EDF (Électricité De France) is the world's largest electricity generator, with an installed capacity of about 130 GW. In order to safely operate the plants, optimize natural resources and fulfill ecological requirement, EDF has installed, since 1946, a sensor network dedicated to the monitoring of hydro-climatologic parameters.

In the context of non-intrusive methods for measuring flood discharge (LSPIV, SVR[1]), understanding the depth-averaged to surface velocity ratio is crucial. The depth-averaged to surface velocity ratio is here called α. This study analyzes a substantial sample of gaugings data (current meters and ADCP methods), totaling around 6,500 observations collected at various EDF sites. For current meters measurements, three methods are employed to compute α : fitting of a log- and a power-law and using the measured surface velocity. For ADCP measurements, three methods are applied to approach α : fitting of power-power, constant-no slip and 3-point-no slip law by using the Qrame[2] application.

This study aims at creating an alpha coefficient database (classified by riverbed, hydraulic radius, etc.) directly usable for non-intrusive streamflow measurements. 

How to cite: Perriaud, T., Morlot, T., and Hauet, A.: Velocity profile and depth-averaged to surface velocity in natural streams: a review over a large sample of rivers using current meters and ADCP measurements., EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-22428, https://doi.org/10.5194/egusphere-egu24-22428, 2024.

In the last decades, Acoustic Doppler Current Profilers (ADCP) has become the most widely used tool for measuring discharge of rivers and canals. Discharge is a key information for many risk studies, structures dimensioning and even impact assessments. These values must therefore be correctly estimated. Quality assurance and quality control (QA/QC) procedures have been established to ensure that hydrological services share the best practices. Also, to ensure that ADCP tools are working properly, services has to regularly check that the equipment is properly calibrated.

The lack of references value most of the time makes the task difficult. To make sure that ADCP is properly calibrated, interlaboratory testing are frequently organized. Large-scale intercomparaisons are particularly interesting because of the diversity of models and practices but it also makes them more complicated to organize. The Sault-Brénaz intercomparaison was definitively a big one with more than 120 European participants with 16 RiverPro, 15 M9, 15 StreamPro and 12 RS5 for a total of 160 measurements with 1870 transects among 4 sessions. Due to hydrological conditions, the protocol had to be adapted. Measurements took place on small straight canal of the Rhone river with a discharge of around 2m³/s.

Following QA/QC procedures, participant had to post-process their data with the QRevInt open-source software. QRevInt provides many quality filters and computes uncertainty following OURSIN method. Then, to compute interlaboratory results, the QRame software has been used. This open-source software has been developed to apply QRevInt with default settings to a set of ADCP discharge measurements and to retrieve post-processed discharge and uncertainty results. When the dataset is actually an ADCP interlaboratory experiment, the empirical discharge uncertainty, for a given number of transects taken in the average, can be computed by application of the standard interlaboratory method.

Results show that discharge varied slightly over time, particularly between sessions. To exploit further all the discharge results, different approaches to homogenizing data were tested. This issue of varying discharge over time is a common issue for interlaboratory experiments. A generalizable solution would enable experiments in extended conditions. Also, interlaboratory experiments permit to validate uncertainty computations. The greater the number of intercomparisons and the wider the measurement conditions, the more robust uncertainty models will be.

How to cite: Despax, A., Calmel, B., Le Coz, J., Hauet, A., and Mueller, D.: An ADCP large-scale international intercomparaison: Sault-Brénaz 2023, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-9754, https://doi.org/10.5194/egusphere-egu24-9754, 2024.

The transboundary Mekong River, spanning approximately 4800 km with numerous tributaries and floodplains, serves as a vital resource for power generation, fisheries, and agriculture. Despite its significance, the river's productivity faces disruption due to inadequate cooperation among riparian countries regarding data sharing, the uneven distribution of gauging stations, and data gaps for many parts of the river length. This disparity poses challenges in accurately modeling the river's natural runoff, flow characteristics, and the flooded area, navigating through mountainous and relatively flat terrains.

To address this, we have developed an integrated modeling framework comprising a physically-based hydrological model and a hydrodynamic model. For 2500 km of the Mekong River’s mainstream, a highly accurate hydrodynamic model was developed. The produced velocity, water level, and discharge data were compared with gauging stations with continuous data records, showing high accuracy with NSE exceeding 0.93. Additionally, a point-by-point comparison of the yielded water level and discharge data by the hydrodynamic model was conducted with the low-resolution recorded data for stations lacking continuous time series data. Results indicated a high accuracy with an average NSE greater than 0.91, demonstrating the model's precision in capturing the dynamic behavior of the Mekong River.

