•Study about discharge inversion in the context of remotly sensed observations of rivers.•The concept of an effective bathymetry-friction pair is introduced.•Several inverse formulations are proposed ...from the zero inertia shallow water equations, with an inversion algorithm.•Inversions of topography and friction pairs, and discharge are performed and analyzed on a set of synthetic and real river flows.
The future SWOT mission (Surface Water and Ocean Topography) will provide cartographic measurements of inland water surfaces (elevation, widths and slope) at an unprecedented spatial and temporal resolution. Given synthetic SWOT like data, forward flow models of hierarchical-complexity are revisited and few inverse formulations are derived and assessed for retrieving the river low flow bathymetry, roughness and discharge (A0,K,Q). The concept of an effective low flow bathymetry A0 (the real one being never observed) and roughness K, hence an effective river dynamics description, is introduced. The few inverse models elaborated for inferring (A0,K,Q) are analyzed in two contexts: (1) only remotely sensed observations of the water surface (surface elevation, width and slope) are available; (2) one additional water depth measurement (or estimate) is available. The inverse models elaborated are independent of data acquisition dynamics; they are assessed on 91 synthetic test cases sampling a wide range of steady-state river flows (the Froude number varying between 0.05 and 0.5 for 1km reaches) and in the case of a flood on the Garonne River (France) characterized by large spatio-temporal variabilities. It is demonstrated that the most complete shallow-water like model allowing to separate the roughness and bathymetry terms is the so-called low Froude model. In Case (1), the resulting RMSE on infered discharges are on the order of 15% for first guess errors larger than 50%. An important feature of the present inverse methods is the fairly good accuracy of the discharge Q obtained, while the identified roughness coefficient K includes the measurement errors and the misfit of physics between the real flow and the hypothesis on which the inverse models rely; the later neglecting the unobserved temporal variations of the flow and the inertia effects. A compensation phenomena between the indentifiedvalues of K and the unobserved bathymetry A0 is highlighted, while the present inverse models lead to an effective river dynamics model that is accurate in the range of the discharge variability observed. In Case (2), the effective bathymetry profile for 80km of the Garonne River is retrieved with 1% relative error only. Next, accurate effective topography-friction pairs and also discharge can be inferred. Finally, defining river reaches from the observation grid tends to average the river properties in each reach, hence tends to smooth the hydraulic variability.
•Unit-width-discharge (Q/w) explains spatiotemporal current velocity variation.•Unit water surface area CO2 flux (FCO2) varies closely with river hydraulics.•Unit channel length CO2 flux (FlCO2) ...depends largely on river discharge.•River hydraulics explains larger part of CO2 flux variation than pCO2.•Current velocity is the most important factor determining k600 in large rivers.
Although CO2 emission from river surfaces is largely hydrology-related, the hydrological impact on riverine CO2 emission is rarely characterized in previous studies. Relying on daily hydrology and monthly-resolved partial pressure of water dissolved CO2 (pCO2), this study characterized the influence of spatiotemporal hydraulic variability on CO2 emission from confined low-gradient rivers of the Yangtze. Results indicate that unit-width-discharge (Q/w) explained well along-mainstem, inter-tributary and intra-annual (monthly) variations of key hydraulic characteristics (e.g., current velocity and water depth). CO2 flux normalized to unit water surface area (FCO2) varied closely with channel hydraulics (especially current velocity), which explained 48.9–67.9% and 68.2–72.0% of the spatial and temporal FCO2 variations, respectively, larger than the parts explained by pCO2 (32.1–51.1% and 28.0–31.8%). Despite high sensitivity of FCO2 to river hydraulics, CO2 flux normalized to unit channel length (FlCO2) depended primarily on channel discharge, which determines ultimately the amount of CO2 exported from the catchment and emitted to the atmosphere, regardless of the complex interplays among channel hydraulic characteristics. Limited variances under different k600-v parameterizations indicate reliability of the major conclusions in this paper.
