The sampling of met-ocean variables is crucial for a plethora of applications. In coastal areas, the management of coastal activities and shipping lanes have to account for variations on mean sea level, wave parameters and current velocities, and coastal defences need to be designed according to the severe sea states they will most likely have to face. Similarly, off-shore engineering projects are expected to stand against forces driven by waves that might occur, e.g., once in ten thousand years. The assessment of design waves relies on statistical extrapolations that need to be fed with reliable and continuous wave data. Therefore, it would be appropriate to extend as much as possible the possible ways of sampling waves. In this regard, this thesis first addresses the reliability of HF-radar wave measures, through a practical case study in the Gulf of Naples. Radar data are compared to the outcomes of two numerical models: one providing the wave parameters on a regional scale, and the other specifically developed for the area of investigation over finer resolutions. Both the models are previously validated against a buoy installed offshore the gulf (taken as reference), which is placed outside the radar domain and therefore cannot be employed for a direct comparison with the latter. The agreement between the models and the HF-radars is evaluated through error indexes computed on the significant wave heights, mean period and mean incoming directions. Results show a reasonable consistency between HF-radar and models measures, leaving room for further investigations on the use of such devices. The aforementioned study refers to hindcast data provided by the Department of Civil, Chemical and Environmental Engineering of the University of Genoa (Italy). The hindcast was developed through a third generation wave model defined over the whole Mediterranean Sea, outputting the most significant wave parameters on a hourly base in the 1979-2018 period. Such data, being continuously defined over a long period, allow also to perform reliable analysis of the extreme waves for given locations. In particular, beyond the analysis of HF-radar wave measurements, this thesis proposes two insights in the framework of the so-called extreme value analysis (EVA). First, a ``bottom-up'' approach for the identification and classification of the atmospheric processes producing extreme wave conditions is revisited, and applied to several locations selected among the Italian buoy network. A methodology is given for classifying samples of significant wave height peaks in homogeneous subsets, related to the climatic forcing driving the most severe wave states. Subsequently, the study shows how to compute the overall extreme values distribution of significant wave height starting from the distributions fitted to each single subset previously detected. From the obtained results, it is concluded that the proposed methodology is capable of identifying clearly differentiated subsets, driven by homogeneous atmospheric processes: two well-known cyclonic systems are identified as most likely responsible of the extreme conditions detected in the investigated locations. These systems are analyzed in the context of the Mediterranean Sea atmospheric climatology, and compared with those figured out by previous researches developed in similar frameworks. Then, it is proved that the high return period quantiles for the significant wave height are consistent with those resulting from the usual computational scheme of the EVA. Finally, a simple model for evaluating non-stationarity in extreme waves is discussed, and possible implications are analyzed through practical examples. This model takes advantage of a linear slope estimate that allows to quantify the rate of change of a given time series of data, lowering the weight of possible outliers. The reliability of this slope is proved against two other methods that are not bounded by the linear trend hypothesis, which in fact could represent a too limiting assumption. This study is applied to series of significant wave height annual statistics over the whole Mediterranean Sea. Trend tests are applied on the series carried out from the hindcast locations, and show that the modified linear slope is sound and reliable. Hence, it is shown how such index can be employed to evaluate for the need of non-stationary EVA rather than the common stationary ones, i.e. when significant divergences between the two models may arise. Finally, the linear slope estimates are used to assess the spatial distribution of historical long-term trend in the Mediterranean Sea, showing interesting analogies with previous works defined over similar locations.
|Titolo della tesi:||New methodologies for the characterization of extreme sea states: applications in the Mediterranean Sea|
|Data di discussione:||6-apr-2020|
|Appare nelle tipologie:||Tesi di dottorato|
File in questo prodotto:
|phdunige_4314598_1.pdf||This file is contains the first two Chapters of the thesis (Introduction and State of the Art)||Tesi di dottorato||Open Access Visualizza/Apri|
|phdunige_4314598_2.pdf||This file is contains Chapters 3 and 4 of the thesis (analysis of wave radar measures and Extreme Value Analysis based on Atmospheric Patterns Classification)||Tesi di dottorato||Open Access Visualizza/Apri|
|phdunige_4314598_3.pdf||This file is contains Chapter 5 (on trends detection in time series of wave data) and the general conclusions of the thesis||Tesi di dottorato||Open Access Visualizza/Apri|