Monday, January 30, 2017

KFSM Earthquake Insurance Interview

Just finished an interview with Charlie Hanna of KFSM channel 5 in Ft. Smith, AR. Topic was earthquake insurance. It will air sometime in the next couple of days. Below is my whiteboard drawing made for the interview.

Explanation of the figure: In N and NE Oklahoma horizontal wells are drilled into unconventional reservoirs, typically the Mississippian Lime. These wells produce vast quantities of formation (salt) water mixed with a small percentage of oil. The oil is separated and the water needs to be disposed. For this purpose a deeper nearby vertical salt water disposal well pumps the water down into the 500 million year old Arbuckle formation that sits on granite basement rocks. The basement and Arbuckle often have ancient inactive faults from continental collisions that formed the North American continent. Salt water injection that is too rapid, at too high a pressure or from wells too close together can reactive the faults and be felt as earthquakes. The earthquake waves radiate out from the fault and, sometimes, can be felt 100 miles away in Fayetteville, AR where I live. The hilly terrain of NW Arkansas scatters and focusses the earthquake waves so that the felt effect varies widely from one neighborhood to the next.


Sketch of NE OK earthquake as felt in Fayetteville, AR

Sunday, January 29, 2017

Seismic interpretation bootcamp

Yesterday I taught a 1-day seismic interpretation bootcamp short course for the Tulsa Geological Society. Everyone worked on their own laptop running OpendTect software and the Netherland's offshore F3 Demo data from the dGB website.

Kudos to Shane Matson for organizing and the 25 who attended. Lots of familiar faces in the room, including Dick Banks, Jamie Woolsey, and Eric Gross. The OSU-Tulsa campus as an excellent venue. It was a great day!

Bootcampers using OpendTect in Tulsa.
Bootcamp advertisement

Thursday, January 26, 2017

Lyell and Darwin


As some Seismos readers know, a passion of mine is collecting rare books. I have been at it since the late 1970s and now have about 120 titles that one could term rare.

The photo above is a very nice part of the collection, completed only this week with he addition of Charles Darwin's The Origin of Species. The 1876 version of the 6th edition of Origin is the basis for all modern reprinting of his most famous book. It was the last edition edited by Darwin himself before his death in 1882. Only 1250 copies were printed, making 1876 the smallest print run except the 1859 first edition of Origin that also numbered 1250.

Charles Lyell was an older contemporary of Darwin and a close colleague. They are together again on my bookshelf.

From left-to-right in the photo, we have:

Lyell, 1834, Principles of Geology, 4 vol (3rd ed)
Lyell, 1885, Students Elements of Geology (4th ed)
Lyell, 1865, Elements of Geology (6th ed)
Lyell, 1863, Geological Evidences of the Antiquity of Man (1st ed)
Darwin, 1876, The Origin of Species (6th ed)
Darwin, 1871, The Descent of Man (1st ed, 2nd issue)

A rather shabby first edition of Origin sold at Sotheby's in December 2016 for $102K.

Wednesday, January 11, 2017

Seismic Sherlock

When one works with students teaching seismic interpretation it can be a sinking feeling. All the basic elements get covered in the right order. Faults are picked, horizons are tracked, amplitude is extracted, attributes are applied. High tech, lots of workflow, parameters and button pushing. Then a question comes up and the student answer makes it clear something is missing. Perhaps after picking 20 faults the student has no clue of the structural style. Or noise is closely interpreted to have geologic meaning. Of course every mistaken concept is discussed and corrected, but it leaves the teaching like someone standing before a crashing surf having explained a few pebbles.

I've been thinking there may be a better way to jump start the learning process about seismic interpretation. No cubes, or N-level software applications, not even a colored pencil. Perhaps we need to think of a seismic section like a dead body in the morgue and the teacher is like one of those Victorian doctors who lectures to a classroom in the round as his assistant dissects the corpse, or  Sherlock explaining the geological implications and geophysical limitations of a seismic section...

