Flood events are recognized as the third most damaging natural hazard globally, affecting a significant amount of people annually. Furthermore, urban areas involve highest concentration of valuable items. Thus, improved modelling of urban flooding has been identified as a research priority. In Europe, a key step of implementation of the EU Flood Directive (2007/60/EC) is the preparation of flood hazard maps. A choice is needed in relation to diverse available methods for map preparation, any method having advantages and disadvantages. Moreover, mapping long rivers poses a further challenge compared to mapping short reaches of interest. Hence, the main research question of this thesis work is how one may tackle most effectively/conveniently the hazard mapping of long watercourses. This thesis performs a comparative study of two mapping methods in long rivers for three different return periods. The chosen watercourses to work with are the Adda river in Valtellina, for a length of 95 km, and the Lambro river with a length of 140 km. Earlier 1D hydraulic studies are available for both rivers. The first approach employed is based on two 0D methods (TIRANTI and RAPIDE); both methods perform spatial interpolation of water elevations over a known terrain geometry, the first starting from the results of a 1D model and the second starting from the contour of a flooded area. The second approach applied is a 2D hydraulic modelling using an optimized solver that highly reduces the computation time compared to traditional 2D models. Steady flow was simulated in a 70-km reach of the Adda river to obtain water depth and velocity maps. As a result, three water depth maps were produced for each modelling approach. For the 0D approaches, that were applied to both rivers, the TIRANTI method was preferred to the RAPIDE due to its better time performance and exploitation of available information; however, RAPIDE is obviously recommended for reaches where TIRANTI cannot be applied due to lacking information. The Adda river was modelled in both 0D and 2D, therefore the two approaches were compared for this river. Their applications were similar in terms of total time invested. A comparison of water elevations showed that the approaches globally returned similar values, even though the hydraulic method predicted, generally, higher water depth in the areas near the river and much wider flood extents than the 0D model. This could be due to the roughness values employed in the 2D model, or to area clipping performed by the Autorità di Distretto del fiume Po providing expected contours of the flooded areas. Hydraulic model calibration and unsteady modelling are proposed as future work, after detailed hydrologic data are available. The method recommended in this thesis is the enhanced 2D coding. However, the 0D methods are also a viable option when data is not available or reliable.
Le alluvioni sono globalmente la terza calamità naturale in termini di entità della popolazione annualmente colpita. Le aree urbane includono alte concentrazioni di elementi esposti; quindi, la ricerca di sempre migliori capacità modellistiche è una priorità. In Europa, un punto chiave dell’implementazione della Direttiva UE sulle inondazioni (2007/60/CE) è la preparazione delle mappe di pericolosità alluvionale. Si pone la necessità di scegliere tra diversi metodi applicabili per la preparazione delle mappe, ciascuno con propri vantaggi e limitazioni. Inoltre, la mappatura di fiumi estersi pone difficoltà aggiuntive rispetto a quella di brevi tratti di interesse. Perciò, la principale domanda affrontata da questo lavoro di tesi è come si possa affrontare efficacemente/convenientemente la mappatura della pericolosità per lunghi corsi d’acqua. Questa tesi propone uno studio comparativo di due approcci alla mappatura per tre diversi periodi di ritorno. I corsi d’acqua scelti sono il fiume Adda in Valtellina, lungo 95 km, e il fiume Lambro di 140 km. Per entrambi si hanno precedenti studi idraulici monodimensionali. Il primo approccio si basa su due metodi 0D (TIRANTI e RAPIDE); entrambi conducono interpolazioni spaziali di quote idriche su un terreno di geometria nota, il primo basandosi sui risultati di un modello 1D e il secondo basandosi sul solo perimetro dell’area allagata. Il secondo approccio applicato è un modello idrodinamico 2D ad alta efficienza, che riduce notevolmente il tempo di calcolo rispetto ai modelli 2D tradizionali. Il modello è stato usato in moto permanente per 70 km del fiume Adda, ottenendo le mappe di tirante idrico e