Geostatistical Filtering of Refraction Data – Methods and Comparison of Acquisition Systems

Geostatistical filtering techniques, based on factorial kriging, are applied to improve the quality of seismic data such as velocity cubes or interpreted horizons. The classical application to 4D seismic data sets and history matching is particularly interesting for the identification of areas untouched by the oil&gas production process. More generally, when two seismic data sets are measuring the same area, hence the same underlying geology, it is advisable to extract jointly the artifacts contained in both data sets. A factorial cokriging method is then used to interpolate at the best the common component using both sets. The paper aims at illustrating the benefit of these techniques to mapping of near surface horizon from refraction seismic surveys. After reminders of the principles of geostatistical filtering techniques, the paper illustrates the benefit of using the proposed methodology for mapping the top of a near surface karstic reservoir by refraction surveying. Two different acquisition geometries (in-line shots and cross-line shots) provide the information on the structure of the top of the aquifer. The geostatistical filtering by merging both data sets allows to extract the relevant part of the information and to get more accurate maps of the geological structure.