Upheaval Dome Seismic Refraction Analysis

Preliminary work in progress by John Louie, 8/7/95.
On-line document at http://www.seismo.unr.edu/ftp/pub/louie/dome/refraction.html

Setting

This paper shows a preliminary analysis of the seismic refraction data collected during January 1995 at the Upheaval Dome structure in Canyonlands National Park, southeastern Utah. Upheaval Dome has been proposed to be the deeply-eroded remnant of a 10 km wide meteor impact that may have occurred between 5 and 100 Ma. However, it also bears some resemblance to the many salt anticlines that occur in the region, such as the Moab anticline. The salt diapirs arise from the Paradox Formation, which also underlies Upheaval Dome. Thus the 1995 seismic refraction experiment, and a companion seismic reflection profiling experiment, were designed to constrain the structure of the top of the Paradox Formation below and near Upheaval Dome, at minimal cost and environmental impact. Deformation driven by salt diapirism from below should show increasing amounts of deformation with increasing depth, while deformation driven by cratering from above should show the greatest amount of deformation at the surface.


Fig. 1: Map of the January 1995 refraction deployment (click on the image for a viewable Acrobat PDF file, 1.2 Mbyte). The yellow stars show three of the five 700 lb hammer source placements. AR0 is at the Upheaval Picnic Area, TH is at the Wilhite Trail Head, and WH is on Horsethief Point. Additional source locations at Willow Flat and the Island in the Sky Visitor Center are not shown on this map. The blue triangles show the locations where strings of eight 8 Hz reflection geophones were placed for each of the three channels per RefTek recorder. The RefTek serial numbers are given. The map shows channel locations only for the 10 recorders that yielded data, out of twelve placed in the field.


Fig. 2: Dual-tool acoustic log for the wildcat petroleum exploration well Buck Mesa #1, located in the ring syncline of Upheaval Dome. (A viewable Acrobat PDF file is available.)

At about 2000 ft depth sandstones averaging 4300 m/s possibly transition to limestones averaging 5400 m/s. The top of the Paradox salt is at a depth of 4000 ft, where the log shows massive salt apparently averaging 4700 m/s, with harder and softer interbeds. Note that the logs, from two different acoustic tool lengths, differ substantially only within the intervals of massive salt, a sign of significant borehole erosion and widening during drilling (with water-based mud), or of pervasive in-situ fracturing. Given the >5 Ma age of the Upheaval Dome structure, fractures should have healed. Thus it is more likely that the bulk velocity of the massive salt intervals, in situ, is closer to the value of 5500-6000 m/s indicated by the harder interbeds.

Data Stacks

Due to the location of the refraction survey within a National Park, and an area of extreme topographic relief, a specialized design was executed for the seismic refraction survey. A source consisting of a trailer-mounted Bison EWG 700 lb enhanced weight-drop mechanism could be placed only along passable roads. More intense energy sources such as explosives could not be used. Given the distances of 1 to 10 km to be traversed by the refractions needed to undershoot Upheaval Dome, each seismic record of a source location would have to sum up hundreds of separate impacts by the weight-drop source, to assure that the refraction arrivals would be visible above the noise level.

This strategy required a period of continuous recording, more than one day, to execute all the impacts, and very accurate timing coordination between the source and the far-flung receivers. We used twelve RefTek recorders provided by the Incorporated Research Institutions for Seismology (IRIS), each having a GPS satellite clock receiver. One recorder was placed at the source to record the GPS time of each impact, and the GPS clock receivers at each receiver away from the source noted GPS times on the refraction data. The coordination with GPS time allowed the receivers and source to not require any communication among them, and produced timing accuracies at the millisecond level, better than the 8 ms data sampling interval.

The continuous data from the source had to be picked for the impact times, and the continuous data from the receivers had to be gathered into time windows of 10 s following each impact. Each source location received between 160 and 867 pickable impacts. The sections shown below are the products of vertical stacking of each receiver channel on the post-impact time window. The gather from each channel was plotted to pick the interval of lowest-noise impacts over which to stack. To distances of about 5 km, plots of the gathers before stacking showed visible wave arrivals, but more distant arrivals were usually invisible until after stacking.


Fig. 3: Stacked traces across Upheaval Dome from the Upheaval Picnic Area (AR0) hammer source position. (A viewable Acrobat PDF file is available.)


