Global and regional hydrological systems have been significantly impacted by climate change, which is also increasing the frequency of extreme rainfall and droughts, changing precipitation patterns, and raising the risk of flooding. In regions like northern Italy, where densely populated urban areas rely on precise climate data for long-term risk management, these changes have been especially obvious. Therefore, accurate precipitation projections are essential for protecting infrastructure, water resources, and urban resilience. However, traditional ensemble methods, which are commonly used in climate studies, frequently are unable to faithfully represent current precipitation conditions, resulting in being inconvenient to be used for future projections. Because of this limitation, there is a major methodological gap that calls for an improvement of better ensemble and calibration techniques. This study evaluates different approaches to improve the performance of Euro-CORDEX Regional Climate Models (RCMs) at a monthly spatial resolution of 0.11° (~12.5 km). In particular, two representative sites in the Lombardy Region: Milan (with two stations: Palazzo di Brera and Palazzo di Marino) and Pavia (with one precipitation station located at Ponte della Libertà), are used as case studies for testing the proposed approaches, using observational precipitation datasets from ARPA-Lombardia, ISPRA, Metropolitana milanese and five RCMs. Five time windows are defined for the future projection analysis: historical time window (1985–2005), Current time window (2006-2024) and future time window divided in 3 approximately equally time periods which are near future window (2026-2050), mid future window (2051-2075) and far future window (2076-2100). Furthermore, two Representative Concentration Pathways (RCPs) for future greenhouse gases concentration scenarios, namely the RCP4.5 (intermediate) and RCP8.5 (worse scenario) were considered in the analysis. Moreover, two new techniques were accessed, in addition to the traditional ensemble mean and ensemble median, for combining the predictions of RCMs, namely: (i) Weighted Normalized Mean Approach, and (ii) Calibrated Approach. Bias correction was applied to improve the accuracy and spatial consistency of future projections. In addition, the study introduces a new bias-correction procedure that uses current observed precipitation rather than historical climate observations, addressing mismatch resulted on by swiftly changing climate conditions and improves the physical accuracy of corrected projections. The results proved that both the weighted and mean ensemble approaches substantially improve the representation of current precipitation conditions compared with the other ensemble approached, and that the new bias correction method increase confidence in future precipitation. These refined approaches strengthen the methodological basis for hydrological modelling, enhance the reliability of climate scenario analysis, and contribute to more informed climate-resilient planning for urban areas such as Milan, Pavia, and other regions facing similar climatic pressures.
Il cambiamento climatico ha influenzato in modo significativo i sistemi idrologici a scala globale e regionale, aumentando la frequenza di eventi estremi come precipitazioni intense e periodi di siccità, modificando i regimi delle precipitazioni e accrescendo il rischio di alluvioni. In regioni come l’Italia settentrionale, dove aree urbane densamente popolate dipendono da dati climatici accurati per la gestione dei rischi a lungo termine, questi cambiamenti risultano particolarmente evidenti. Per questo motivo, proiezioni affidabili delle precipitazioni sono fondamentali per la protezione delle infrastrutture, delle risorse idriche e della resilienza urbana. Tuttavia, i metodi tradizionali di ensemble, comunemente utilizzati negli studi climatici, spesso non riescono a rappresentare in modo accurato le condizioni attuali delle precipitazioni, risultando quindi poco affidabili per le proiezioni future. Questa limitazione evidenzia una significativa lacuna metodologica che richiede il miglioramento delle tecniche di combinazione e calibrazione dei modelli. Il presente studio valuta diversi approcci per migliorare le prestazioni dei Regional Climate Models (RCMs) del progetto Euro-CORDEX con una risoluzione spaziale mensile di 0,11° (circa 12,5 km). In particolare, sono stati selezionati due siti rappresentativi nella regione Lombardia come casi di studio: Milano (con due stazioni: Palazzo di Brera e Palazzo di Marino) e Pavia (con una stazione pluviometrica situata a Ponte della Libertà). L’analisi è stata condotta utilizzando dati osservativi di precipitazione provenienti da ARPA-Lombardia, ISPRA e Metropolitana Milanese, insieme alle simulazioni di cinque RCM. Per l’analisi delle proiezioni future sono state definite cinque finestre temporali: un periodo storico (1985–2005), un periodo attuale (2006–2024) e un periodo futuro suddiviso in tre intervalli di durata approssimativamente simile: futuro prossimo (2026–2050), futuro intermedio (2051–2075) e futuro lontano (2076–2100). Inoltre, nell’analisi sono stati considerati due scenari di concentrazione rappresentativa dei gas serra (Representative Concentration Pathways), ovvero RCP4.5 (scenario intermedio) e RCP8.5 (scenario peggiore). Oltre ai tradizionali metodi di ensemble mean ed ensemble median, sono state analizzate due nuove tecniche per combinare le previsioni dei modelli climatici regionali: (i) l’approccio della media normalizzata pesata (Weighted Normalized Mean) e (ii) l’approccio calibrato (Calibrated Approach). Per migliorare l’accuratezza e la coerenza spaziale delle proiezioni future è stata applicata una procedura di bias correction. Inoltre, lo studio introduce una nuova metodologia di correzione del bias che utilizza le osservazioni di precipitazione del periodo attuale invece dei dati climatici storici, affrontando le discrepanze generate dai rapidi cambiamenti climatici e migliorando l’accuratezza fisica delle proiezioni corrette. I risultati dimostrano che sia l’approccio della media pesata sia l’ensemble mean migliorano significativamente la rappresentazione delle condizioni attuali di precipitazione rispetto agli altri metodi di ensemble, e che la nuova procedura di bias correction aumenta la fiducia nelle proiezioni future delle precipitazioni. Questi approcci raffinati rafforzano la base metodologica per la modellazione idrologica, migliorano l’affidabilità dell’analisi degli scenari climatici e contribuiscono a una pianificazione più consapevole e resiliente al cambiamento climatico per aree urbane come Milano, Pavia e altre regioni che affrontano pressioni climatiche simili.
