A small section fragments corresponding to well defined rocks intervals determined with the information of the well  log 302,  were used as primary patterns for each seismic class. In total ten rocks types were linked to seismic patterns based on Multi-D attributes. The attributes that were chosen consisted of seismic section data and section transformed by high frequency filtering, attributes analysis, deconvolution with phase shift elimination, coherent summation and mixing on 5 traces base. Oil saturated siltstones corresponded to class 7.

Classified section (see figure) covers four stratigraphic units and shows very complicated distribution of rock types along with unconformity and truncations of the beds associated in some cases with small faults.

Each class in each point of the section obtained with Topology program is characterized by the value of wave shape (Multi-D attribute) similarity with corresponding primary pattern. If the result of classification for some point is not valid then contrast of classes is small and this number will be equal to 1/N, where N is the total number of the classes. In our case N=10. On the lower section of the figure, the yellow and red colors highlight the zones where probability of finding oil saturated siltstones is more than three times bigger than "noise level" equal to 1/10.

The results obtained with PMDI subsystem allowed to explain the testing results of the wells 305 and 314 and to delineate the parts of the section  with the maximal probability of finding oil saturated rocks.