In this study, we elaborate on and evaluate a new reduced basis method for model reduction of the shallow water equations using Proper Orthogonal Decomposition (POD) and artificial Neural Networks ...(NNs). The method begins with the POD technique to construct reduced bases from high-resolution solutions, followed by training two deep NNs to learn associated coefficients in the reduced bases. The approach follows an offline–online strategy: the POD reduced basis, along with the training of the NNs, is performed in an offline stage, and then the surrogate model can be used in an online stage for real-time predictions. The method takes into account the POD-based projection error, enabling the attainment of higher accuracy while preserving a limited number of POD modes, even in the delicate situation of convection-dominated flow problems. This point is crucial in our approach since it enables to limit the output dimension of the NNs, thus providing the opportunity to employ smaller NNs (with less parameters). This may lead to the utilization of potentially smaller datasets (i.e., the snapshots), better generalization of the obtained model, and simpler explainability. The process is non-intrusive: it does not require opening the high-resolution model code. The method is evaluated on a real-world test case aimed at simulating inundation of the Aude river (Southern France). The results show that the proposed method provides satisfying accuracy for the hydraulic variables (water elevation, discharge) compared to the reference high-resolution 2D shallow water model, with quite small dimension NNs. Overall, the method is promising, particularly for performing real-time simulations of large floodplains hydrodynamics.
In mountainous torrents of the Mediterranean environment the riparian vegetation is strongly influenced by the presence of engineering control works, since these structures bring heavy modifications ...in channel geometry, hydraulic regime and bed sediment size. Previous investigations have shown high linear correlations between physical (section shape, profile slope, specific discharge, surface and subsurface size of the channel bed) and vegetation (development, structure and biodiversity) indicators in headwater channels with check dams of Calabrian (Southern Italy) torrents. Based on these findings, this study applies multivariate statistical techniques (Principal Component Analysis and Partial Least Square Regression) to identify in the same study headwaters new synthetic explanatory variables, representative of the different transects (upstream, downstream or intermediate, compared to the check dam location) and develop predictive models of riparian vegetation characteristics.
The Principal Component Analysis has provided a simple parameter (the first Principal Component, explaining about 60% of the total variance), which is able to discriminate the physical and vegetal characteristics of the different transects close to check dams, thus reducing the large number of factors influencing the fluvial processes. Moreover, cover, height and transversal variability of riparian vegetation have a very high influence (loadings over 0.73) on this component, while its biodiversity is correlated to the second Principal Component (loadings over 0.63). The Partial Least Square Regression has shown that it is possible to estimate with fair accuracy (minimum r2 of 0.70) the development, structure as well as transversal variability of the riparian vegetation, starting from the physical features of the channel. These models may be important in the planning steps of new check dams, since their effects on the development and growth of vegetation upstream and downstream can be forecasted before their installation, at least for the quantification of the order of magnitude of the check dam impacts on torrent ecology.
Display omitted
•Synthetic indicators identify physical and vegetation characteristics close to check dams.•Channel geometry, bed sediment size and riparian vegetation depend on transect location.•Changes in physical and vegetal parameters are less significant among torrents.•PCA identifies a new synthetic indicator of physical and vegetation characteristics.•PLSR provides predictions of the vegetation characteristics close to check dams.
This paper presents an innovative deep learning (DL) framework to (a) automatically identify river geometry and flood extent, and (b) predict river flooding depth. To do that, U-Net, an advanced ...convolutional neural network (CNN), was modified and given the designation of U-NetRiver. With the modification, the model received an input composite image with two bands of ground elevation and flooding discharge, and the output was water depth. The model was trained and validated based on the outputs from iRIC (a two-dimensional hydraulic model) for a segment of the Green River in the state of Utah. The results showed that the U-NetRiver could identify the river shape and wetted areas for flooded regions automatically. The maximum difference of predicted river depth obtained from U-NetRiver and the one obtained from the hydraulic model was 2.7 m. This result suggests a 29% improvement in prediction of the maximum flood depth in the river.
•U-NetRiver, could successfully predict river flooding depth and extent.•The model accurately and automatically detected river geometry and shape.•The model is more objective as it minimized human involvement in flood prediction.•The model decreased the error in predicted river maximum flooding depth by 29%.
