Hybrid model

class hy2dl.modelzoo.hybrid.Hybrid(cfg: Config)

Bases: Module

Wrapper to combine a deep learning model with a conceptual hydrological models [1].

Hybrid model in which a conceptual hydrological model is parameterized using a LSTM network.

Parameters:

cfg (Config) – Configuration file.

References

[1]

Acuña Espinoza, E., Loritz, R., Álvarez Chaves, M., Bäuerle, N., and Ehret, U.: To Bucket or not to Bucket? Analyzing the performance and interpretability of hybrid hydrological models with dynamic parameterization, Hydrology and Earth System Sciences, 28, 2705–2719, https://doi.org/10.5194/hess-28-2705-2024, 2024.

forward(sample: dict[str, Tensor | dict[str, Tensor]]) → dict[str, Tensor]

Forward pass on hybrid model.

In the forward pass, each element of the batch is associated with a basin. Therefore, the conceptual model is done to run multiple basins in parallel, and also multiple entities of the model at the same time.

Parameters:

sample (dict[str, torch.Tensor]) – Dictionary with the different tensors that will be used for the forward pass.

Returns:

pred

Return type:

dict[str, torch.Tensor]