Improving the spatial resolution of effective elastic thickness estimation with the fan wavelet transform

Abstract

We show here a simple technique to improve the spatial resolution of the fan wavelet method for effective elastic thickness (Te) estimation that we have previously developed. The technique involves reducing the number of significant oscillations within the Gaussian window of the Morlet wavelet from approximately five to three or fewer (while making an additional correction for its no-longer-zero mean value). Testing with synthetic models and data over South America indicates that the accompanying reduction in wavenumber resolution does not seriously affect the accuracy of the Te estimates. Comparison against the more widely-used multitaper Fourier transform approach shows that the enhanced wavelet method not only improves upon the multitaper method's spatial resolution, but also is computationally much faster and requires the arbitrary variation of only one parameter compared to three for the multitaper method. Finally, we present a modified method to compute the predicted coherence using the multitaper method that, while not improving its spatial resolution, does improve the bias of recovered Te estimates.