Tuesday 1st of july – from 14:00 to 19:00
Session A1 :Climate change: knowing and perceiving
Development and Implementation of a Compact, Non-Intrusive Water Surface Temperature Monitoring Station for River Ecosystems
ALESNARD-EVANGELISTA Eliot, RIOU Yannick, POISSON Jean-Christophe, FÉRET Thibaut, REBILLARD Jean-Pierre, MORIN Guillaume
France
Short abstract : Water temperature is a key indicator of river health, affecting both aquatic ecosystems and water quality. This project is led by vorteX-io in collaboration and with the support of the Agence de l’eau Adour-Garonne We introduce an innovative, compact, and non-intrusive station for continuous multi-parameter measurements, such as surface water temperature monitoring, specifically designed for diverse river settings and with limited ecological impact. Equipped with a high-temporal-resolution temperature sensor, this station enables precise temperature tracking without disrupting natural habitats and combines the assessement of water quantity and quality.
Our main objective was to create an affordable, portable solution that can be easily deployed on bridges over watercourses. Field tests at multiple sites confirmed the station’s accuracy, durability, and adaptability under varying environmental conditions. Preliminary results showed strong agreement with immerged thermal sensors, leading to the ongoing network densification with 150 stations. IIf the results meet the project expectations that are rather ambitious, we will consider a large-scale deployment.
This presentation will share key findings, practical insights, and the device’s potential contributions to river management, restoration efforts, and climate adaptation. Our approach promotes cost-effective, sustainable monitoring, providing valuable localized data for continuous assessment of river ecosystems on a large scale.
A Deep Learning approach for reconstructing suspended sediment load and forecasting under various climate change scenarios
HAMADENE Taha, NICOULAUD-GOUIN Valérie, LEPAGE Hugo, FOULADIRAD Mitra
France
In recent years, numerous studies have demonstrated the ability of machine learning (ML) and deep learning (DL) models to accurately predict suspended sediment load (SSL). As part of the Rhône Sediment Observatory (OSR), we propose a novel approach that utilizes streamflow and rainfall data collected from different stations for predicting SSL on each Rhône tributaries. This approach is based on a DL model that combines a Convolutional Layer (CNN) with a Long Short-Term Memory (LSTM) layer, enabling it to capture both spatial and temporal features of input variables. To highlight the advantages of using a complex model, we evaluate its performance against the random forest (RF) model and an empirical approach (Simplified Rating Curve Approach, SiRCA). The CNN-LSTM model shows the best results, outperforming other models in most tributaries, with R² values close to 1. The RF model also outperforms SiRCA, with R² values ranging from 0.8 to 0.94, compared to a lower range of R² from 0.2 to 0.7 for SiRCA. The robustness of our methodology allows us to use it to assess the impact of climate change on sediment dynamics by applying future streamflow and rainfall projections based on various climate scenarios from the Explore2 project.
Session A2 : Hydrological change and climate change
Method for evaluating the flows of a hydro system in low-water conditions: GraviSec scales. First application to the Gardon and Cèze rivers (Gard, France)
MARTIN Philippe
France
Short abstract : Drought situations have a hydrological component, affecting both rivers and springs. In order to adjust restrictive instructions (prefectoral, etc.) as effectively as possible, and for them to be understood and accepted by the population and local stakeholders, it is necessary to have objective assessments that complement aesthetic approaches. With this in mind, we proposed the GraviSec scales as part of the HydroPop program (Ag RMC; ZABR). Two approaches were then developed. One is based on statistics of recorded flows (Banque Hydro). This involves identifying all low-flow events on the same day, regardless of their date. An assessment is then made of the most frequent daily flow (“master” hydrograph: MH) to be expected in the hundred or so days of low water preceding the low-water period ss, and integrated into a table of correspondence (TC) between different variables, including flow, water level, frequency, rating and colour code. This solution is retroactive (cycle whose low-water level is known). It allows the drought level to be contextualised. The second solution is to establish a probabilistic evaluation of the flows for each day of the low-water period, but for which the low-water level is not known, although a start date is chosen (possibly mid-April in a French Mediterranean climate). We assume that each daily series is Gaussian. This is often only true if we perform an anamorphosis on the variable. This solution requires a larger number of cycles but the probabilized hydrograph (PH) can be built as the recession progresses.
