We tested a numerical inversion of the Ricker wavelet as an initial guess for a bidirectional deconvolution scheme. This Ricker wavelet inversion has two symmetric parts, which makes it suitable to be the initial guess. We tested our inversion of the Ricker wavelet on a field data example, and the the inversion of the Ricker wavelet initial guess did improve the bidirectional deconvolution result (with a suitable
parameter). We found there is some instability in our scheme if we use different filter length for our deconvolution filter and do not have a explanation of it. We will continue researching on this.