loading page

Multiscale Learnable Physical Modeling and Data Assimilation Framework: Application to High-Resolution Regionalized Hydrological Simulation of Flash Flood
  • +3
  • Ngo Nghi Truyen Huynh,
  • Pierre-André Garambois,
  • François Colleoni,
  • Benjamin Renard,
  • Jérôme Monnier,
  • Hélène Roux
Ngo Nghi Truyen Huynh
INRAE, Aix Marseille Univ, RECOVER
Author Profile
Pierre-André Garambois
INRAE, Aix Marseille Univ, RECOVER

Corresponding Author:[email protected]

Author Profile
François Colleoni
INRAE, Aix Marseille Univ, RECOVER
Author Profile
Benjamin Renard
Author Profile
Jérôme Monnier
INSA Toulouse
Author Profile
Hélène Roux
Institut de Mécaniques des Fluides de Toulouse
Author Profile


To advance the discovery of scale-relevant hydrological laws while better exploiting massive multi-source data, combining machine learning and process-based modeling is compelling, as recently demonstrated in lumped hydrological modeling. This article introduces MLPM-PR, a new and powerful framework standing for Multiscale spatially distributed Learnable Physical Modeling and learnable Parameter Regionalization with data assimilation. MLPM-PR crucially builds on a differentiable model that couples (i) two neural networks for processes learning and parameters regionalization, (ii) a grid-based differentiable conceptual hydrological model, and (iii) a simple kinematic wave routing. The approach is tested on a challenging flash flood-prone multi-catchment modeling setup at high spatio-temporal resolution (1km, 1h). Discharge prediction performances highlight the accuracy and robustness of MLPM-PR, both spatially and temporally compared to classical approaches. The physical interpretability of spatially distributed parameters and internal states shows the nuanced behavior of the hybrid model and its adaptability to diverse hydrological responses.