environmentaltools.temporal.ensemble_dt

environmentaltools.temporal.ensemble_dt(models: dict, percentiles='equally')[source]

Compute ensemble of temporal dependency parameters from multiple models.

Combines VAR model parameters from multiple Regional Climate Models (RCMs) or different model realizations using weighted or equal averaging.

Parameters:

  • models (dict) – Dictionary with model identifiers as keys and file paths as values. Each file should contain temporal dependency parameters (B, Q matrices)

  • percentiles (str or list, optional) –

    Weighting scheme for ensemble:

    • ’equally’: Equal weight for all models (default)

    • list of float: Weight for each model (must sum to 1)

Returns:

Ensemble temporal dependency parameters containing:

  • Bnp.ndarray

    Averaged autoregression coefficient matrix

  • Qnp.ndarray

    Averaged covariance matrix of residuals

  • idint

    Model order

Return type:

dict

Notes

The ensemble is computed by:

  1. Reading B and Q matrices from each model

  2. Aligning matrix dimensions (padding with zeros if needed)

  3. Computing weighted average (or simple average if ‘equally’)

For equal weights: B_ensemble = mean(B_i)

For custom weights: B_ensemble = sum(w_i * B_i)

Examples

>>> models = {
...     'model1': 'path/to/model1.json',
...     'model2': 'path/to/model2.json',
...     'model3': 'path/to/model3.json'
... }
>>> # Equal weighting
>>> ensemble = ensemble_dt(models, percentiles='equally')
>>>
>>> # Custom weighting
>>> ensemble = ensemble_dt(models, percentiles=[0.5, 0.3, 0.2])