Filtern
Dokumenttyp
Gehört zur Bibliographie
- ja (7) (entfernen)
Schlagworte
- Complex networks (2)
- Event synchronization (2)
- Indian summer monsoon (2)
- Baiu (1)
- Extreme rainfall (1)
- Extremereignisse (1)
- Indischer Sommer-Monsun (1)
- Japan | landslides (1)
- Paleoclimatology (1)
- Paläoklimatologie (1)
- Rainfall patterns (1)
- Tropical storms (1)
- complex networks (1)
- databases (1)
- extreme events (1)
- komplexe Netzwerke (1)
- nichtlineare Datenanalyse (1)
- nonlinear time series analysis (1)
- random forest (1)
Um Extremereignisse in der Dynamik des indischen Sommermonsuns (ISM) in der geologischen Vergangenheit zu identifizieren, schlage ich einen neuartigen Ansatz basierend auf der Quantifikation von Fluktuationen in einem nichtlinearen Ähnlichkeitsmaß vor. Dieser reagiert empfindlich auf Zeitabschnitte mit deutlichen Veränderungen in der dynamischen Komplexität kurzer Zeitreihen. Ein mathematischer Zusammenhang zwischen dem neuen Maß und dynamischen Invarianten des zugrundeliegenden Systems wie fraktalen Dimensionen und Lyapunovexponenten wird analytisch hergeleitet. Weiterhin entwickle ich einen statistischen Test zur Schätzung der Signifikanz der so identifizierten dynamischen Übergänge. Die Stärken der Methode werden durch die Aufdeckung von Bifurkationsstrukturen in paradigmatischen Modellsystemen nachgewiesen, wobei im Vergleich zu den traditionellen Lyapunovexponenten eine Identifikation komplexerer dynamischer Übergänge möglich ist. Wir wenden die neu entwickelte Methode zur Analyse realer Messdaten an, um ausgeprägte dynamische Veränderungen auf Zeitskalen von Jahrtausenden in Klimaproxydaten des südasiatischen Sommermonsunsystems während des Pleistozäns aufzuspüren. Dabei zeigt sich, dass viele dieser Übergänge durch den externen Einfluss der veränderlichen Sonneneinstrahlung, sowie durch dem Klimasystem interne Einflussfaktoren auf das Monsunsystem (Eiszeitzyklen der nördlichen Hemisphäre und Einsatz der tropischenWalkerzirkulation) induziert werden. Trotz seiner Anwendbarkeit auf allgemeine Zeitreihen ist der diskutierte Ansatz besonders zur Untersuchung von kurzen Paläoklimazeitreihen geeignet. Die während des ISM über dem indischen Subkontinent fallenden Niederschläge treten, bedingt durch die zugrundeliegende Dynamik der atmosphärischen Zirkulation und topographische Einflüsse, in äußerst komplexen, raumzeitlichen Mustern auf. Ich stelle eine detaillierte Analyse der Sommermonsunniederschläge über der indischen Halbinsel vor, die auf Ereignissynchronisation (ES) beruht, einem Maß für die nichtlineare Korrelation von Punktprozessen wie Niederschlagsereignissen. Mit hierarchischen Clusteringalgorithmen identifiziere ich zunächst Regionen mit besonders kohärenten oder homogenen Monsunniederschlägen. Dabei können auch die Zeitverzögerungsmuster von Regenereignissen rekonstruiert werden. Darüber hinaus führe ich weitere Analysen auf Basis der Theorie komplexer Netzwerke durch. Diese Studien ermöglichen wertvolle Einsichten in räumliche Organisation, Skalen und Strukturen von starken Niederschlagsereignissen oberhalb der 90% und 94% Perzentilen während des ISM (Juni bis September). Weiterhin untersuche ich den Einfluss von verschiedenen, kritischen synoptischen Systemen der Atmosphäre sowie der steilen Topographie des Himalayas auf diese Niederschlagsmuster. Die vorgestellte Methode ist nicht nur geeignet, die Struktur extremer Niederschlagsereignisse zu visualisieren, sondern kann darüber hinaus über der Region atmosphärische Transportwege von Wasserdampf und Feuchtigkeitssenken auf dekadischen Skalen identifizieren.Weiterhin wird ein einfaches, auf komplexen Netzwerken basierendes Verfahren zur Entschlüsselung der räumlichen Feinstruktur und Zeitentwicklung von Monsunniederschlagsextremen während der vergangenen 60 Jahre vorgestellt.
Analysis of spatial and temporal extreme monsoonal rainfall over South Asia using complex networks
(2012)
We present a detailed analysis of summer monsoon rainfall over the Indian peninsular using nonlinear spatial correlations. This analysis is carried out employing the tools of complex networks and a measure of nonlinear correlation for point processes such as rainfall, called event synchronization. This study provides valuable insights into the spatial organization, scales, and structure of the 90th and 94th percentile rainfall events during the Indian summer monsoon (June-September). We furthermore analyse the influence of different critical synoptic atmospheric systems and the impact of the steep Himalayan topography on rainfall patterns. The presented method not only helps us in visualising the structure of the extreme-event rainfall fields, but also identifies the water vapor pathways and decadal-scale moisture sinks over the region. Furthermore a simple scheme based on complex networks is presented to decipher the spatial intricacies and temporal evolution of monsoonal rainfall patterns over the last 6 decades.
