Pathfinder dataset

In October 1995, Pathfinder Oil and Gas, Inc., made available to me a portion of a 3-D land data set. The original prestack data set was recorded on an irregular, 17.5 square-mile grid. The sampling rate in time is 2 milliseconds over a 3 second total time recorded. There are 65 wells drilled on the site of the survey. The available data set comprises a poststack time-migrated 3-D seismic cube (3.6 squared miles) and six sets of well-log curves recorded at six different locations on the survey site. Figure 4 shows the interval velocity map for a particular horizon of the survey and the six well locations. The two southern wells (E and F) are dry holes.

 

horizon
Figure 4 Interval velocity map for one of the main reflectors on the site of the survey. In the four wells at the center of the figure, oil has been observed, and two of them are producing oil. The wells E and F at the bottom of the figure are dry holes.

As of the date of this article, I have not received the corresponding 3-D prestack shot gathers and the stacking velocity model used to migrate the seismic data. I will also obtain four more well logs. The well-log curves available are two sonic logs, six gamma ray logs, as well as other curves (induction, self-potential, porosity, and density). No checkshots have been recorded in any of these wells.

The geology for the area is not structurally complex: at the depth of the target zone, the geology shows prograding carbonates and clastic sediments embedded in thin sand layers in a shaly environment. The prograding of the carbonated series makes it difficult to predict the extent of the oil reservoir reached by the four northern wells. Several dry holes have been drilled already in this area because of the inaccuracy of the velocity model used for the time-to-depth conversion.

For this data set I worked with the six well-log curves available. I blocked by hand all the gamma ray curves in order to do visual correlations between the various wells. I defined nine major interfaces on one curve and retrieved them on the five other curves. These interfaces bound regions of the curve where the gamma ray values are roughly constant (high-frequency oscillations around a mean value). I also blocked by hand the two sonic curves and retrieved the same nine interfaces as determined on the gamma ray. Figure 5 shows a portion of the sonic and gamma ray curves around the target zone. The high-frequency component of the original logs has been removed by median filtering on a running window. The four curves present strong similarities exploitable for interpretation and visual correlations. I then plotted several scattergrams showing the existence of a correlation between the sonic and the gamma ray curves at the two wells where both pieces of information were available. I also observed a strong correlation ($\rho_{linear} \geq 0.94$) between the blocked sonic and the gamma ray logs. Figure 6 shows this correlation on two scattergrams of sonic median filtered values plotted against gamma ray median filtered values for the nine layers defined around the target zone.

 

dt-gr.88-93
Figure 5 Sonic and gamma ray curves for wells E and F. The high-frequency component has been filtered out by median filtering on a running window.

 

sub-dt-gr
Figure 6 Scattergrams of sonic versus gamma ray values for nine layers around the target zone.