Weather observations form a crucial part of meteorological forecasts and are also used as input to climate models and for monitoring long-term climate trends and changes. But coverage at sea is much less dense spatially and temporally, especially at high latitudes. Passive underwater acoustics has been established for several decades as a good and reliable tool to monitor the weather. The frequency spectra of different weather types have been well documented and acoustic disdrometers have been developed in several research environments. Nevertheless, many questions remain about the validity of some of these measurements, the best analysis approaches and the combination of different weather processes. This is particularly true in polar regions, where the loud and complex acoustics of ice-related processes adds to the difficulty of the task.
This thesis focuses on the analysis of a broadband dataset acquired in an Arctic fjord (Kongsfjord, Svalbard), in summer 2007. Taken at 6 locations from the mouth of the fjord to the glaciers at its termination, measurements cover the combination of varying levels of rain (from none to light rain), wind (from none to 11 km/h), ice (from none to growlers and bergy bits), shipping (from none to a large cruise ship) and animal activity (including whales and diving seabirds). The recordings covered frequencies from 100 Hz to 48 kHz. Principal-Components Analysis identified 3 distinct frequency bands mostly related to noise from wind, rain and ice. Laboratory-based tank experiments were conducted to assess the physical sources of these components (from 100 Hz to 100 kHz), confirming the acoustic role of ice and the relevance of the frequency bands identified by Principal-Component Analysis. These experiments also identified for the first time the role of higher (up to 45 kHz) acoustic frequencies in the identification of ice-related processes such as scraping, colliding and melting.
Principal-Component Analysis is shown to be a valuable and rigorous tool for identifying weather processes at sea, especially in complex combinations of wind, rain, ice and other factors. Analyses at frequencies higher than generally used also offer the potential of identifying specific processes associated to the melting of glaciers and icebergs. This has paved the way for field measurements at glaciers around Svalbard in summer 2009. The approach presented here is now considered by the Meteorological Office (UK) for inclusion on operational present-weather sensors attached to moored or drifting buoys in polar and high-latitude regions in general.
|Date of Award||1 Jan 2011|
|Supervisor||Philippe Blondel (Supervisor)|