Required Education : Ingénier or M2
Start date : 1 September 2022
Mission duration : 12 months
Deadline for applications : 1 September 2022
The project aims at making use of Earth Observation (EO) data from satellite imagery to set up bi-dimensional (2D) hydrodynamics numerical models on selected catchments of interest. These models will be further used in the context of remote sensing (RS)-derived flood extent and flood forecasting.
Several international initiatives have joined efforts in monitoring and modelling river hydrodynamics, in order to provide Decision Support System services with accurate flood. Currently, decision making and risk assessment both rely on the use of RS-EO products and hydrodynamic models.
RS products that have tremendously developed over the last decade, arguably offer opportunities to improve our ability to monitor and forecast flooding. The observation of inland waters benefits from several altimetry mission efforts that provide along-track WSE from nadir altimeters, from TOPEX/Poseidon (1992-2006), to the Sentinel program (i.e. Sentinel-3A (2016), 3B (2018) and Sentinel-6 (2020)). Additionally, the up-coming SWOT mission uses wide-swath altimetry technology and will enable the retrieval of river level, width and slope at an unprecedented resolution with a global coverage and a 21-day repeat orbit. SAR data have increasingly become one of the most efficient ways to map and monitor flood extents in near-real time over large areas, due to their all-weather day-and-night imaging. SAR-derived flood extent maps are an important information source for an effective flood disaster management by supporting humanitarian relief organizations and decision makers. RS data provide valuable distributed calibration and validation data for hydraulic models of fluvial flood processes and support the derivation of spatially accurate hazard maps in terms of flood prevention activities, insurance risk management, and spatial planning. The merits of RS-derived flood information can therefore be greatly enhanced by combining it with other data types, from various EO satellite missions and from numerical modelling. For instance, one of the major limitations of satellite EO of flood extent is the rather limited revisit time and the model can be used as a time interpolator between EO data.
River hydrodynamic models, that solve the Shallow Water (SW) equations (depth-averaged Navier Stokes equations), are used to predict river water surface elevation (WSE) and discharge and therefore to allow assessing flood risks. Additionally, the combination of RS data with hydrodynamic models through Data Assimilation (DA) algorithms has therefore been investigated in the literature to overcome the limitations of both incomplete and uncertain sources of knowledge on the river and flood plain dynamics. However, the implementation of such hydrodynamic numerical models requires knowledge of topography and bathymetry data, description of the land cover (vegetation class, urbanization) as well as water level/discharge observations, both for imposing boundary conditions and for calibrating unknown parameters (e.g. friction). Inaccurate input data and parameters represent sources of uncertainty in hydrodynamic simulations that necessarily translate into uncertainties in the predictions and reduce model performance. Flood modelling quality in urban and peri-urban areas rely on very precise Digital Elevation Models (DEMs), traditionally generated by local LiDAR or field surveys. A product widely used by the community is MERIT, that even though it is at quite low resolution, presents a global coverage. For more detailed data, topography and bathymetry high quality data from IGN LiDAR campaign at high resolution (1m or 5m) are available over the French territory (https://geoservices.ign.fr/rgealti). High-resolution (HR) to very high-resolution (VHR) products for DEM can also be generated from optical imagery such as SPOT, Pléïades, Sentinel-2 or Worldview. Along the last decades, DEMs derived from TanDEM-X or TerraSAR-X acquisitions have been evaluated in urban areas and applied in flood modelling. Given the abrupt nature of high-density built-up areas, tri-stereo acquisitions provide a substantial improvement on DEM completeness. Flood modelling quality also depends on the description of friction that is related to the vegetation classes, for instance as described in the OSO-IOTA2 product for France or by ESA at global scale (https://viewer.esa-worldcover.org/worldcover/).
The objective of the Space2Hydro proposal is to set up 2D hydrodynamic models using HR and VHR topographical data made available from the EO community with the software Telemac (http://opentelemac.org) on catchments of interest where RS flood observations are available.
- Work Plan:
CERFACS has been working with Telemac for more than 10 years, focusing on the development of peripheral functionalities dedicated to uncertainty quantification and reduction with ensemble-based DA techniques. Telemac is an open-source code, developed by a Consortium that gathers French and international partners, among which EDF, ARTELIA and CERFACS. While having gained a considerable expertise in the use of Telemac for flood modelling over the last decade, CERFACS has so far, a limited expertise in building hydrodynamic models for new test cases or using new input data sets. Indeed, UQ and DA strategies have usually been applied to existing test cases provided by EDF, SCHAPI or SPCs in the context of collaborative projects. The hydrology team at CERFACS is now willing to gain expertise in this domain in order to be able to demonstrate the merits of using high-quality data acquired from space, e.g. for topography, land cover description and discharge estimation. These modelling efforts will allow us to meet the challenges of several on-going projects at CERFACS related to hydrology and remote sensing.
We describe below the organization of the Space2Hydro project in 4 tasks detailed with associated expected skills:
- Gathering high resolution topography and bathymetry community product (MERIT) or space data from Pléiades, TerraSAR-X/TanDEM-X missions over the Garonne catchment (France), Po (Italy), Mahanadu (India), Bramaputra (India), Gange (India), Phnom Penh (Cambodia), Congo (Africa). This is a preliminary list of sites of interest, but this must be completed and eventually reconsidered depending on the SWIFTS funding. This requires to get in touch with colleagues at INRAE, LEGOS, ARTELIA, first to gather in situ data but also to focus the efforts on catchments of interest. This step requires skills on remote sensing data management and visualization.
