Secondary microseism generation mechanisms and microseism derived ocean wave parameters, NE Atlantic, West of Ireland.

Thursday, 18 December 2014
Sarah Elizabeth Donne1, Christopher J Bean1, Ivan Lokmer1, Miguel Nicolau2 and Michael O'Neill2, (1)University College Dublin, Dublin, Ireland, (2)Computational and Adaptive Systems Laboratory, University College Dublin, Dublin, Ireland
Ocean waves, driven by atmospheric processes, generate faint continuous Earth vibrations known as microseisms (Bromirski, 1999). Under certain conditions, ocean waves travelling in opposite directions may interact with one another producing a partial or full standing wave. This wave-wave interaction produces a pressure profile, unattenuated with depth, which exerts a pressure change at the seafloor, resulting in secondary microseisms in the 0.1-0.33 Hz band.

There are clear correlations between microseism amplitude and storm and ocean wave intensity. We aim to determine ocean wave heights in the Northeast Atlantic offshore Ireland at individual buoy locations, using terrestrially recorded microseism signals. Two evolutionary approaches are used: Artificial Neural Networks (ANN) and Grammatical Evolution (GE). These systems learn to interpret particular input patterns and corresponding outputs and expose the often complex underlying relationship between them. They learn by example and are therefore entirely data driven so data selection is extremely important for the success of the methods. An analysis and comparison of the performance of these methods for a five month period in 2013 will be presented showing that ocean wave characteristics may be reconstructed using microseism amplitudes, adopting a purely data driven approach.

There are periods during the year when the estimations made from both the GE and ANN are delayed in time by 10 to 20 hours when compared to the target buoy measurements. These delays hold important information about the totality of the conditions needed for microseism generation, an analysis of which will be presented.