velocità. Quindi, sono state prodotte tre mappe di tirante per ciascun approccio. Riguardo ai modelli 0D, che sono stati applicati a entrambi i fiumi, TIRANTI è preferibile a RAPIDE date le sue migliori prestazioni in termini di tempo e la possibilità di sfruttare le informazioni disponibili; RAPIDE è comunque raccomandato nelle situazioni con carenza di informazioni. Il fiume Adda è stato modellato sia in 0D che in 2D, consentendo un confronto tra i due approcci. Le loro applicazioni sono simili in termini di tempo totale investito. Le quote idriche ottenute sono generalmente simili, anche se il metodo idraulico ha stimato tiranti idrici maggiori nelle aree vicino al fiume e aree allagate molto più ampie. Questo potrebbe essere dovuto alla scabrezza impiegata nel modello 2D, o alla perimetrazione preliminare delle aree inondabili condotta dall’Autorità di Distretto del fiume Po. Sviluppi futuri comprenderanno la calibrazione del modello idraulico e una modellazione in moto vario, necessitando però informazioni idrologiche di dettaglio. In sintesi, in questa tesi si raccomanda la simulazione 2D ottimizzata, ma i metodi 0D sono comunque un’opzione vantaggiosa quando i dati non sono disponibili o affidabili.
Mapping urban risk : flood hazard assessment methods for long watercourses and their application in rivers of northern Italy
SILVA MUÑOZ, NATALIA FERNANDA
2018/2019
Abstract
Flood events are recognized as the third most damaging natural hazard globally, affecting a significant amount of people annually. Furthermore, urban areas involve highest concentration of valuable items. Thus, improved modelling of urban flooding has been identified as a research priority. In Europe, a key step of implementation of the EU Flood Directive (2007/60/EC) is the preparation of flood hazard maps. A choice is needed in relation to diverse available methods for map preparation, any method having advantages and disadvantages. Moreover, mapping long rivers poses a further challenge compared to mapping short reaches of interest. Hence, the main research question of this thesis work is how one may tackle most effectively/conveniently the hazard mapping of long watercourses. This thesis performs a comparative study of two mapping methods in long rivers for three different return periods. The chosen watercourses to work with are the Adda river in Valtellina, for a length of 95 km, and the Lambro river with a length of 140 km. Earlier 1D hydraulic studies are available for both rivers. The first approach employed is based on two 0D methods (TIRANTI and RAPIDE); both methods perform spatial interpolation of water elevations over a known terrain geometry, the first starting from the results of a 1D model and the second starting from the contour of a flooded area. The second approach applied is a 2D hydraulic modelling using an optimized solver that highly reduces the computation time compared to traditional 2D models. Steady flow was simulated in a 70-km reach of the Adda river to obtain water depth and velocity maps. As a result, three water depth maps were produced for each modelling approach. For the 0D approaches, that were applied to both rivers, the TIRANTI method was preferred to the RAPIDE due to its better time performance and exploitation of available information; however, RAPIDE is obviously recommended for reaches where TIRANTI cannot be applied due to lacking information. The Adda river was modelled in both 0D and 2D, therefore the two approaches were compared for this river. Their applications were similar in terms of total time invested. A comparison of water elevations showed that the approaches globally returned similar values, even though the hydraulic method predicted, generally, higher water depth in the areas near the river and much wider flood extents than the 0D model. This could be due to the roughness values employed in the 2D model, or to area clipping performed by the Autorità di Distretto del fiume Po providing expected contours of the flooded areas. Hydraulic model calibration and unsteady modelling are proposed as future work, after detailed hydrologic data are available. The method recommended in this thesis is the enhanced 2D coding. However, the 0D methods are also a viable option when data is not available or reliable.File | Dimensione | Formato | |
---|---|---|---|
Thesis_NataliaSilva_nm894360_final.pdf
accessibile in internet per tutti
Descrizione: Thesis_NataliaSilva_nm894360_final
Dimensione
170 MB
Formato
Adobe PDF
|
170 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/154138