Curve Smoothing and Gamma Spiking

Curve Smoothing and Gamma Spiking in Correlating with MWD Gamma Logs

Successful correlation is all about pattern recognition. When the gamma is spiking (both high and low) it is the overall underlying pattern that must be sought. Most E-logs utilize curve-smoothing filters for the curves so that thin-bed geology may not be seen except perhaps in a high-resolution section. Curve values are averaged over a certain interval (usually 1 or 2 ft) to smooth the curves. Ideally, the high-res section would possibly be better to correlate against. A curve without smoothing is more jagged and the difference between a smoothed and non-smoothed curve is perhaps sort of analogous to smooth gamma vs. spiking gamma. However, it should be noted that spiking gamma may or may not be “correct.” It may be accurate in cases of a recurring zone of spiking gamma but it may not be accurate where spiking is random and generally not correlatable in small scale.

I do not know why gamma spikes in MWD tools as a few inquiries have gotten several possible answers: the way the mud is pulsing though the tool, minerals, interfering electrical signal, etc. It is true that certain zones do have recurring spiking gamma in multiple wells and multiple times in the same well, and those cases may well have to do with thin bed geology and minerals. Other cases occur after a zone has been crossed several times (as well as in other wells) with no spiking. In those cases it is likely due to with the mud/tool interactions. It seems possibly to often increase in those cases later in the well if my memory serves me. More recently I have seen spiking associated with tightly folded zones as well as what appeared to be highly faulted zones. Perhaps fractures and/or fracture fill could induce spiking as well.

Spikes can be filtered out simply by removing data points. It is a “no-brainer” when there is a single or just a few out-of-place spikes but when there are more and thicker spikes it is sometimes a judgment call whether it is an anomalous spike or not. 

After one sees a certain zone several times in a small area and its character becomes well worked out, one can work from a known correlation there and see the spiking and correlate around it. Thus, defining the character of specific zones is important. However, it is also true that a zone may show different gamma when passing through it at different times, particularly when the gamma becomes known to be “spiking.” Gamma character may also vary when passing through the same zone at different angles relative to dip.

Many geologists will actually name zones within target intervals according to their characteristics such as ROP, “rattyness,” gamma readings, gamma shape, samples, etc. Stratigraphically, zones can be defined by fining upward or coarsening upward sequences as lobes are often defined by particle size which in turn is often associated with energy conditions during deposition. Relative particle size can also sometimes be roughly correlated with silica content and relative brittleness of the rock. This is maybe more the case in tight sands rather than in shales. It is also sometimes the case in limestone stringers in shales where the hottest shale occurs just above the limestone indicating inundation and a return to low-energy, slow deposition, and more anoxic conditions conducive to high TOC shale development. The limey zones may display changing particle-size and energy conditions either up or down but it seems sometimes they coarsen upward or increase in shallow water/high energy conditions as they go up and are then inundated, resulting in hot black high TOC shale deposition.

The bottom line is that definitive curve matching is the key to successful correlation. Sometimes small zones will thin and thicken resulting in off correlations. There may also be deflections off of hard zones that can seemingly create anomalous zones. When a curve is not matching what is expected from recent avg. dip – an investigative process ensues where all of these possibilities are examined and gradually interpretations are developed with variable degrees of certainty.