Accesso e analisi dell'affidabilità dei modelli climatici regionali nel Nord Italia su scala mensile.
IKUZWE, BIENFAIT DARIUS
2024/2025
Abstract
Global and regional hydrological systems have been significantly impacted by climate change, which is also increasing the frequency of extreme rainfall and droughts, changing precipitation patterns, and raising the risk of flooding. In regions like northern Italy, where densely populated urban areas rely on precise climate data for long-term risk management, these changes have been especially obvious. Therefore, accurate precipitation projections are essential for protecting infrastructure, water resources, and urban resilience. However, traditional ensemble methods, which are commonly used in climate studies, frequently are unable to faithfully represent current precipitation conditions, resulting in being inconvenient to be used for future projections. Because of this limitation, there is a major methodological gap that calls for an improvement of better ensemble and calibration techniques. This study evaluates different approaches to improve the performance of Euro-CORDEX Regional Climate Models (RCMs) at a monthly spatial resolution of 0.11° (~12.5 km). In particular, two representative sites in the Lombardy Region: Milan (with two stations: Palazzo di Brera and Palazzo di Marino) and Pavia (with one precipitation station located at Ponte della Libertà), are used as case studies for testing the proposed approaches, using observational precipitation datasets from ARPA-Lombardia, ISPRA, Metropolitana milanese and five RCMs. Five time windows are defined for the future projection analysis: historical time window (1985–2005), Current time window (2006-2024) and future time window divided in 3 approximately equally time periods which are near future window (2026-2050), mid future window (2051-2075) and far future window (2076-2100). Furthermore, two Representative Concentration Pathways (RCPs) for future greenhouse gases concentration scenarios, namely the RCP4.5 (intermediate) and RCP8.5 (worse scenario) were considered in the analysis. Moreover, two new techniques were accessed, in addition to the traditional ensemble mean and ensemble median, for combining the predictions of RCMs, namely: (i) Weighted Normalized Mean Approach, and (ii) Calibrated Approach. Bias correction was applied to improve the accuracy and spatial consistency of future projections. In addition, the study introduces a new bias-correction procedure that uses current observed precipitation rather than historical climate observations, addressing mismatch resulted on by swiftly changing climate conditions and improves the physical accuracy of corrected projections. The results proved that both the weighted and mean ensemble approaches substantially improve the representation of current precipitation conditions compared with the other ensemble approached, and that the new bias correction method increase confidence in future precipitation. These refined approaches strengthen the methodological basis for hydrological modelling, enhance the reliability of climate scenario analysis, and contribute to more informed climate-resilient planning for urban areas such as Milan, Pavia, and other regions facing similar climatic pressures.| File | Dimensione | Formato | |
|---|---|---|---|
|
Bienfait Darius IKUZWE-Thesis..pdf
accesso aperto
Descrizione: This study was done under Supervision of Professor Enrico Creaco and Professor Carlo Giudiccianni. Below is the Pdf file of the thesis.
Dimensione
4.58 MB
Formato
Adobe PDF
|
4.58 MB | Adobe PDF | Visualizza/Apri |
È consentito all'utente scaricare e condividere i documenti disponibili a testo pieno in UNITESI UNIPV nel rispetto della licenza Creative Commons del tipo CC BY NC ND.
Per maggiori informazioni e per verifiche sull'eventuale disponibilità del file scrivere a: unitesi@unipv.it.
https://hdl.handle.net/20.500.14239/33921