next up previous print clean
Next: A filtering method Up: Noise attenuation by filtering Previous: A short review of

Proposed solutions to attenuate coherent noise

This section proposes two methods for approximating the noise covariance operator ${\bf C_d^{-1}}$ when coherent noise is present in the data. But first, let us define ${\bf d}$ as the sum of signal ${\bf s}$and coherent noise ${\bf n}$ as follows:
\begin{displaymath}
{\bf d} ={\bf s}+{\bf n},\end{displaymath} (18)
where n is made of the inconsistent part (or modeling uncertainties part) of the data d for any given operator L. The goal of this Chapter is to estimate earth parameters that are not biased by the noise present in the data.