sup3r.models.multi_step.SolarMultiStepGan

class SolarMultiStepGan(spatial_solar_models, spatial_wind_models, temporal_solar_models, t_enhance=None, temporal_pad=0)

Bases: MultiStepGan

Special multi step model for solar clearsky ratio super resolution.

This model takes in two parallel models for wind-only and solar-only spatial super resolutions, then combines them into a 3-channel high-spatial-resolution input (clearsky_ratio, U_200m, V_200m) for a solar temporal super resolution model.

Parameters:

• spatial_solar_models (MultiStepGan) – A loaded MultiStepGan object representing the one or more spatial super resolution steps in this composite MultiStepGan model that inputs and outputs clearsky_ratio

• spatial_wind_models (MultiStepGan) – A loaded MultiStepGan object representing the one or more spatial super resolution steps in this composite MultiStepGan model that inputs and outputs wind u/v features and must include U_200m + V_200m as output features.

• temporal_solar_models (MultiStepGan) – A loaded MultiStepGan object representing the one or more (spatio)temporal super resolution steps in this composite SolarMultiStepGan model. This is the final step in the custom solar downscaling methodology.

• t_enhance (int | None) – Optional argument to fix or update the temporal enhancement of the model. This can be used with temporal_pad to manipulate the output shape to match whatever padded shape the sup3r forward pass module expects.

• temporal_pad (int) – Optional reflected padding of the generated output array.

Methods

generate(low_res[, norm_in, un_norm_out, ...])	Use the generator model to generate high res data from low res input.
get_s_enhance_from_layers()	Compute factor by which model will enhance spatial resolution from layer attributes.
get_t_enhance_from_layers()	Compute factor by which model will enhance temporal resolution from layer attributes.
load(spatial_solar_model_dirs, ...[, ...])	Load the GANs with its sub-networks from a previously saved-to output directory.
preflight()	Run some preflight checks to make sure the loaded models can work together.
save_params(out_dir)
seed([s])	Set the random seed for reproducible results.
set_model_params(**kwargs)	Set parameters used for training the model
temporal_pad(hi_res[, mode])	Optionally add temporal padding to the 5D generated output array

Attributes

hr_exo_features	Get list of high-resolution exogenous filter names the model uses.
hr_out_features	Get the list of output feature names that the last solar spatiotemporal generative model in this SolarMultiStepGan outputs.
idf_solar	Get an array of feature indices for the subset of features required for the spatial_solar_models.
idf_wind	Get an array of feature indices for the subset of features required for the spatial_wind_models.
idf_wind_out	Get an array of spatial_wind_models output feature indices that are required for input to the temporal_solar_models.
input_dims	Get dimension of model generator input.
input_resolution	'...km', 'temporal': '...min'}.
is_4d	Check if model expects spatial only input
is_5d	Check if model expects spatiotemporal input
lr_features	Get a list of low-resolution features input to the generative model.
means	Get the data normalization mean values.
meta	Get a tuple of meta data dictionaries for all models
model_params	Get a tuple of model parameters for all models
models	Get an ordered tuple of the Sup3rGan models that are part of this MultiStepGan
output_resolution	Resolution of output data.
s_enhance	Factor by which model will enhance spatial resolution.
smoothed_features	Get the list of smoothed input feature names that the generative model was trained on.
smoothing	Value of smoothing parameter used in gaussian filtering of coarsened high res data.
spatial_solar_models	Get the MultiStepGan object for the spatial-only solar model(s)
spatial_wind_models	Get the MultiStepGan object for the spatial-only wind model(s)
stdevs	Get the data normalization standard deviation values.
t_enhance	Factor by which model will enhance temporal resolution.
temporal_solar_models	Get the MultiStepGan object for the (spatio)temporal model(s)
version_record	Get a tuple of version records from all models

preflight()

Run some preflight checks to make sure the loaded models can work together.

property spatial_solar_models

Get the MultiStepGan object for the spatial-only solar model(s)

Returns:

MultiStepGan

property spatial_wind_models

Get the MultiStepGan object for the spatial-only wind model(s)

Returns:

MultiStepGan

property temporal_solar_models

Get the MultiStepGan object for the (spatio)temporal model(s)

Returns:

MultiStepGan

property meta

Get a tuple of meta data dictionaries for all models

Returns:

tuple

property lr_features

Get a list of low-resolution features input to the generative model. This includes low-resolution features that might be supplied exogenously at inference time but that were in the low-res batches during training

property hr_out_features

Get the list of output feature names that the last solar spatiotemporal generative model in this SolarMultiStepGan outputs.

property idf_wind

Get an array of feature indices for the subset of features required for the spatial_wind_models. This excludes topography which is assumed to be provided as exogenous_data.

property idf_wind_out

Get an array of spatial_wind_models output feature indices that are required for input to the temporal_solar_models. Typically this is the indices of U_200m + V_200m from the output features of spatial_wind_models

property idf_solar

Get an array of feature indices for the subset of features required for the spatial_solar_models. This excludes topography which is assumed to be provided as exogenous_data.

generate(low_res, norm_in=True, un_norm_out=True, exogenous_data=None)

Use the generator model to generate high res data from low res input. This is the public generate function.

Parameters:

• low_res (np.ndarray) – Low-resolution input data to the 1st step spatial GAN, which is a 4D array of shape: (temporal, spatial_1, spatial_2, n_features). This should include all of the self.lr_features which is a concatenation of both the solar and wind spatial model features. The topography feature might be removed from this input and present in the exogenous_data input.

