Estimation of Inflows and Effective Channel from Satellite Observations: From Local to Hydrographic Network Scale

With the upcoming SWOT satellite mission, which should provide spatially dense river surface elevations, widths and slopes observations globally, comes the need to pertinently use such data into hydrodynamic models, from the reach to hydrographic network scales. Based on the HiVDI (Hierarchical Variational Discharge Inversion) modeling strategy ([1,2], DassFlow software¹), this work tackles the forward and inverse modeling capabilities of distributed channel parameters and inflows (in the 1D Saint-Venant model) from multisatellite observations of river surface. Several synthetic cases are designed to study fluvial and torrential flows signatures and assess the inference capabilities of model parameters (inflows, bathymetry, friction) given different observation patterns. Accurate inferences of both inflows and distributed channel parameters (bathymetry-friction) is achievable even with a minimum spatial observability between inflows. A sensitivity analysis of the inferences to prior hydraulic parameter values and to regularization parameters is performed. Next a real case is studied: 871km of the Negro river (Amazon basin) including complex multichannel reaches, 21 tributaries and backwater controls from major confluences. An effective modeling approach is proposed using (i) WS elevations from ENVISAT observations and dense in situ GPS flow lines, (ii) average river top widths from optical imagery, (iii) upstream and lateral flows from the MGB large-scale hydrological model [3]. The calibrated effective hydraulic model closely fits satellite altimetry observations of WS signatures and contains real-like spatial variabilities and flood wave propagations (frequential features analyzed with identifiability maps [2]). Synthetic SWOT observations are generated from the simulated flowlines and the identifiability of model parameters (579 bathymetry points, 17 friction patches and 22 upstream and lateral hydrographs) is tested using the HiVDI computational inverse method and given hydraulically coherent prior guesses and regularization parameter values. Inferences of channel parameters carried out on this fine hydraulic model applied at large scale give satisfying results considering the challenging inverse problems solved globally in space and time, even with noisy SWOT data. Inferences of spatially distributed temporal parameters (lateral inflows) give satisfying results as well, with even small scale hydrograph variations being infered accurately.

This study brings insights in:

1. the hydraulic visibility of multiple inflows hydrographs signature at large scale with SWOT;

2. the simultaneous identifiability of spatially distributed channel parameters and inflows by assimilation of satellite altimetry data;

3. the need to further taylor and scale hydrodynamic models and assimilation methods to improve potential information feedbacks to hydrological modules in integrated chains.

References:

[1] Larnier, Monnier, Garambois, Verley. (2019) River discharge and bathymetry estimations from SWOT altimetry measurements.

[2] Brisset, Monnier, Garambois, Roux. (2018) On the assimilation of altimetric data in 1d Saint-Venant river flow models. AWR, doi: 10.1016/j.advwatres.2018.06.004.

[3] Paiva, Buarque, Collischonn, et al. Large-scale hydrologic and hydrodynamic modeling of the amazon river basin. WRR, doi: 10.1002/wrcr.20067.