Climate change and anthropogenic pressures are altering the balance of riverine systems, increasing the risk of degradation and compromising their functionality. In this context, integrated and sustainable water resources management represents a key element to ensure efficiency and resilience at the basin scale. However, ongoing transformations profoundly affect hydrological dynamics, making the planning, coordination, and implementation of management strategies increasingly complex. Therefore, this study aims to strengthen water resources management in the Mediterranean region, with particular reference to the Imera Meridionale River Basin (IMRB) in Sicily, Italy, through the integration of seasonal flow forecasts into decision-making processes in contexts characterized by competing water uses. The first part of the study focuses on the development of seasonal streamflow forecasting models using statistical and dynamic climate models. The former group was developed based on local and global predictors selected according to statistically significant correlations with seasonal streamflow, and subsequently used to calibrate statistical models such as Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR). The dynamic models, in contrast, rely on seasonal climate forecast products generated by dedicated climate models that provide short- and long-term global climate predictions, coupled with a semi-distributed, physically based hydrological model, namely the Soil and Water Assessment Tool (SWAT). Despite not explicitly representing the underlying physics, statistical models outperformed the dynamic models, particularly in capturing summer flow peaks, whereas dynamic models, although physically-based, struggled to reproduce peak flows, especially during the winter season. The final part of the thesis focuses on the development of a forecast-informed reservoir operation (FIRO) system for the upstream sub-basin of the Olivo Dam, located within the IMRB, using ensemble forecasts. A risk-based stochastic model predictive control (SMPC) framework was developed for multisectoral basin management under uncertainty. The SMPC framework is implemented on a receding time horizon and updated whenever a new forecast becomes available. The SMPC system was compared with three reference operational scenarios: deterministic model predictive control using climatological forecasts (DMPC-CLIM), deterministic model predictive control using perfect forecasts derived from pseudo-observations of inflows as future predictions (DMPC-PERF), and the standard operating policy (SOP). Results indicate that SMPC overall ensures a more balanced system performance under uncertainty compared to the other approaches. Overall, the proposed framework demonstrates the potential of integrating seasonal forecasting with stochastic predictive control to enhance the robustness and operational resilience of water management systems in Mediterranean environments.
I cambiamenti climatici e le pressioni antropiche stanno alterando gli equilibri dei sistemi fluviali, aumentando il rischio di degrado e compromettendone la funzionalità. In questo contesto, una gestione integrata e sostenibile delle risorse idriche rappresenta un elemento chiave per garantire l’efficienza e la resilienza a scala di bacino. Tuttavia, le trasformazioni in atto modificano profondamente le dinamiche idrologiche, rendendo più complessi i processi di pianificazione, coordinamento e attuazione delle strategie gestionali. Pertanto, il presente studio mira a rafforzare la gestione delle risorse idriche nella regione mediterranea, con particolare riferimento al bacino del fiume Imera Meridionale (IMRB), in Sicilia (Italia), attraverso l’integrazione di previsioni stagionali di deflusso nei processi decisionali in contesti caratterizzati da usi idrici concorrenti. La prima parte è incentrata sullo sviluppo di modelli di previsione del deflusso stagionale utilizzando modelli climatici statistici e modelli dinamici. Il primo gruppo è stato sviluppato sulla base di predittori locali e globali, selezionati in base a valori di correlazione con il deflusso stagionale statisticamente significativi , poi adottati per calibrare modelli statistici del tipo Principal Component Regression (PCR) e Partial Least Squares Regression (PLSR). I modelli dinamici, invece, si basano su prodotti di previsione climatica stagionale ricavati da appositi modelli climatici che forniscono previsioni climatiche a breve e lungo termine a livello globale, accoppiati a un modello idrologico semi-distribuito e fisicamente basato, e cioè il modello Soil and Water Assessment Tool (SWAT). Pur ignorando la fisica sottostante, i modelli statistici hanno mostrato superiorità rispetto ai modelli dinamici, in particolare nel catturare i picchi di portata estivi, mentre i modelli dinamici, pur basandosi sulla fisica del sistema, faticano a catturare i picchi, soprattutto nella stagione invernale. La