The hydrodynamic model's results were then used to fill data gaps in stations with significant data deficiencies, allowing the production of reliable data and sufficient gauging network distribution for the entire basin. These datasets, combined with recorded gauging data, served as the calibration stations for the developed physically-based hydrological model. This calibration aimed to assess the impacts of climate change on natural runoff, encompassing not only the mainstream but also tributaries and lake floodplains of the Mekong River. Findings revealed a discernible declining trend in natural runoff within the Mekong River over the specified four-decade period.

This enhanced modeling capability is particularly crucial for accurately simulating dynamic river flows with insufficient continuous data. Our comprehensive approach contributes to a more precise understanding of the Mekong River's complex hydrological dynamics, supporting informed decision-making for sustainable resource management.

How to cite: Morovati, K., Zhang, K., and Tian, F.: Integrated Approach to Mekong River Flow Modeling: Data Gaps and Climate Trends, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-1606, https://doi.org/10.5194/egusphere-egu24-1606, 2024.

River flow forecasting has been the focus of many researchers for many years.  The methods evolved from simple statistical methods to highly sophisticated mathematical models.  In recent years, due to the advancement of computers and artificial algorithms, new methods have become increasingly reliable and easier to use.  One of the promising artificial intelligence methods is the Extreme Gradient Boosting (XGBoost) model.  XGBoost is a scalable, distributed gradient-boosting decision tree machine learning library.  It provides parallel tree boosting and is the leading machine learning library for regression, classification, and ranking problems.  Three different algorithms of XGBoost were used in this research and the results were compared.  These algorithms were Random Search, Grid Search, and CatBoost. The proposed models were conducted in a station located Pò River basin which is the longest river in Italy, and it flows from the Cottian Alps and ends at a delta projecting into the Adriatic Sea new Venice.  The data were divided into training and validation sets.  The statistical indicators included mean square error, Nash-Sutcliffe efficiency, and mean absolute error were calculated for each set to compare the efficiency of each algorithm.  These indicators showed that XGBoost using random search algorithm had better performance, although the other algorithms were also acceptable predictions.  In general, the XGBoost model could be used as a reliable tool to forecast the river flow at locations with enough historical data.

How to cite: Golmohammadi, G., Razdar, B., Mohammadi, K., Grossi, G., and Javadi, S.: A comprehensive method based on machine learning schemes in predicting river flow, case study: Po River, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4402, https://doi.org/10.5194/egusphere-egu24-4402, 2024.

We present the latest version of the global reach-scale “ICESat-2 River Surface Slope” (IRIS) dataset, which comprises average and extreme water surface slopes (WSS) derived from observations of the ICESat-2 satellite between October 2018 and August 2023 as a supplement to 130,283 reaches from the “SWOT Mission River Database” (SWORD). To gain full advantage of ICESat-2’s accurate and unique measurement geometry with six parallel lidar beams, the WSS is determined across pairs of beams or along individual beams, depending on the intersection angle of spacecraft orbit and river centerline. Combining both approaches maximizes spatial and temporal coverage. IRIS can be used to research river dynamics, estimate river discharge, and correct water level time series from satellite altimetry for shifting ground tracks. Additionally, we compare IRIS with observations from the recently launched SWOT mission. 

How to cite: Scherer, D., Schwatke, C., Dettmering, D., and Seitz, F.: IRIS: Global Reach-Scale River Surface Slopes from the ICESat-2 Satellite , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-5805, https://doi.org/10.5194/egusphere-egu24-5805, 2024.

Rating curve relationship is vital to hydrological studies, such as flood control and other water-related decision-making processes. Traditionally, rating curve are estimated by using single- or multiple-gauging observations, which is time-consuming, costly, and lacks spatial resolution. Hydraulic models are usually a reliable method to quickly derive the stage-discharge relation for discharge estimation, especially for assessing more reliable high-flow rating relations in extrapolation beyond gauge observation. To establish such models, hydraulic parameters such as water surface elevation, bathymetry, and bed roughness are needed, but they are mostly not available in remote and inaccessible regions. Drone-borne hydrometric monitoring technologies can be deployed to address this problem.

