The problem of the ghost
A source that is placed under the surface will generate energy in all directions. The up going wave field goes towards the surface, and is reflected and propagated down in the media. If the source is placed close to the surface, the ghost is regarded as a short periodic signal. It will be added to it’s original down propagating wave field and contribute to changes in the signal form.
List of uncertainties:
-Dependent on Reflection coefficient knowledge.
-Waves is affecting both ghost delay times, and reflection coefficients.
-Streamer depth measurement.
-Angle of incidence dependent.
-The receiver ghost parameters will vary both spatially and temporally within a shot record.
-Difficult to estimate as affected by attenuation.
What is good?
Due to the above mentioned ”problems” and the fact that the reflection coefficient is never at unity, there is never hard zeroes in the recorded frequency spectra and a deghosted spectra can be reconstructed using various techniques.
Inverse deghosting operator method.
The method is based upon inverting for the optimum ghost parameters (reflection coefficient, delay) from the data.
It is a stable operator if you have a reasonable correct estimate of the ghost delay and angular reflection coefficient.
The reconstructor of the operator is specially sensitive and suffers from variation in streamer depth and sea waves hight.
It can be computed in XT domain considering local constant ray, angle or in linear (tau-P) domain to account for actual wave propagation.
The optimum deghosting parameters can be analyzed and filtered in a surface consistent manner.
Wavelet extraction method.
Variation of traditional blind deconvolution method.
Based on the assumption of white reflectivity.
Suffers of effect of geology on the reflectivity.
Joint Inverse deghosting and Wavelet extraction method.
This is our preferred approach involves:
1.Inverse receiver deghosting (either in local XT or Tau-P).
2.Inverse source deghosting (either in local XT or Tau-P).
3.Spatially and temporally variant wavelet estimation and deconvolution (either in local XT or Tau-P).
We have done deghosting for many different configurations:
Normal marine configuration Source 6-7 m Receiver 7-9 m
Normal Site survey source 2-3 m Receiver 4 m
Deep towed Site survey (Slanted in the beginning of the streamer) Source 2-3 m Receiver 10-20 m
updated in 2014