Modelling the data Data reduction Noise evaluation

Electric field decomposition

The electric field amplitudes recovered from the raw time series using the least squares fitting procedure were screened for a signal to noise ratio greater than two. Most of the data passed this screening with the exception of those from the exceptionally noisy ELF Kermit; no tow 1 signals were recovered, while only a small amount of the data from tow 2, recorded at the shortest range (3-4 km), were included in the modelling. After screening for signal to noise ratio, the remaining amplitudes were corrected for the height of the transmitter above the seafloor assuming simple exponential decay of the fields, corrected to take into account the frequency response of the instruments, and finally normalised by the source dipole moment, calculated from DASI logger data and calibrations during pre-deployment deck tests.

The data recorded by the two orthogonal channels of the ELF instruments were further decomposed to yield polarisation ellipse parameters (Smith & Ward, 1974; Constable & Cox, 1996). The complete dataset is shown in Figure. 3. Error bars show one standard deviation errors associated with the least squares fitting procedure. The minimum error was set to 20% for the 0.35 Hz and 0.75 Hz data, and 30% for the 11 Hz data, based on an internal scatter in the data similar to that observed by Evans (1991), in the CSEM data collected at 13°N on the East Pacific Rise. It is assumed that this scatter is caused by lateral resistivity variations in the crust over which the source is moving (Evans etal, 1991; Unsworth, 1994) due to water or sediment filled cracks, metalliferous deposits or variations in temperature or porosity. The larger scatter in the higher frequency 11 Hz data is consistent with this assumption.


Modelling the data Data reduction Noise evaluation

Go to first page