Master Eau et Littoral

20/04/2026

Séminaire Scientifique-Equipe ECO (Génie de l'Eau, Climat et Océanographie) du LaGEA/ENSTP/UNSTIM

Email : [email protected]

Jeudi 23 Avril 2026 à partir de 19h


Titre : Optimisation multi-modèles pour la prévision.

Par : HOUNNONDAHO Z. Freddy (LaGEA/ENSTP & ED-STM/UNSTIM)

Pour participer au séminaire sur Google Meet, cliquez sur ce lien :
https://meet.google.com/swz-qdpf-xeh


Abstract:
The Ouémé basin, in Benin, is exposed to increasing hydroclimatic variability that weakens hydrological systems and amplifies flood risks, particularly at the Bonou outlet. In this context, improving the reliability of streamflow forecasts is a major challenge. This study proposes a multi-model approach aimed at reducing the uncertainties associated with the use of a single hydrological model. Three conceptual models (HBV, GR4J and SAC-SMA) were thus calibrated over seven sub-basins of the Ouémé using the Differential Evolution and L-BFGS-B algorithms, testing three objective functions (NSE, logNSE, KGE). Several combination strategies were then compared: the corrected GrangerRamanathan method (GRC), a Random Forest Regressor, and a hierarchical semi-distributed approach implemented through the airGRiwrm package. Several relevant criteria (NSE, KGE and CRPSS for the probabilistic dimension) were used to assess the robustness of our approach. The three models adequately reproduce the seasonal dynamics, with station-dependent performance (KGE around 0.85), and KGE emerges as the best calibration criterion. Multimodel combinations consistently outperform individual models: the Random Forest better captures flood peaks and seasonal transitions, while the GRC performs better in terms of interpretability. At Bonou, the combination markedly improves the representation of extreme events. The multimodel approach therefore constitutes an effective lever for operational hydrological forecasting in tropical basins, reconciling statistical robustness with process representation.

20/04/2026

01/03/2026

01/03/2026

Séminaire Scientifique-Equipe ECO (Génie de l'Eau, Climat et Océanographie) du LaGEA/ENSTP/UNSTIM

Email : [email protected]

Jeudi 5 mars 2026 à partir de 19h


Titre :
Enhancing conceptual rainfall–runoff modeling in data-scarce catchments using machine learning: Kolmogorov–Arnold networks compared to LSTMs.

Par : Ezéckiel Houénafa SIANOU (Université Pan-Africaine, Kenya)

Pour participer au séminaire sur Google Meet, cliquez sur ce lien :
https://meet.google.com/swz-qdpf-xeh


Abstract:
Abstract: Rainfall-runoff modeling in data-scarce catchments—where only precipitation and potential evapotranspiration data are available—remains a major challenge due to the complexity of hydrological processes. Conceptual models such as GR6J are well suited to these contexts because of their parsimonious structure and robustness when dealing with limited data records. This study evaluates the potential of the Kolmogorov–Arnold Network (KAN) to improve rainfall–runoff modeling under data-scarce conditions. A two-stage hybrid framework is proposed in which GR6J residuals are modeled using a KAN architecture, combined with wavelet-based preprocessing to extract multiple frequency components. Results from the Ouémé at Savè basin (Benin) and the Yala basin (Kenya) show that KAN underperforms LSTM in standalone settings but outperforms it in hybrid configurations. The GR6J–Wavelet-KAN model improves NSE from 0.86 to 0.93 in Savè and from 0.68 to 0.76 in Yala, with better performance for high flows. These findings confirm the potential of KAN-based hybrid approaches for rainfall–runoff modeling in data-scarce environments.

14/02/2026

Séminaire Scientifique-Equipe ECO (Génie de l'Eau, Climat et Océanographie) du LaGEA/ENSTP/UNSTIM

Email : [email protected]

Jeudi 19 février 2026 à partir de 19h


Titre :
VARIATION SPATIO-TEMPORELLE DE LA SALINITE DES EAUX DE LA LAGUNE DE PORTO-NOVO (BENIN)

Par : AGBANDEGBA Ulrich (ENSTP/UNSTIM)

Pour participer au séminaire sur Google Meet, cliquez sur ce lien :
https://meet.google.com/swz-qdpf-xeh


Abstract:
This study analyzes the spatio-temporal variability of salinity in the Porto-Novo Lagoon (Benin) during the year 2024. The data are derived from monthly CTD profiles conducted at 36 stations distributed across the whole lake. The results reveal a marked seasonal dynamic, characterized by a salinization phase during the dry season (January–April), with maximum values close to 12 PSU under the influence of marine intrusions. In contrast, the rainy season (May–November) is dominated by near-complete desalination (~0 PSU), related to freshwater inputs from the Ouémé River.
A persistent southwest–northeast spatial gradient is observed, reflecting the combined influence of marine and fluvial forcings. Vertical stratification remains generally weak, with a maximum recorded in March (0.32 PSU).
These findings highlight the strong sensitivity of the lagoon to seasonal hydrological variations, providing essential insights for the sustainable management of this ecosystem facing anthropogenic and climate-related pressures.

