Date of Award
Doctor of Philosophy in Nursing
This dissertation presents a two-part study of persistent sea surface temperature (SST) fronts based on global satellite observations. Its main goal is to achieve a better description of these features at the global scale, while reducing the bias in defining them. To this end, an algorithm was developed to detect and track persistent fronts by relying on frontal probabilities. Frontal probability (or frontal frequency) can be defined as the probability to observe a front relative to the cloud cover. Long-term mean (29 years) and seasonally averaged frequency maps were produced, showing the location of over 2,000 SST fronts for each map. The algorithm, denoted automated detection algorithm or PFDA, is based on a simple concept, which is the determination of local maxima of frontal probability by scanning the fields along lines of constant latitude and of constant longitude. The ADA agrees well with satellite observations, and performs well in detecting fronts with complex shapes or western boundary currents. This method can be applied to other types of datasets, such as chlorophyll fronts, and presents the crucial advantage to remove subjectivity in finding fronts. Frontal properties, derived from the Cayula-Cornillon algorithm were added to the geographic coordinates of the fronts. Furthermore, the bathymetry and bathymetry gradient magnitude are also part of the information provided by the PFDA at each frontal pixel.
Subsequently, we applied multivariate statistical analysis tools to evaluate the presence of patterns among persistent fronts. We performed a principal component analysis, followed by a k-means clustering procedure to partition fronts into different types. Unsupervised machine learning applied to the problem of oceanic fronts resulted in nine clusters, namely four clusters corresponding to shelf and shelf-break fronts, two clusters representing the subpolar frontal system, one matching the Kuroshio and Gulf Stream extensions, and the two last ones grouping the boundary currents and upwelling fronts respectively. The partitioning was conducted based on the characteristics of the fronts, in an objective manner. Be- cause this study was performed globally, we gained significant insights on poorly- documented and newly-found SST fronts, but also on fronts that were previously studied. These results also shifts the traditional way of understanding oceanic fronts. In particular, it is interesting to see coastal and equatorial upwelling fronts in the same cluster, just at it is intriguing to observe western and eastern boundary currents being part of one unique cluster. While this study is mostly focused on the statistical description of persistent SST fronts, its findings will certainly bring a broader understanding of ocean's submesocale dynamics, and is expected to benefit the Oceanography community in the future.
Mauzole, Yackar L., "Dynamical Typology of Sea Surface Temperature (SST) Fronts Based on Satellite Observations" (2017). Open Access Dissertations. Paper 592.