parte finale della tesi si concentra sullo sviluppo di un sistema di gestione del bacino basato sulle previsioni (FIRO) per il sottobacino a monte della diga Olivo, situato all'interno dell'IMRB, utilizzando previsioni di ensemble. È stato sviluppato un framework di controllo predittivo basato su modelli stocastici (SMPC) basato sul rischio nella gestione multisettoriale del bacino in condizioni di incertezza. Il framework SMPC viene implementato in un orizzonte temporale regressivo e aggiornato ogni volta che viene emessa una nuova previsione. Il sistema SMPC è stato confrontato con tre scenari operativi di riferimento, quali il controllo predittivo tramite modello deterministico con previsione climatologica (DMPC-CLIM), il controllo predittivo tramite modello deterministico con previsione perfetta utilizzando le pseudo-osservazioni dei volumi in ingresso al serbatoio come previsione futura (DMPC-PERF) e la politica operativa standard (SOP). Il risultato indica che SMPC garantisce nel complesso un funzionamento più bilanciato del sistema in condizioni di incertezza rispetto agli altri modelli. Nel complesso, il framework proposto dimostra il potenziale dell’integrazione tra previsione stagionale e controllo predittivo stocastico nel migliorare la robustezza e la resilienza operativa dei sistemi di gestione idrica in ambiente mediterraneo.
Sviluppo di quadri modellistici per la gestione sostenibile e resiliente ai cambiamenti climatici dei bacini idrografici / Tekle, Shewandagn Lemma. - (2026 May 12).
Sviluppo di quadri modellistici per la gestione sostenibile e resiliente ai cambiamenti climatici dei bacini idrografici
TEKLE, SHEWANDAGN LEMMA
2026-05-12
Abstract
Climate change and anthropogenic pressures are altering the balance of riverine systems, increasing the risk of degradation and compromising their functionality. In this context, integrated and sustainable water resources management represents a key element to ensure efficiency and resilience at the basin scale. However, ongoing transformations profoundly affect hydrological dynamics, making the planning, coordination, and implementation of management strategies increasingly complex. Therefore, this study aims to strengthen water resources management in the Mediterranean region, with particular reference to the Imera Meridionale River Basin (IMRB) in Sicily, Italy, through the integration of seasonal flow forecasts into decision-making processes in contexts characterized by competing water uses. The first part of the study focuses on the development of seasonal streamflow forecasting models using statistical and dynamic climate models. The former group was developed based on local and global predictors selected according to statistically significant correlations with seasonal streamflow, and subsequently used to calibrate statistical models such as Principal Component Regression (PCR) and Partial Least Squares Regression (PLSR). The dynamic models, in contrast, rely on seasonal climate forecast products generated by dedicated climate models that provide short- and long-term global climate predictions, coupled with a semi-distributed, physically based hydrological model, namely the Soil and Water Assessment Tool (SWAT). Despite not explicitly representing the underlying physics, statistical models outperformed the dynamic models, particularly in capturing summer flow peaks, whereas dynamic models, although physically-based, struggled to reproduce peak flows, especially during the winter season. The final part of the thesis focuses on the development of a forecast-informed reservoir operation (FIRO) system for the upstream sub-basin of the Olivo Dam, located within the IMRB, using ensemble forecasts. A risk-based stochastic model predictive control (SMPC) framework was developed for multisectoral basin management under uncertainty. The SMPC framework is implemented on a receding time horizon and updated whenever a new forecast becomes available. The SMPC system was compared with three reference operational scenarios: deterministic model predictive control using climatological forecasts (DMPC-CLIM), deterministic model predictive control using perfect forecasts derived from pseudo-observations of inflows as future predictions (DMPC-PERF), and the standard operating policy (SOP). Results indicate that SMPC overall ensures a more balanced system performance under uncertainty compared to the other approaches. Overall, the proposed framework demonstrates the potential of integrating seasonal forecasting with stochastic predictive control to enhance the robustness and operational resilience of water management systems in Mediterranean environments.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tekle_DIsseration_Final_April2026.pdf
embargo fino al 12/05/2027
Descrizione: Tesi def
Tipologia:
Tesi di dottorato
Dimensione
55.17 MB
Formato
Adobe PDF
|
55.17 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.


