Efficient number of calibrated cross sections bottom levels on a hydrodynamic model using the SCE-UA algorithm. Case study: Madeira River

Abstract

Hydrodynamic models are important tools for simulating river water level and flow. A considerable fraction of the hydrodynamic model errors are related to parameters uncertainties. As cross sections bottom levels considerably affect water level simulation, this parameter has to be well estimated for flood studies. Automatic calibration performance and processing time depend on the search space dimension, which is related to the number of calibrated parameters. This paper shows the application of the Shuffled Complex Evolution (SCE-UA) optimization algorithm to assess the number of cross sections bottom levels used in calibration. Also was evaluated the extent of algorithm exploration regarding computational processing time and accuracy. It was tested the calibration of 2, 4, 7 and 10 cross sections bottom levels (2PAR, 4PAR, 7PAR and 10PAR calibration configurations) of a 1,100 km reach of the Madeira River. 7PAR and 10PAR representation had better fitness (lower objective function value) on cross sections used for calibration; however, the error on other cross sections (2 validation gauging stations) was higher than 2PAR and 4PAR calibration. The short number (5) of gauging stations used in calibration has limited the number of calibrated parameters to represent adequately the river level profile. Finally, this paper shows a contribution for the parsimonious selection of parameters regarding the spatial distribution of observation sites used in calibration.

Assimilation of Satellite Altimetry Data for Effective River Bathymetry

“…The KF method barely corrected the riverbed elevation far from the observation sites, and the bathymetry errors remained high (as discussed in section ). The SCE‐UA poor performance assimilating JASON 2 appeared to be caused by overfitting (Brêda et al, ), because there was an excessive number of calibrated parameters (12 at least) compared to the number of virtual stations (4). Because the main source of WSE errors is clearly related to underestimation of the riverbed elevation, a simple interpolation method was able to achieve satisfactory results.…”

“…The SCE‐UA parameters followed the recommendations of Duan et al (), except for the number of complexes, which varies with the number of model parameters of the optimization process. The number of complexes is related to the algorithm probability of finding the global optimum, and it was selected in accordance to Brêda et al ().…”

On the discretization of river networks for large scale hydrologic-hydrodynamic models

“…Moreover, a key improvement of recent studies, albeit initially proposed for 2D flood inundation modeling, is the use of the explicit local inertia approximation of the shallow water equations (Bates et al, 2010), which has been an interesting alternative to the full Saint-Venant hydrodynamic equations due to its relative efficiency and easier implementation. The ability to speed up model computations together with code parallelization (Yamazaki et al, 2013) can be crucial either for handling finer model resolutions or when several model runs are made necessary, for instance, for uncertainty assessment in ensemble flood forecasting (e.g., Pappenberger et al, 2005), data assimilation (e.g., Brêda et al, 2017) or parameter estimation (e.g., Dung et al, 2011).…”