As one of the primary objectives of the Horizon Europe UAWOS project, which is dedicated to developing an Unmanned Airborne Water Observing System for providing key hydrometric variables at high spatial resolution/coverage, and data-based products/services to enhance management and decision-making, this work centers on integrating hydraulic modeling with the unmanned airborne water observing system to establish the rating curve relationship. Water surface elevation data is derived by radar altimetry, bathymetry data by water penetrating radar and sonar, and Doppler radar for surface velocity. By utilizing the surface velocity and water surface elevation data, in conjunction with shallow-water equations, a bathymetry estimation algorithm is used to interpolate the bathymetry from the observed cross-section to the whole simulated river channel. We also come up with a method to directly retrieve the river roughness parameter from the UAV drone observation data.

As a whole, these methods collectively establish a framework that is easily to use to estimate the rating curve in remote regions. The study shows how information from high spatial resolution and coverage hydrometric variables derived by drone-borne hydrometric monitoring technologies can improve rating curve estimates from models.

How to cite: Hu, X., Tuo, Y., Broich, K., Merk, F., and Disse, M.: A Contactless Rapid Rating Curve Assessment Based On Drone-borne Measurement, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-11189, https://doi.org/10.5194/egusphere-egu24-11189, 2024.

Many hydrologists face the situation that they have no, or very limited, information about the uncertainty in the discharge data they are using. Data uncertainty is rarely communicated by monitoring agencies and data providers – and is often not available on request. This means that data users typically treat data as if they are error-free, whereas in reality there can be large uncertainties and errors.

However, the absence of metadata and ‘hard’ information about data uncertainty does not mean that there is no information about the data uncertainty. Instead, we can use other types of ‘soft’ information to understand the likelihood that discharge data in a particular location are uncertain. For example, if high flows are of short duration (i.e., a few hours) and the rainfall-runoff lag time is short, it is practically quite difficult to manage to gauge high flows, leading to likely extrapolation of stage–discharge rating curves and large high flow uncertainty. A second example is if a river is ice-covered during the winter season, then most of the winter water-level time series is subjectively estimated, leading to substantial uncertainty in winter low flows. Such soft information about data uncertainty is well known by field hydrologists and data uncertainty experts but is not as commonly known in the wider hydrological community. In this presentation we focus on uncertainty in discharge data calculated from stage–discharge rating curves and aim to share – and to encourage sharing – of soft information about data uncertainty sources, to promote more informed decisions on data uncertainty in hydrological studies.

We summarize the soft information about discharge data uncertainty as a perceptual model of uncertainty. Our perceptual model divides the soft information into three categories: station characteristics, climate and flow regime, and catchment characteristics. For each category we present and describe different types of soft information, the uncertainty sources and impacts they can inform us about, and sources for each soft information type (e.g., photos, satellite images, land use). We find that soft information can inform us about three main types of uncertainty sources: uncertainty related to the hydraulic control, uncertainty related to incomplete gauging of the full flow range, and uncertainty due to measurement error.

Our generalised perceptual model can be seen as a smorgasbord of information about uncertainty sources, where the soft information can be considered as relevant to a particular dataset and can inform us if high or low data uncertainty is likely. We believe that a key benefit of the type of generalized perceptual model of uncertainty we present is to facilitate dialogue on, and understanding of, possible sources of observational uncertainties and their impacts.  We encourage others to complement our perceptual model of discharge data uncertainty based on experience from different regions and for other discharge monitoring techniques such as index-velocity stations or drone/camera-based methods.

How to cite: Westerberg, I. and Huseby Karlsen, R.: Sharing perceptual models of uncertainty – on the use of soft information about discharge data uncertainty, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16356, https://doi.org/10.5194/egusphere-egu24-16356, 2024.

Evapotranspiration (ET) is one of the main environmental variables for the study of land-atmosphere interactions due to its interconnection with the energy and water balance at the land surface. Despite the dedicated effort of the remote sensing community to estimate the magnitude of ET at global scales, the uncertainties and differences between products are still very large, especially when comparing ET products with different spatial resolutions. Here, we designed a round-robin experiment to determine the product or products most suitable for future integration in hydrological modeling. The evaluation is performed using eddy covariance measurements as reference and point-scale downscaling (PSD) benchmarking criteria to identify the added value of the high-resolution products. The eddy covariance measurements of latent and sensible heat fluxes are known to not close the surface energy budget. Therefore, the use of eddy covariance measurements as a reference could have important consequences for the later use of ET products in assimilation approaches. Therefore, two main strategies to deal with the energy balance closure problem are considered here. Firstly, we considered three energy balance closure scenarios – (i) assigning the energy balance residuum to sensible heat flux; (ii) distributing the residuum to both turbulent energy fluxes by preserving their ratio, i.e. Bowen ratio; (iii) assigning the entire residuum to latent heat flux. While the first case has no impact on ET, the two remaining ones lead to an increase in ET. Secondly, the use of the triple collocation method, which does not require a reference dataset, will be explored to complement these results. Despite these efforts to identify the best ET product for the integration of satellite ET products in hydrological models, we cannot conclude that the products reaching the best metrics in this evaluation will be the products adding more value to the assimilation approach. Therefore, further experiments should be designed to test if the products selected in the round-robin exercise are indeed improving the performance of hydrological models or on the contrary other ET products are more suitable for assimilation approaches. We acknowledge support from AdAgriF - Advanced methods of greenhouse gases emission reduction and sequestration in agriculture and forest landscape for climate change mitigation (CZ.02.01.01/00/22_008/0004635).

