earthkit.hydro.data_structures

Module contents

class earthkit.hydro.data_structures.RiverNetwork(river_network_storage: RiverNetworkStorage)

Bases: object

A class representing a river network for hydrological processing.

n_nodes

The number of nodes in the river network.

Type:

int

n_edges

The number of nodes in the river network.

Type:

int

sinks

Nodes with no downstream connections.

Type:

array-like

sources

Nodes with no upstream connections.

Type:

array-like

bifurcates

Whether the river network has bifurcations.

Type:

bool

shape

The size of the river network grid. None if the network is vector-based.

Type:

tuple

mask

Flattened 1D indices on the raster grid corresponding to river network nodes.

Type:

array-like

array_backend

The array backend of the river network.

Type:

str

device

The device of the river network.

Type:

str

return_type

The default return type of the river network. Either “gridded” or “masked”.

Type:

str

export(fpath='river_network.joblib', compression=1)

Save the river network to a local file.

Parameters:

• fpath (str, optional) – The filepath specifying where to save the RiverNetwork. Default is ‘river_network.joblib’.

• compression (str, optional) – The compression factor used for saving. Default is 1.

Return type:

None

set_default_return_type(return_type)

Set the default return type for the river network.

Parameters:

return_type (str) – The default return_type to use.

Return type:

None

to_device(device=None, array_backend=None)

Change the RiverNetwork’s array backend and/or move it to a different device.

Parameters:

• device (str, optional) – The device to which to transfer. Default is None, which is ‘cpu’ for all backends except cupy, which is ‘gpu’.

• array_backend (str, optional) – The array backend. One of “numpy”, “np”, “cupy”, “cp”, “pytorch”, “torch”, “jax”, “jnp”, “tensorflow”, “tf”, “mlx” or “mx”. Default is None, which uses self.array_backend.

Returns:

The modified RiverNetwork.

Return type:

RiverNetwork