08/02/2026

Séminaire Scientifique-Equipe ECO (Génie de l'Eau, Climat et Océanographie) du LaGEA/ENSTP/UNSTIM

Pour nous contacter : [email protected]

Jeudi 12 Février 2026 à partir de 19h


Titre :
MODELISATION DES DEBITS DE CRUE DANS LE BASSIN DU MONO A ATHIEME GRACE A L’APPRENTISSAGE PROFOND

Par : Jude Seruch AGBODOYETIN (ENSGEMM/UNSTIM)

Pour participer au séminaire sur Google Meet, cliquez sur ce lien :
https://meet.google.com/swz-qdpf-xeh


Abstract:
Recurrent floods in the Mono River Basin at Athieme, exacerbated by cli mate variability, call for tools capable of providing reliable multi-day forecasts. This study aims to develop and evaluate deep learning models for multi-horizon (up to 5 days) flood discharge prediction — a key challenge for early warning systems. Based on hydro-climatic data from 1987 to 2023, two models were compared: a simple Long Short-Term Memory (LSTM) network and a Sequence-to-Sequence (Seq2Seq) model with an attention mechanism, specifically designed to generate forecast sequences. In parallel, a frequency analysis of maximum discharges confirmed the relevance of the Generalized Extreme Value (GEV) distribution for modeling such events. The results demonstrate the clear superiority of the Seq2Seq model with attention mechanism across all prediction horizons. For a one-day-ahead forecast (t+1), it achieves remarkable accuracy (R2 = 0.97). Although performance naturally decreases with longer horizons, the model maintains significant robustness up to five days ahead (t+5), consistently outperforming the LSTM model. This superiority also extends to binary flood detection, where the Seq2Seq model more effectively identifies high-risk events over multiple days — a critical feature for operational early warning systems. These findings validate the high potential of Sequence-to-Sequence architectures for multi horizon flood forecasting, providing a promising tool to strengthen flood risk management in the region.

20/01/2026

🎓🌊 Sortie de terrain – Master 2 Eau et Littoral – ENSTP | UNSTIM
📍 Grand-Popo | 📅 17 et 18 décembre 2025

Dans le cadre de la formation du Master Eau & Littoral – ENSTP | UNSTIM de l'Université Nationale des Sciences, Technologies, Ingénierie et Mathématiques (UNSTIM Abomey), les étudiants ont effectué, les 17 et 18 décembre 2025, une sortie de terrain sur la lagune côtière de Grand-Popo, en interaction directe avec le fleuve Mono, principal cours d’eau alimentant ce système lagunaire.

Cette activité pédagogique a permis de consolider les enseignements théoriques à travers la collecte de données in situ relatives à la qualité et à la dynamique des eaux lagunaires. Les mesures ont porté sur plusieurs paramètres physico-chimiques et hydrodynamiques, notamment la salinité, l’oxygène dissous (O₂), la turbidité et le débit.

Les étudiants ont été formés à l’utilisation d’instruments scientifiques spécialisés, parmi lesquels :
• une sonde CTD (Conductivité – Température – Profondeur),
• un appareil multiparamètre pour l’analyse de la qualité de l’eau,
• un ADCP (Acoustic Doppler Current Profiler) pour la mesure des courants et du débit.

La sortie a été coordonnée par Dr Da-Allada Y. Casimir et Dr Ezekiel OBADA, avec des travaux sur l’eau encadrés par Dr Victor Okpeitcha. Le Master Eau et Littoral leur adresse ses sincères remerciements pour la qualité de l’encadrement scientifique et la richesse des enseignements pratiques dispensés.
Cette immersion de terrain illustre l’approche scientifique et professionnalisante du Master Eau et Littoral et met en évidence l’importance des observations in situ pour une gestion durable des zones côtières, en particulier dans le continuum fleuve Mono – lagune – océan au Bénin.