Hydrokinetic energy resource assessment is a crucial prerequisite for strategic turbine deployment and energy extraction. Despite advancements in analytical tools, resource assessment is often ...completed without detailed investigation of spatial and temporal flow variation and implications on optimal turbine placement. A case study was conducted on the Rivière Rouge, Québec, Canada to estimate the hydrokinetic energy resource, to locate the optimal turbine placement, and to study the impact of seasonal flow variation. The primary optimal turbine location did not change, but the second, third, and fourth optimal locations were impacted. Assuming a hypothetical deployment of one turbine with a 1 m2 swept area, the theoretical hydrokinetic energy resource for the site was 21.8 MWh per year in the optimal turbine locations and 6.2 MWh per year using the reach-averaged velocity. This difference illustrates the need to consider the entire velocity flow field in hydrokinetic energy assessments. To conduct the assessment, field data were collected with an acoustic Doppler current profiler and a global positioning system for hydrodynamic model generation, calibration, and validation using the software TELEMAC-2D. The mean absolute percentage errors of the model in the areas of interest were 14.8% for calibration and 22.9% and 19.4% for validation.
•Optimal hydrokinetic turbine placement varied temporally with flow condition.•Spatially, average and maximum hydrokinetic energy values varied greatly.•Entire velocity flow field must be utilized in hydrokinetic assessments.•High flow period contributed 72% to the total annual energy production.•Spatially intense data collection techniques required for hydrokinetic assessments.
Computational Fluid Dynamics (CFD) is increasingly used to study a wide variety of complex Environmental Fluid Mechanics (EFM) processes, such as water flow and turbulent mixing of contaminants in ...rivers and estuaries and wind flow and air pollution dispersion in urban areas. However, the accuracy and reliability of CFD modeling and the correct use of CFD results can easily be compromised. In 2006, Jakeman et al. set out ten iterative steps of good disciplined model practice to develop purposeful, credible models from data and a priori knowledge, in consort with end-users, with every stage open to critical review and revision (Jakeman et al., 2006). This paper discusses the application of the ten-steps approach to CFD for EFM in three parts. In the first part, the existing best practice guidelines for CFD applications in this area are reviewed and positioned in the ten-steps framework. The second and third part present a retrospective analysis of two case studies in the light of the ten-steps approach: (1) contaminant dispersion due to transverse turbulent mixing in a shallow water flow and (2) coupled urban wind flow and indoor natural ventilation of the Amsterdam ArenA football stadium. It is shown that the existing best practice guidelines for CFD mainly focus on the last steps in the ten-steps framework. The reasons for this focus are outlined and the value of the additional – preceding – steps is discussed. The retrospective analysis of the case studies indicates that the ten-steps approach is very well applicable to CFD for EFM and that it provides a comprehensive framework that encompasses and extends the existing best practice guidelines.
Display omitted
► Application of the ten-steps approach to CFD in Environmental Fluid Mechanics (EFM). ► Existing best practice guidelines (BPG) in CFD mainly focus on only last three steps. ► Reason: specific character and disadvantages of CFD: need for verification and validation. ► More attention should be given to the other steps in CFD studies of EFM. ► Two case studies show that ten-steps approach is very well applicable to CFD in EFM and that it extends existing BPG.
Floods, bridge scour, and flood-associated loads have caused over sixty percent of bridge failures in the U.S. Current practices for the vulnerability assessment of instream bridges under the effect ...of such flood largely rely on qualitative methods, such as visual inspection, without considering uncertainties associated with structural behaviors and flood loads. Recently, numerical methods have been investigated to quantitatively consider such uncertainty effects by adapting fragility analysis concept that has been well established in the earthquake engineering area. However, river hydraulics, geotechnical uncertainties of foundation, variable scour-depth effects, and their significance in structural fragility of bridges have rarely been systematically investigated. This study proposes a comprehensive fragility analysis framework that can effectively incorporate both flow hydraulics and geotechnical uncertainties, in addition to commonly considered components in flood-fragility analysis of bridges. The significance of flow hydraulics and geotechnical uncertainties has been demonstrated through a real-bridge case study. Conventional fragility curves with maximum scour depth may not represent actual vulnerability during floods, as the scour may not reach to the maximum in many cases. Therefore, fragility surface with two intensity measures, i.e. flow discharges and scour depths, is introduced for real-time vulnerability assessment during floods in this study.