• norm_in (bool) – Flag to normalize low_res input data if the self.means, self.stdevs attributes are available. The generator should always received normalized data with mean=0 stdev=1.

• un_norm_out (bool) – Flag to un-normalize synthetically generated output data to physical units

• exogenous_data (ExoData) – class:ExoData object with data arrays for each exogenous data step. Each array has 3D or 4D shape: (spatial_1, spatial_2, n_features) (temporal, spatial_1, spatial_2, n_features) It’s assumed that the spatial_solar_models do not require exogenous_data and only use clearsky_ratio.

Returns:

hi_res (ndarray) – Synthetically generated high-resolution data output from the 2nd step (spatio)temporal GAN with a 5D array shape: (1, spatial_1, spatial_2, n_temporal, n_features)

temporal_pad(hi_res, mode='reflect')

Optionally add temporal padding to the 5D generated output array

Parameters:

• hi_res (ndarray) – Synthetically generated high-resolution data output from the 2nd step (spatio)temporal GAN with a 5D array shape: (1, spatial_1, spatial_2, n_temporal, n_features)

• mode (str) – Padding mode for np.pad()

Returns:

hi_res (ndarray) – Synthetically generated high-resolution data output from the 2nd step (spatio)temporal GAN with a 5D array shape: (1, spatial_1, spatial_2, n_temporal, n_features) With the temporal axis padded with self._temporal_pad on either side.

classmethod load(spatial_solar_model_dirs, spatial_wind_model_dirs, temporal_solar_model_dirs, t_enhance=None, temporal_pad=0, verbose=True)

Load the GANs with its sub-networks from a previously saved-to output directory.

Parameters:

• spatial_solar_model_dirs (str | list | tuple) – An ordered list/tuple of one or more directories containing trained + saved Sup3rGan models created using the Sup3rGan.save() method. This must contain only spatial solar models that input/output 4D tensors.

• spatial_wind_model_dirs (str | list | tuple) – An ordered list/tuple of one or more directories containing trained + saved Sup3rGan models created using the Sup3rGan.save() method. This must contain only spatial wind models that input/output 4D tensors.

• temporal_solar_model_dirs (str | list | tuple) – An ordered list/tuple of one or more directories containing trained + saved Sup3rGan models created using the Sup3rGan.save() method. This must contain only (spatio)temporal solar models that input/output 5D tensors that are the concatenated output of the spatial_solar_models and the spatial_wind_models.

• t_enhance (int | None) – Optional argument to fix or update the temporal enhancement of the model. This can be used with temporal_pad to manipulate the output shape to match whatever padded shape the sup3r forward pass module expects.

• temporal_pad (int) – Optional reflected padding of the generated output array.

• verbose (bool) – Flag to log information about the loaded model.

Returns:

out (SolarMultiStepGan) – Returns a pretrained gan model that was previously saved to model_dirs

get_s_enhance_from_layers()

Compute factor by which model will enhance spatial resolution from layer attributes. Used in model training during high res coarsening

get_t_enhance_from_layers()

Compute factor by which model will enhance temporal resolution from layer attributes. Used in model training during high res coarsening

property hr_exo_features

Get list of high-resolution exogenous filter names the model uses. For the multi-step model, each entry in this list corresponds to one of the single-step models and is itself a list of hr_exo_features for that model.

property input_dims

Get dimension of model generator input. This is usually 4D for spatial models and 5D for spatiotemporal models. This gives the input to the first step if the model is multi-step. Returns 5 for linear models.

Returns:

int

property input_resolution

‘…km’, ‘temporal’: ‘…min’}. The numbers are required to be integers in the units specified. The units are not strict as long as the resolution of the exogenous data, when extracting exogenous data, is specified in the same units.

Type:

Resolution of input data. Given as a dictionary {‘spatial’

property is_4d

Check if model expects spatial only input

property is_5d

Check if model expects spatiotemporal input

property means

Get the data normalization mean values. This is a tuple of means from all models.

Returns:

tuple | np.ndarray

property model_params

Get a tuple of model parameters for all models

Returns:

tuple

property models

Get an ordered tuple of the Sup3rGan models that are part of this MultiStepGan

property output_resolution

Resolution of output data. Given as a dictionary {‘spatial’: ‘…km’, ‘temporal’: ‘…min’}. This is computed from the input resolution and the enhancement factors.

property s_enhance

Factor by which model will enhance spatial resolution. Used in model training during high res coarsening

save_params(out_dir)

Parameters:

out_dir (str) – Directory to save linear model params. This directory will be created if it does not already exist.

static seed(s=0)

Set the random seed for reproducible results.

Parameters:

s (int) – Random seed

set_model_params(**kwargs)

Set parameters used for training the model

Parameters:

kwargs (dict) – Keyword arguments including ‘input_resolution’, ‘lr_features’, ‘hr_exo_features’, ‘hr_out_features’, ‘smoothed_features’, ‘s_enhance’, ‘t_enhance’, ‘smoothing’

property smoothed_features

Get the list of smoothed input feature names that the generative model was trained on.

property smoothing

Value of smoothing parameter used in gaussian filtering of coarsened high res data.

property stdevs

Get the data normalization standard deviation values. This is a tuple of stdevs from all models.

Returns:

tuple | np.ndarray

property t_enhance

Factor by which model will enhance temporal resolution. Used in model training during high res coarsening

property version_record

Get a tuple of version records from all models

Returns:

tuple