In this study, we provide a comprehensive analysis of trends in the extremes during the Indian summer monsoon (ISM) months (June to September) at different temporal and spatial scales. Our goal is to identify and quantify spatiotemporal patterns and trends that have emerged during the recent decades and may be associated with changing climatic conditions. Our analysis primarily relies on quantile regression that avoids making any subjective choices on spatial, temporal, or intensity pattern of extreme rainfall events. Our analysis divides the Indian monsoon region into climatic compartments that show different and partly opposing trends. These include strong trends toward intensified droughts in Northwest India, parts of Peninsular India, and Myanmar; in contrast, parts of Pakistan, Northwest Himalaya, and Central India show increased extreme daily rain intensity leading to higher flood vulnerability. Our analysis helps explain previously contradicting results of trends in average ISM rainfall.
We propose a novel approach based on the fluctuation of similarity to identify regimes of distinct dynamical complexity in short time series. A statistical test is developed to estimate the significance of the identified transitions. Our method is verified by uncovering bifurcation structures in several paradigmatic models, providing more complex transitions compared with traditional Lyapunov exponents. In a real-world situation, we apply this method to identify millennial-scale dynamical transitions in Plio-Pleistocene proxy records of the South Asian summer monsoon system. We infer that many of these transitions are induced by the external forcing of the solar insolation and are also affected by internal forcing on Monsoonal dynamics, i.e., the glaciation cycles of the Northern Hemisphere and the onset of the Walker circulation.
Frequent and intense rainfall events demand innovative techniques to better predict the extreme rainfall dynamics. This task requires essentially the assessment of the basic types of atmospheric processes that trigger extreme rainfall, and then to examine the differences between those processes, which may help to identify key patterns to improve predictive algorithms. We employ tools from network theory to compare the spatial features of extreme rainfall over the Japanese archipelago and surrounding areas caused by two atmospheric processes: the Baiu front, which occurs mainly in June and July (JJ), and the tropical storms from August to November (ASON). We infer from complex networks of satellite-derived rainfall data, which are based on the nonlinear correlation measure of event synchronization. We compare the spatial scales involved in both systems and identify different regions which receive rainfall due to the large spatial scale of the Baiu and tropical storm systems. We observed that the spatial scales involved in the Baiu driven rainfall extremes, including the synoptic processes behind the frontal development, are larger than tropical storms, which even have long tracks during extratropical transitions. We further delineate regions of coherent rainfall during the two seasons based on network communities, identifying the horizontal (east-west) rainfall bands during JJ over the Japanese archipelago, while during ASON these bands align with the island arc of Japan.
Landslide hazard models aim at mitigating landslide impact by providing probabilistic forecasting, and the accuracy of these models hinges on landslide databases for model training and testing.
Landslide databases at times lack information on the underlying triggering mechanism, making these inventories almost unusable in hazard models.
We developed a Python-based unique library, Landsifier, that contains three different machine-Learning frameworks for assessing the likely triggering mechanisms of individual landslides or entire inventories based on landslide geometry.
Two of these methods only use the 2D landslide planforms, and the third utilizes the 3D shape of landslides relying on an underlying digital elevation model (DEM). The base method extracts geometric properties of landslide polygons as a feature space for the shallow learner - random forest (RF).
An alternative method relies on landslide planform images as an input for the deep learning algorithm - convolutional neural network (CNN).
The last framework extracts topological properties of 3D landslides through topological data analysis (TDA) and then feeds these properties as a feature space to the random forest classifier.
We tested all three interchangeable methods on several inventories with known triggers spread over the Japanese archipelago. To demonstrate the effectiveness of developed methods, we used two testing configurations.
The first configuration merges all the available data for the k-fold cross-validation, whereas the second configuration excludes one inventory during the training phase to use as the sole testing inventory.
Our geometric-feature-based method performs satisfactorily, with classification accuracies varying between 67 % and 92 %. We have introduced a more straightforward but data-intensive CNN alternative, as it inputs only landslide images without manual feature selection.
CNN eases the scripting process without losing classification accuracy. Using topological features from 3D landslides (extracted through TDA) in the RF classifier improves classification accuracy by 12 % on average.
TDA also requires less training data. However, the landscape autocorrelation could easily bias TDA-based classification. Finally, we implemented the three methods on an inventory without any triggering information to showcase a real-world application.
Electronic databases of landslides seldom include the triggering mechanisms, rendering these inventories unusable for landslide hazard modeling. We present a method for classifying the triggering mechanisms of landslides in existing inventories, thus, allowing these inventories to aid in landslide hazard modeling corresponding to the correct event chain. Our method uses various geometric characteristics of landslides as the feature space for the machine-learning classifier random forest, resulting in accurate and robust classifications of landslide triggers. We applied the method to six landslide inventories spread over the Japanese archipelago in several different tests and training configurations to demonstrate the effectiveness of our approach. We achieved mean accuracy ranging from 67% to 92%. We also provide an illustrative example of a real-world usage scenario for our method using an additional inventory with unknown ground truth. Furthermore, our feature importance analysis indicates that landslides having identical trigger mechanisms exhibit similar geometric properties.