Options
Simulating Flood‐Induced Riverbed Transience Using Unmanned Aerial Vehicles, Physically Based Hydrological Modeling, and the Ensemble Kalman Filter
Auteur(s)
Tang, Qi
Kurtz, W.
Vereecken, H.
Hendricks Franssen, Harrie-Jan
Date de parution
2018-11
In
Water Resources Research
Vol.
11
No
54
De la page
9342
A la page
9363
Revu par les pairs
1
Résumé
Abstract Flood events can change the riverbed topography as well as the riverbed texture and structure,
which in turn can influence the riverbed hydraulic conductivity (Krb) and river-aquifer exchange fluxes. A
major flood event occurred in the Emme River in Switzerland in 2014, with major implications for the riverbed
structure. The event was simulated with the fully integrated hydrological model HydroGeoSphere. The aim
was to investigate the effect of the spatial and temporal variability of riverbed topography and Krb on
predictions of hydraulic states and fluxes and to test whether data assimilation (DA) based on the ensemble
Kalman filter (EnKF) can better reproduce flood-induced changes to hydraulic states and parameters with the
help of riverbed topography changes recorded with an unmanned aerial vehicle (UAV) and through-water
photogrammetry. The performance of DA was assessed by evaluating the reproduction of the hydraulic
states for the year 2015. While the prediction of surface water discharge was not affected much by the
changes in riverbed topography and in Krb, using the UAV-derived postflood instead of the preflood riverbed
topography reduced the root-mean-square error of predicted heads (RMSE [h]) by 24%. If, in addition to
using the postflood riverbed topography, also Krb and aquifer hydraulic conductivity (Kaq) were updated
through DA after the flood, the RMSE (h) was reduced by 55%. We demonstrate how updating of Krb and Kaq
based on EnKF and UAV-based observations of riverbed topography transience after a major flood event
strongly improve predictions of postflood hydraulic states.
which in turn can influence the riverbed hydraulic conductivity (Krb) and river-aquifer exchange fluxes. A
major flood event occurred in the Emme River in Switzerland in 2014, with major implications for the riverbed
structure. The event was simulated with the fully integrated hydrological model HydroGeoSphere. The aim
was to investigate the effect of the spatial and temporal variability of riverbed topography and Krb on
predictions of hydraulic states and fluxes and to test whether data assimilation (DA) based on the ensemble
Kalman filter (EnKF) can better reproduce flood-induced changes to hydraulic states and parameters with the
help of riverbed topography changes recorded with an unmanned aerial vehicle (UAV) and through-water
photogrammetry. The performance of DA was assessed by evaluating the reproduction of the hydraulic
states for the year 2015. While the prediction of surface water discharge was not affected much by the
changes in riverbed topography and in Krb, using the UAV-derived postflood instead of the preflood riverbed
topography reduced the root-mean-square error of predicted heads (RMSE [h]) by 24%. If, in addition to
using the postflood riverbed topography, also Krb and aquifer hydraulic conductivity (Kaq) were updated
through DA after the flood, the RMSE (h) was reduced by 55%. We demonstrate how updating of Krb and Kaq
based on EnKF and UAV-based observations of riverbed topography transience after a major flood event
strongly improve predictions of postflood hydraulic states.
Identifiants
Type de publication
journal article
Dossier(s) à télécharger