CACTUS: an interactive hydrological modelling tool for simulating climate change and land management scenarios in a customisable catchment area
MIMEAU Louise, HÉRAUT Louis, VIDAL Jean-Philippe, BRANGER Flora
France
Short abstract : Spatial hydrological modelling is a useful tool for water management in local areas. In particular, it can be used to test solutions for adapting to climate change by simulating the hydrological response of the catchment to different scenarios. However, setting up a spatialised hydrological model on a real catchment and using it to simulate climate change scenarios or adaptation scenarios can take several months of work.
CACTUS is a hydrological modelling tool developed for water management stakeholders. Its aim is to facilitate consultation and the co-construction of collective solutions. CACTUS contains an interactive interface that makes it very easy to carry out hydrological simulations for a simplified, customisable virtual catchment, and then to quickly test climate change and development scenarios on a simplified case study. It can therefore be used at the start of a project to explore various local strategies for adapting to climate change, or it can be used as a teaching aid for scientific mediation during discussions with local elected representatives or water users to answer their questions about hydrological processes and hydrological modelling. Primary analyses of the hydrological simulations obtained with CACTUS show that the general hydrological behaviour of the catchment simulated by CACTUS is comparable to that obtained with other, more realistic catchment hydrological models.
Multi-scenario method in forecast mode for the operational implementation of efficient flood management
CHAPON benoit, PORTIER Lucie, PICOUET Cécile
France
Short abstract : Among all the questions posed to the scientific world about the consequences of climate change, flood forecasting plays an important role.
However, flood risk analysis too often comes up against the correct definition of the flood scenario. What is the right rain scenario? what is the right hydrological scenario? What is the correct initial state? After work to define a multi-scenario method for forecasting the “Good” rain scenario (Is Rivers 2022), using a distributed event model, an evolution of the method is taking shape to integrate the definition good hydrological scenarios through a good understanding of the initial state. This is then organized around a continuous hydrological model.
The evolution of the method then involves the migration of working tools from event mode to continuous mode and this is what the work presented here proposes to begin.
Glacier melt contributions to future streamflow in the Rhône bassin
CHAMPAGNE Olivier, LEMOINE Anthony, GOUTTEVIN Isabelle, CONDOM Thomas, DELAYGUE Gilles, CAUVY-FRAUNIÉ Sophie, BRANGER Flora
France
Short abstract : The alps are impacted by dramatic changes in the context of global warming with large implications for hydrology. The Rhône bassin, draining a large part of the french and Swiss Alps, has already been the subject of hydrological modelling using J2000-Rhone. In this study, we present the integration of a glacier algorithm in the hydrological model J2000-Rhône, the validation of snowmelt, icemelt and streamflow, and the future projections of these processes. The results show that snowmelt, icemelt and streamflow are satisfactorly simulated by J2000-glaciers in the Rhone basin. By the end of the 21st century, the major changes will be a large increase of streamflow in winter but a decrease in summer associated to earlier snowmelt, a decrease of precipitation and glacier shrinkage. On the Arve and upper Rhône catchments, the remaining glaciers will still be crucial to sustain the streamflow in dry summers.
Construction of a hydrological model construction coupling farm dams and irrigation to estimate the cumulative impact on a regional scale.
PELLERIN Nathan, BRANGER Flora, VIDAL Jean Philippe, MIMEAU Louise, BROWN Ninon
France
Short abstract : Farm dams are a controversial solution to ensure water supply for summer agricultural needs, relying on withdrawals made during the winter period. Current knowledge about their cumulative impact at the watershed scale, their influence on the ecological quality of rivers, and their sustainability as a long-term solution remains insufficient. The aim of this study is to investigate the sustainability of farm dams from an eco-hydrological perspective, under global change and at the regional scale, in order to respond to national regulatory issues. The distributed hydrological model JAMS-J2000 is implemented at the regional scale for the Rhône and Loire watersheds, simulating past, present, and future climates. A substitution reservoir component, designed to represent reservoirs used for irrigation purposes, is included in our modeling. The physical characteristics (volume, depth, porosity) as well as the management modes for supply (surface runoff, groundwater or river abstraction) and restitution (seasonality, water restriction rules) in the reservoir can be modulated, enabling exploration of all the reservoir typologies targeted by our study. The farm dams component is coupled with the irrigation component to enable direct withdrawal of irrigation water from the reservoir. Validation of the coupling is based on water balances, improvements to the J2000 model through the coupling, and comparisons with withdrawal data.