer to the name of the spacial component you wish to use. You can also specify which spacial component the model should derive it’s region from. If you have only one spacial component, this information may be omitted, but if you have multiple (as in the example above) you must also specify which component the model should derive it’s region from.

Finally there is the steps part of the workflow. This should be a list, where each list item should be a name of a function you want to run, followed by any arguments you want to pass to that function. You can use the . syntax to call functions on components, or omit this if the function you want to call is defined on the model.

For example, after you have specified the components that should be added to your model in the global key, you can use them in the steps of your workflow like so:

steps:
    - grid.add_data_from_constant:
        constant: 0.01
        name: "c1"
        nodata: -99.0
    - ...

in the example above grid is the name of the component and add_data_from_constant is the function on it that you want to call. Please check the specific component for what functions are available. Note that only functions that have the @hydromt_step decorator are available to use from the yaml workflow.