How to cite: Orság, M., Fischer, M., García-García, A., Peng, J., Samaniego, L., and Trnka, M.: Evaluation of remote sensing actual evapotranspiration products for hydrological modeling applications, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19308, https://doi.org/10.5194/egusphere-egu24-19308, 2024.

Coping with natural disasters such as floods places special demands on the emergency units. From the point of view of command-and-control operators, observations of watercourses are desirable in the event of flooding in order to obtain an accurate picture of the situation. Optical measurement methods using cameras offer thereby advantages as they do not require water contact and hence can be used safely. Therefore, the project "KIWA: Artificial Intelligence (AI) for Flood Warning" (http://kiwa.hydro.tu-dresden.de/) is developing AI-based tools for the robust quantification of water levels, flow velocities and flow rates from surveillance cameras.

In this article, we present the workflow for an exclusive optical measurement of time series of water level and discharge from single images and short video sequences. The basis is a high-precision (i.e., at centimetre level), georeferenced 3D terrain model of the measurement site including the riverbed. The terrain model is created using the structure-from-motion (SfM) technique and georeferenced via ground control points (GCPs) measured with a multiband GNSS receiver. To determine the water level, the water area in the single images is automatically segmented using AI based on convolutional neural networks (CNNs) and then intersected with the terrain model. Changes of the camera geometry influence the measurement accuracy during long-term observations. Therefore, the GCPs are automatically detected in the individual images with an adapted AI-based keypoint detector to frequently update the estimated camera orientation. To estimate the discharge, the water surface flow velocity is determined using short video sequences and applying the particle tracking (PTV) method, whereby the segmented water area narrows down the search area for the particle detection. Afterwards, the "OptiQ" modelling approach is used to derive the discharge times series based on the PTV measurements. Thereby, data filtering and error correction methods are used to achieve continuous time series. 

The methods were developed at three different measuring gauges, whose cameras record single images and videos every 15 minutes over several months. The accuracy of the water level measurement is in the centimetre range, even at night with the support of infrared emitters. Depending on the water level, there are deviations in the flow rate, which average less than 10%.

How to cite: Grundmann, J., Blanch, X., Kutscher, A., Hedel, R., and Eltner, A.: Towards a comprehensive optical workflow for monitoring and estimation of water levels and discharge in watercourses, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-12507, https://doi.org/10.5194/egusphere-egu24-12507, 2024.

In Europe, the majority of the precipitation during winter falls as snow over 1.000 m altitude, and remains stored in the snowpack until the melting season, when it returns in the hydrological cycle and is partly used for irrigation and power generation. Snow cover estimation is then one of the main indicators necessary to evaluate water budget and plan water management, predict possible drought conditions, and drive operational flood prediction. 

The use of Earth Observation (EO) for Snow Cover Area (SCA) estimation has been improved during the last decade thanks to high resolution satellites such as the ESA Sentinels, having a pixel resolution of 10 m. Furthermore, diverse processing techniques, nowadays mature, are used by the different data providers to retrieve SCA and Fractional Snow Cover (FSC) maps from EO data.

The scope of our work is an intercomparison of different medium- to high-resolution EO-retrieved SCA/FSC maps over Europe; we use as a benchmark a vast dataset of in situ data coming from different sources and harmonized by Matiu et al. (2021). 

Regarding the dataset, SCA or FSC maps retrieved from multispectral Sentinel-2 and Landsat-8 images are considered, together with gap-filled maps evaluated integrating Sentinel-1 (Synthetic Aperture Radar) and/or Sentinel-3 (multispectral). A unique dataset of Sentinel-1-retrieved snow depth maps is also used in the exercise. Finally, for a continuity with a previous project on EO snow products, medium-resolution MODIS-retrieved SCA images have been added to our dataset.

