Massively Parallel Seismic Data Wavelet Processing Using Advanced Grid Workflows

Leveraging the huge seismic data collections can be a quite challenging process, especially if the available data comes from large number of sources. Computing Grids enable such processing, giving the users necessary tools to share the data from various countries and sources. Processing this data not only gives results related to the earthquakes themselves, but also it reveals the geological features of the observed regions. Using the gLite base Grid, we propose a framework for massively parallel wavelet data processing of the seismic waveforms using advanced Grid workflows. Such workflows enable users to use the power of the Grid more easily and to achieve better performance. In the process of the data processing we use seamlessly several different grid services (AMGA, LFC ...) to locate the necessary data and to extract the needed information. The Grid application uses waveform data from several earthquakes from the same recording station. For the processing we use continuous wavelet transformation in order to capture the characteristics of the earth crust following the path from the earthquake origin towards the station. These features are recorded and later are classified using pattern matching to identify important characteristics of some specific seismic region as seen from that specific station.