Here is the patient (click to enlarge).
Data fade at the top of the section is due to shallow event muting and subsequent loss of CMP fold in the gather that has been summed to make each poststack trace we are interpreting. The mute cut is uneven, as evidenced by the irregular depth (that is, time) of the fade. Any interpretation in the mute zone is extremely hazardous, amplitudes are unreliable, structure is uncertain because missing offsets in the gathers make velocity analysis highly suspect. Only fools and lunatics interpret in the mute zone. Note the strong event at 250 ms on the left end of the line is undisturbed, no obvious lateral time or amplitude disruptions. At the same depth on the right side are lateral amplitude bursts with a 'sting of pearls' look. This is characteristic of spatial aliasing that arrises from poststack traces that are too far apart for the frequency and velocity at that location. The spatial aliasing effect in this area of the data is doubtless due to shallow (and therefore low) velocities, the effect largely subsiding by 400 ms.

Revisiting the strong 250 ms event, is shows as a peak-over-trough waveform, assuming black is positive amplitude and polarity is SEG normal. If it is further assumed that the seismic processor has done the job correctly and the wavelet is zero phase, it follows that the 250 ms event must be a seismic thin bed. A thin bed with zero phase wavelet, it is to be recalled, has the appearance of a 90 degree phase shifted wavelet, or a positive-over-negative as observed here. To quantify a bit with 10000 ft/s velocity and 55 Hz dominant frequency in this zone, the vertical resolution is about 45 ft, so the waveform of this event indicates it is less than 45 ft thick. Yet this event is very bright. With some basic investigation, one finds the surface geology is Pennsylvanian sandstone and shale, in short clastics. The possible lithologies that could give a strong reflection for a thin bed in an otherwise clastic section are anhydrite or carbonate (limestone or dolomite). Most likely a limestone, but well verification is needed.

At this point it is useful to jump to the deeper section and work back up the 250 ms event. This line is from the US mid-continent so we can safely assume the basement rocks are igneous, specifically granite. Not all granites are created equal with respect to velocity and density, but a fair guess is 21000 ft/s (6400 m/s) for P-wave velocity and 2.65 g/cc for density. Granitic basement is often layered and this is the case here. An autocorrelation test would quickly revel absence of multiples, so the basement layering is real. You can on the fact that the top of the basement is an unconformity surface.

The basement reflection event here is at 800 ms on the left, represented by a strong weaker trailing peak. If we are to have a negative amplitude event then what overlies the basement must have a higher impedance than granite. A tall order, but dolomite has just this property. Also, lateral variability of the basement reflection amplitude makes one suspect there is a zone of weathered granite or conglomerate at the basement contact. Notice in the center of the line there is an irregular group of hills representing erosional remnants or later uplift. In any case, we see the basement reflector fade to nearly nothing in this region, indicating unaltered granite overlain by tight dolomite. The impedance of these two rock types being almost equal, a case of matched impedance making the granite interface almost invisible.

The peak (dark) event on the LHS at 680 ms is likely a shale contact with tight dolomite below. Lateral variability of the event amplitude is probably related to thickness changes in the shale. On the left side, the disruption of event 680 may not be a geological reality. Immediately above we see a deep sinkhole indicating a limestone unconformity (limestone because it is far more soluble than dolomite); this unconformity extends across the entire line. If the sinkhole is filled with low-velocity rubble it will have the effect of a velocity sag. This is a false structure. On the other hand, if it were a velocity sag all deeper events should sag equally. But they do not, suggesting it is a real dissolution feature at the limestone unconformity and slowly healing with depth. How deep is this sinkhole? It looks to be about 30 ms

And speaking of the limestone unconformity at 610 ms (on the left), it is a doozy. In fact, the Missippippian-Pennsylvanian boundary that occurs in most places around the world. Typically, the overlying Penn rocks are clastics (shale and sandstone)Across the line the amplitude is not only irregular, but often changes polarity. It is also structurally irregular, pocked with hills and valleys. Here we have classic paleokarst on a limestone unconformity surface. Karst in the modern world is a landscape characterized by subsurface drainage through fracture networks and collapse features. Dissolution occurs due to atmospheric and soil biota CO2 forming acidic ground water that dissolves the limestone pretty effectively. The same process that rounds out the crisp lettering in old limestone cemetery markers. Once buried, the altered limestone near the unconformity  translates into highly variable acoustic impedance. For example, on the far right the unconformity pops as a strong negative (white) event, surely an area of deeply weathered limestone with associated high porosity and low acoustic impedance. But in the center of the line, we see the unconformity as a tight positive (black) event. Here we suspect overlying sandstone and shale in direct contact with hard, unaltered limestone.