The results can be used to correctly interpret the accuracy of the EO datasets as well as the processing methodologies. From the comparison it can be evaluated the possibility of merging the different dataset in order to enhance the temporal resolution to a sub-weekly effective revisit time.

How to cite: Di Paolo, F., Dall'Amico, M., Stradiotti, P., and Samaniego, L.: A Round Robin Exercise for an intercomparison of Snow Cover Area maps retrieved from Earth Observation, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-17533, https://doi.org/10.5194/egusphere-egu24-17533, 2024.

Soil moisture (SM) is a fundamental hydrological variable for understanding processes in the land-atmosphere, biological, or geophysical domains and an output of many hydrological or land surface models. It is peculiar in that its variability reflects distinct hydrological processes moving from the field scale, where local topography plays a role, to the regional scale, where meteorological forcing is the main control. Correctly representing this variety of processes is a complex modeling task often alleviated by integrating information from well-established Earth Observation (EO) systems, which produce SM data with near global coverage at coarse (10-25 km) resolution. Still, the increasing need for fine scale (1 km, 1 day) simulations of the water cycle is to be met by EO data of similarly high resolution. 

High resolution satellite-based SM data is now available from several sources. 1km datasets are multiplying following simultaneous efforts to retrieve SM from backscatter measurements of the Sentinel-1 mission with various inversion models. At the same time, physical or statistical relationships are leveraged to down-scale coarse resolution products by ingesting data from distinct observational sources, coming from the mentioned Sentinel-1 or the optical domain. However, while products of the first type are confronted with the limited sensitivity of C-band microwave to SM and reduced spatial and temporal availability, down-scaled products might retain much of the original signal and fail the fine-scale process representation. The question of which of these resources can preferably be integrated to reliably improve high-resolution modelling is therefore an open one. 

In this study we perform a round robin (i.e., inter-comparative) assessment of the most prominent high-resolution SM products in the EO landscape. While adapting validation and error characterization techniques and tools (e.g., the Quality Assurance for SM service) that are routinely used at the coarse scale, we address the partial lack of 1km scale reference measurements through the application of an emerging high resolution validation framework. Such a framework demonstrates that metrics for high resolution benchmarking can be reliably retrieved with only sparse, point-scale measurements. The first results suggest that the true spatial SM heterogeneity might explain a minimum noise tradeoff between coarse- and high-resolution EO products. This work is a fundamental step to assess the current state-of-art in EO and its maturity for integration in high-resolution water cycle modelling.

How to cite: Stradiotti, P., Dorigo, W., and Samaniego, L.: Exploring the relative scale of uncertainty in high-resolution soil moisture remote sensing products towards model integration , EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-15519, https://doi.org/10.5194/egusphere-egu24-15519, 2024.

Water and energy balances have been monitored at a scale which is comparable with remote sensing one in three North-West Italy sites. One step has been to evaluate the performance of a land surface model, in this work the Community Land Model. The measurements taken at the horizontal hundreds meters scale are also compared with vertical profiles of local sensors of soil moisture.

At the grassland mountain site (2600 m asl) the eddy covariance data are taken from 6 years, while the 25 m high mast eddy covariance in the forest from 3 years. The scintillometer and cosmic ray in the vineyard have been installed from one year.

The main result is to have different land cover monitored at about 1 km scale, and to see that the uncalibrated simulations with CLM are following quite well the data in most cases. Also the comparison of cosmic ray and point soil moisture time series will be discussed. The future work will be the comparison with satellite data.

This work is a part of the project NODES which has received unding from the MUR-M4C2 1.5 of PNRR grant agreement no. ECS00000036

How to cite: Ferraris, S., Gentile, A., Gisolo, D., Canone, D., Bechis, S., Heery, B., Marcella, B., Capello, G., Maaschwitz, G., Myagkov, A., Gazzola, E., and Stevanato, L.: Eddy covariance, scintillometer, and cosmic ray 1 km scale measurements at three sites (grassland, forest, and vineyard) in North-West Italy compared with CLM simulations, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-19632, https://doi.org/10.5194/egusphere-egu24-19632, 2024.

