credit.ocean.samudra_data

credit.ocean.samudra_data#

`extract_wet_mask`(data, outputs, hist)
`get_inference_steps`(time_config[, time_delta, hist])	Get the number of inference/rollout steps for the given time configuration.
`convert_tensor_out_to_dict`(→ Dict[str, torch.Tensor])
`get_norm_unnorm_dicts`(data[, input_type, ...])
`compute_anomalies`(→ xarray.Dataset)	Compute the anomalies of a data variable.
`rename_vars`(→ xarray.Dataset)	Rename variables if required.
`validate_data`(→ Tuple[xarray.Dataset, xarray.Dataset, ...)	Validate the data such that we have the correct format for training/evaluation.

credit.ocean.samudra_data.get_inference_steps(time_config, time_delta=5, hist=1)#

Get the number of inference/rollout steps for the given time configuration.

Parameters:

Returns:

Number of rollout steps

Return type:

num_steps

credit.ocean.samudra_data.convert_tensor_out_to_dict(tensor_out: torch.Tensor) → Dict[str, torch.Tensor]#

credit.ocean.samudra_data.get_norm_unnorm_dicts(data: torch.Tensor, input_type: str = 'target', prognostic_channels: int = 0, hist: int = 1)#

credit.ocean.samudra_data.compute_anomalies(data: xarray.Dataset, var: str) → xarray.Dataset#: Compute the anomalies of a data variable.

credit.ocean.samudra_data.rename_vars(data: xarray.Dataset) → xarray.Dataset#: Rename variables if required.

credit.ocean.samudra_data.validate_data(data: xarray.Dataset, data_mean: xarray.Dataset, data_std: xarray.Dataset) → Tuple[xarray.Dataset, xarray.Dataset, xarray.Dataset]#: Validate the data such that we have the correct format for training/evaluation.

class credit.ocean.samudra_data.Normalize#

classmethod init_instance(data_mean: xarray.Dataset, data_std: xarray.Dataset, prognostic_vars: str, boundary_vars: str, wet_mask: torch.Tensor) → Normalize#: Initialize the singleton instance with normalization parameters.

_initialize(data_mean: xarray.Dataset, data_std: xarray.Dataset, prognostic_vars: str, boundary_vars: str, wet_mask: torch.Tensor) → None#: Store normalization parameters and pre-compute numpy arrays.

_to_tensor(array: numpy.ndarray, device: torch.device) → torch.Tensor#: Convert numpy array to tensor on specified device.

normalize_prognostics(data: xarray.Dataset, fill_nan=True, fill_value=0.0) → xarray.Dataset#: Normalize input dataset.

normalize_boundary(data: xarray.Dataset, fill_nan=True, fill_value=0.0) → xarray.Dataset#: Normalize boundary conditions.

unnormalize_prognostics(data: xarray.Dataset) → xarray.Dataset#: Unnormalize output dataset.

normalize_tensor_prognostics(data: torch.Tensor, fill_nan=True, fill_value=0.0) → torch.Tensor#: Normalize output tensor.

unnormalize_tensor_prognostics(data: torch.Tensor) → torch.Tensor#: Unnormalize output tensor.

normalize_numpy_prognostics(data: numpy.ndarray, fill_nan=True, fill_value=0.0) → numpy.ndarray#: Normalize output numpy array.

unnormalize_numpy_prognostics(data: numpy.ndarray) → numpy.ndarray#: Unnormalize output numpy array.