To Bucket or not to Bucket? Analyzing the performance and interpretability of hybrid hydrological models with dynamic parameterization

Hydrological hybrid models have been proposed as an option to combine the enhanced performance of deep learning methods with the interpretability of process-based models. Among the various hybrid methods available, the dynamic parameterization of conceptual models using LSTM networks has shown high potential. We explored this method further to evaluate specifically if the flexibility given by the dynamic parameterization overwrites the physical interpretability of the process-based part. We conducted our study using a subset of CAMELS-GB dataset. First, we show that the hybrid model can reach state-of-the-art performance, fully comparable with a regional LSTM, and surpassing the performance of conceptual models in the same area. We then modified the conceptual model structure to assess if the dynamic parameterization can compensate for structural deficiencies of the model. Our results demonstrated the ability of the deep learning method to effectively compensate for deficiencies and implausible model structures in the hydrological models. This indicates that the hydrological model did not give a strong enough regularization to drop the hybrid model's performance. A model selection based purely on the performance to predict streamflow, for this type of hybrid model, is hence not advisable. However, this does not entail that such hybrid models cannot be used to gain a better understanding of a hydrological system by studying other hydrological fluxes and states than discharge. Comparisons with external data, as well as the internal functioning of the hybrid model, reiterate that if a well-tested model architecture is combined with a LSTM, the deep learning model can learn to operate the process-based model in a consistent matter. In conclusion, this study demonstrated that hybrid models, if set up cautiously, can combine the enhanced performance of deep learning methods while maintaining good interpretability in the process-based part.