Objectif The aim of the project is to improve the resolution of marine seismic data by improved data acquisition methods and corresponding improvements in the data processing. In particular, an accurate description of the angular-dependent source wavefield must be provided and new data processing software developed to utilize this information.The validity of three new methods for specifying the source wavefield for marine seismic data has been tested. The interaction method and the source decomposition method had already been validated before the start the project with the result that the interaction method was viable, successful and likely to be easy to use, and the source decomposition method was unsuccesful. The third method, the scaling method, has been tested within the project, and was very successful, surviving an experiment in which it stood at risk of being refuted. However, it is not easy to use. The conclusion is that the interaction method can be used in production to determine the full wavefield of the source.Also, software has been developed and tested for taking account of source directivity during processing. The source directivity can be specified if the interaction method is used in acquisition. The angular dependent deconvolution can be performed on CMP gathers in the frequency wave number domain ifthe dip of the layers is less than 10 degrees, and if the target depth is very large compared with the dimensions of the source array. With these assumptions, the software that has been developed performed well in tests and was subsequently applied to real data. The software was applied to the data from the Delft Air Gun Experiment, and the expected improvement in resolution over current acquisition and processing methods was tested. It is shown that, even when the far field source signature in the vertical direction is known from measurements, the optimun signature deconvolution of the data is inferior to the angular dependent deconvolution of the data. The increase in resolution at the target depth is noticeable, as expected. It is shown that the conventional statistical approaches to deconvolution fail to extract the known wavelet, probably because the whitness assumption for the earth reflection response is not valid for seismic data.The whole improved data acquisition and processing route was tested by comparing the results with offset-dependent synthetic seismograms calculated from well data. The forward modelling is done by calculating individual plane wave responses using the reflectivity methodand a matrix extension for the elastic case, superposing the plane waves to create the point source response, and then superposing the point source responses using the notional source concept. In this scheme the biggest problem has been the superposition of the plane waves via the Fourier-Bessel transform. Six different schemes have now been developed with various trade-offs between accuracy and speed. Despite this very large effort into the modelling of synthetic seismograms from the well log data, we did not succeed in getting a very good match with the measured data. However, we feel that the real reason for our lack of success here is that the well logs are not accurate.A data set already exists consisting of a seismic line shot over a logged well in the North Sea. The line has been shot with a number of different source configurations and in each case the source wavefield can be determined. The well logs provide a check on the processing of the data in two ways: first, synthetic seismic data can be calculated from the well logs and compared with the real data; second, the real data can be deconvolved and inverted to obtain density and velocity logs that can be compared with the real well logs. Programme(s) ENG-HYDROCARB 1C - Programme (EEC) of Community projects in the hydrocarbons sector, 1973-1985 Thème(s) 1.4 - GEOPHYSICS AND PROSPECTING Appel à propositions Data not available Régime de financement DEM - Demonstration contracts Coordinateur COLLEGE VAN BESTUUR DELFT UNIVERSITY OF TECHNOLO Contribution de l’UE Aucune donnée Adresse BUREAU HOGESCHOOL COLLEGE VAN BESTUUR P.O. BOX 5 2600 Delft Pays-Bas Voir sur la carte Coût total Aucune donnée