Non-perennial streams are those streams that periodically cease to flow in at least one point along their network. The research community well recognizes the importance of such watercourses for they have a global prevalence and provide diverse hydrological functions and ecosystem services. The spatiotemporal pattern of the active drainage network is anything but simple, sometimes showing a very complex pattern. Non-perennial streams, in fact, are often located in heterogeneous environments, in which the combination of climate, morphology, land cover, soil, substrate, and anthropic factors could play a role in the observed drying and wetting patterns. This work combined two techniques, with different spatiotemporal resolutions, to characterize the spatiotemporal extent of the stream network in a 3.7 km² Mediterranean catchment of central Italy. The hydrological status of a set of nodes of the network was derived for the period 2020-2022 from sporadic visual surveys and, most importantly, through the analysis of sub-hourly images collected by 21 cameras distributed along a set of strategic nodes of the network. The latter technique is particularly promising to reconstruct the hydrological dynamics taking place in the target cross-section, as the temporal evolution of the underlying hydrological conditions (wet vs. dry), the water stage, and the corresponding discharge can be inferred from the automatic or manual analysis of the acquired images. The available experimental data  was combined exploiting the hierarchical principle, that postulates the existence of a Bayesian chain based on the local persistency of the nodes that dictates their drying/wetting order during stream retraction/contraction cycles. The results highlighted the complexity of the network dynamics in the study area: while the number of wet nodes decreased during the dry season and increased during the wet season, the local persistency of the nodes showed a highly heterogeneous and non-monotonic pattern, resulting in a dynamically disconnected network. The approach allowed the reconstruction of the entire river network and represented a useful tool to estimate the extent of its wet portion, even in case part of the network could not be inspected. This work represents a novel approach to reconstruct the extension of the wet portion of the stream network in difficult-to-access environments, where traditional techniques might be inadequate.

How to cite: Noto, S., Durighetto, N., Tauro, F., Grimaldi, S., and Botter, G.: Characterizing the space-time evolution of wet channels in a non-perennial Mediterranean catchment exploiting a network of camera traps, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-16913, https://doi.org/10.5194/egusphere-egu24-16913, 2024.

Hydrological modeling is critical for effective water resources management, especially in developing countries such as Morocco where data are scarce. This study aims to improve daily river discharge predictions in 26 Moroccan catchments from 1993 to 2019. It evaluates the GR4J and MISDc models, focusing on optimizing their performances using four optimization techniques: Particle Swarm Optimization (PSO), the Nelder-Mead simplex algorithm (FMIN), Simulated Annealing (SA), and the Genetic Algorithm (GA). The two hydrological models are coupled with six calibration methods to provide the different ranges of uncertainties and to assess their consistency across diverse datasets. The methods include the split-sample or half-half method, the odd/even year method, as well as the calibration on a longer period than validation and vice versa. In addition, the Kling-Gupta Efficiency (KGE) and the relative bias were used as performance criterions. Due to the high elevation of some catchments studied and to the important amount of the snowmelt contribution in the river discharge at their outlets, a snow module incorporation was necessary to assess whether snowmelt impacts runoff or not. The outcomes demonstrate that all algorithms were able to successfully calibrate the GR4J and MISDc models (-0.26<median KGE< 0.34). However, FMIN and PSO demonstrated greater consistency in their performance across all calibration methods and proved to be the most computationally efficient algorithms, making them the best choices in situations requiring both time effectiveness and performance. Despite its slower speed, GA's robustness makes it a viable option under less time-sensitive conditions. The relative bias metric indicates that for the GR4J model, the FMIN, PSO, and GA had comparable and balanced performance, while SA showed greater variability. For the MISDc model, FMIN showed a tendency to slightly underestimate the discharge, while GA and PSO showed higher biases in some cases. In addition, MISDc significantly outperformed GR4J in simulating runoff across all catchments, making it a suitable choice for our region. The integration of a snow module in both models enhanced their performance in some larger pluvio-nival catchments, illustrating the complexity of snow dynamics in hydrological modeling and the need for high resolution data as well as ground measurements.

Keywords: River discharge prediction, GR4J, MISDc, Moroccan catchments, Optimization methods, Data scarcity.

How to cite: Ait naceur, K., El khalki, E. M., Hadri, A., Jaffar, O., Brocca, L., Saidi, M. E. M., Tramblay, Y., and Chehbouni, A.:  Daily Streamflow Simulations Improvement in Data Scarce Watersheds using different Optimization Techniques and Calibration Methods, EGU General Assembly 2024, Vienna, Austria, 14–19 Apr 2024, EGU24-4407, https://doi.org/10.5194/egusphere-egu24-4407, 2024.