- Gathering of in-situ data on discharge for boundary condition description when available (gauged catchments) and selection of time-series for significant flood events over all selected events. This requites skills in hydrodynamics modelling.
- Setting up Telemac models with meshing, calibration of friction and validation steps. This step requires a good knowledge of Telemac2D, mesh generation, and visualization tools of Telemac outputs files.
- Linking up with previously described research project for the use of the developed test cases. This step requires good collaborative and communication skills.
This technical proposal gathers CERFACS, CNES, EDF and Airbus; a consortium that has a complementary expertise in hydrodynamic modelling, remote sensing, data assimilation, machine/deep learning and computational sciences. The project benefits from a facilitated access to remote sensing data (CNES, Airbus) and computational resources at CERFACS.
- H. Nguyen, S. Ricci, C. Fatras, et al. Improvement of flood extent representation with remote sensing data and data assimilation. IEEE Trans. Geosci. Remote Sens., pages 1–22, 2022
- Mirouze, I. and Ricci, S. (2021) SMURF : System for Modelling with Uncertainty Reduction, and Forecasting, Journal of Open Research Software, 9 (1), doi :10.5334/jors.312
- Emery, C. M., Ricci, S. and Piacentini, A. (2021), ToolBox SWOT User's Guide, CECI, Université de Toulouse, CNRS, CERFACS, Toulouse, France – TR-CMGC-21-153, Technical report
- Mirouze, I., Ricci, S. and Goutal, N. (2019)a. The impact of observation densification in an ensemble Kalman Filter., Telemac User Conférence, Toulouse, France, 15-17 Oct., 2019
- Mirouze, I., Ricci, S., Goutal, N., Martin, L. and Desroche, D. (2019)b. Generation and assimilation of synthetical SWOT data with an Ensemble Kalman filter for 1D and 2D hydrodynamics models, ST SWOT, Bordeaux June 2019.
- Peña-Luque, S. Ferrant, M.C.R. Cordeiro, et al. Sentinel-1 &2 mul titemporal water surface detection accuracies, evaluated at regional and reservoirs level. Remote Sens., 13(16), 2021.
- Kettig, S. Baillarin, G. Blanchet, C. Taillan, S. Ricci, T.-H. Nguyen, T. Huang, A. Altinok, N.-T. Chung, G. Valladeai, R. Goeury, A. Roumagnac. The SCO FloodDAM project: New observing strategies for flood detection, alert and rapid mapping, IGARSS 11-16 July 2021.
- Yamazaki et al., “A high-accuracy map of global terrain elevations,” 1562 Geophys. Res. Lett., vol. 44, no. 11, pp. 5844–5853, Jun. 2017.
- C. Mason, G. J.-P. Schumann, J. C. Neal, J. Garcia-Pintado, and P. D. Bates, “Automatic near real-time selection of flood water levels from high resolution synthetic aperture radar images for assimilation into hydraulic models: A case study,” Remote Sens. Environ., vol. 124, pp. 705–716, Sep. 2012.
- Giustarini et al., “Assimilating SAR-derived water level data into a hydraulic model: A case study,” Hydrol. Earth Syst. Sci., vol. 15, no. 7, pp. 2349–2365, Jul. 2011.
- Derksen, D.; Inglada, J.; Michel, J. Geometry Aware Evaluation of Handcrafted Superpixel-Based Features and Convolutional Neural Networks for Land Cover Mapping Using Satellite Imagery. Remote Sens. 2020, 12, 513. https://doi.org/10.3390/rs12030513
- Dottori, L. Mentaschi, A. Bianchi, L. Alfieri, and L. Feyen. Adapt- ing to rising river flood risk in the eu under climate change. Joint Research Centre (JRC), 2020.
- Wood, R. Hostache, J. Neal, et al. Calibration of channel depth and friction parameters in the LISFLOOD-FP hydraulic model using medium-resolution SAR data and identifiability techniques. Hydrol. Earth Syst. Sci., 20(12):4983–4997, 2016.
- Hostache, P. Matgen, G. Schumann, C. Puech, L. Hoffmann, and L. Pfister. Water level estimation and reduction of hydraulic model calibration uncertainties using satellite SAR images of floods. IEEE Trans. Geosci. Remote Sens., 47(2):431–441, 2009.
- Schumann, P.D. Bates, M. S. Horritt, P. Matgen, and F. Pappenberger. Progress in integration of remote sensing–derived flood extent and stage data and hydraulic models. Reviews of Geophysics, 47(4), 2009.
- Voigt, T. Kemper, T. Riedlinger, et al. Satellite image analysis for disaster and crisis-management support. IEEE Trans. Geosci. Remote Sens., 45(6):1520–1528, 2007.
- de Moel, J. van Alphen, and J.C.J.H. Aerts. Flood maps in europe–methods, availability and use. Nat. Hazards Earth Syst. Sci., 9(2):289–301, 2009.