Identification, tracking and analysis of 3-D meteorological features
Loading...
Date issued
Authors
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Reuse License
Abstract
Meteorological data sets, which detail the state of the Earth's atmosphere, have become increasingly complex and diverse over the past decades. The growing volume and quality of these data have created a demand for advanced tools capable of efficiently handling and analyzing large, multidimensional atmospheric data sets. This dissertation meets this demand by combining computer science techniques with meteorological research to develop practical software tools and novel algorithms. These tools aim to help atmospheric scientists extract valuable insights from vast data sets through efficient feature identification, tracking, and analysis.
The main focus of this work is the development of new algorithms for identifying and tracking meteorological features in three dimensions. By utilizing algorithmic geometry, this dissertation introduces low-dimensional, geometry-based descriptors that simplify the analysis of wave-related features such as tropical waves and anomalies along the dynamical tropopause. These descriptors significantly reduce the complexity of the data while preserving important information, making the analysis more efficient and potentially improving the predictability of weather events linked to these features.
A key aspect of this dissertation is the application of the Potential Vorticity (PV) framework, a well-established method for studying atmospheric dynamics. We introduce a novel algorithm for identifying and tracking 3-D PV anomalies along the tropopause, offering new insights that traditional 2-D methods might overlook and addressing the associated challenges. Another application involves identifying and tracking PV features associated with African easterly waves (AEWs). These tropical waves are known to be linked to tropical cyclone development and contribute to high weather prediction uncertainty in tropical West Africa.
To support these advancements, a dedicated software framework has been developed, tailored specifically to the needs of meteorologists. This framework allows for the easy implementation of the proposed algorithms, with configurable options and templates for straightforward feature identification and tracking. Additionally, a near-real-time web display has been created to visualize identified and tracked AEWs, which evolved into a valuable tool for both researchers and operational forecasters.
Summarized, this dissertation bridges the gap between computer science and meteorology, showing how advanced computational techniques can improve the understanding and predictability of complex atmospheric phenomena. The developed tools and algorithms provide a solid foundation for future research and practical applications. They pave the way for more accurate weather predictions and a deeper understanding of meteorological processes.