Fig. 4: Stacked traces across Upheaval Dome from the Wilhite Trail Head (TH) source position, about 5 km further off-end to the southeast. The traces are plotted on the Picnic Area offset axis. (A viewable Acrobat PDF file is available.)


Fig. 5: Stacked traces across Upheaval Dome from the Willow Flat (WI) source position, an additional 4 km further off-end to the southeast of the Wilhite Trail Head. The traces are plotted on the Picnic Area offset axis. (A viewable Acrobat PDF file is available.)


Fig. 6: Stacked traces across Upheaval Dome from the Horsethief Point (WH) source position, about 4 km off the northwest end of the receiver line. This consititutes a reverse shot, but traces are still plotted on the Picnic Area offset axis. (A viewable Acrobat PDF file is available.)


Fig. 7: Stacked traces across Upheaval Dome from a source position at the Island in the Sky Visitor Center (VC), about 25 km northeast of the receiver line. This consititutes a fan shot from the side, but traces are still plotted on the Picnic Area offset axis. The distance was too great for the EWG source, despite stacking over 200 hits into each channel. (A viewable Acrobat PDF file is available.)


Fig. 8: Time-distance plot of first arrivals across Upheaval Dome picked from the Upheaval Picnic Area (AR0; yellow squares), Wilhite Trail Head (TH; yellow diamonds), and Horsethief Point (WH; yellow triangles) hammer source locations. The red symbols locate the profile position of each source, at zero time. (A viewable Acrobat PDF file is available.)

The simplest interpretation of these picks would measure an approximately 4300 m/s surface layer, underlain by a 5300 m/s refractor dipping north at an angle approaching 13 degrees, to explain the differing apparent velocities of the two directions of shooting. The difference is particularly evident in the early arrivals from the Horsethief Point (WH) source location at distances greater than 5 km. These early arrivals are manifest in the WH record above, where the presence of coherent seismic energy establishes them positively. Only late arrivals may be explained as mis-picks due to poor data quality.

The 5300 m/s refractor dipping 13 degrees north would locate at a depth about 1.2 km below Horsetief Point, in accordance with the depth to the Paradox salt below the ring syncline shown in the Buck Mesa #1 well. However, to allow the salt to rise at a 13 degree dip toward Upheaval Dome (thus surfacing in the Dome) would imply the existence of a salt uplift averaging more than 500 m in height and 15 km in diameter. No such broad-scale deformation can be observed in the vicinity of Upheaval Dome; abundant stratigraphic exposure shows that formations more than 2.5 km from the center of the dome have their regional attitude, and are quite undisturbed. For the same reason, the early arrivals cannot be caused by a thickening of the 5400 m/s limestone layer seen capping the salt in the well log.

One hypothesis for the origin of the high-velocity ramp would be a region of cementation centered about Upheaval Dome, producing the high-velocity refraction. The White Rim member of the Cutler Formation, for instance, is very well cemented by calcite near Upheaval Dome, and weathers to spires within the Dome's central depression.

Refraction Solution Methods

We take two approaches to using the picked refraction times to resolve the velocity structure below Upheaval Dome. First is an attempt to use forward-modeling of candidate subsurface structural configurations, to match the observed time picks. The candidate models will include our intuition, and biases. Second is an inversion approach that automatically projects structures from the time picks, yet lacks constraint by most geological knowledge.

Foward-modeling method involves simply drawing velocity polygons in the profile cross section, and then computing the travel time through this 2-d model section from each source location. We use Vidale's (1988) finite-difference solution of the eikonal equation to compute times on gridded velocity models, after 3x3-point smoothing. This approach allows rigorous inclusion of the extreme topopgraphy, as well as accurate computation of refraction times from interfaces having arbitrary attitudes.


Fig. 9a: Simple model velocity section across Upheaval Dome assuming a 1 km/s velocity difference between the Paradox Fm. and overlying rocks, and a 1 km depth to the Paradox below the central depression. A 500 m wide salt diapir is optionally placed below the Dome structure. (A viewable Acrobat PDF file is available for this and the two time images below.)


Fig. 9b: Refraction travel time image showing seismic wave propagation through the model having a salt diapir. The source is located at the Upheaval Picnic Area (purple for very early times), and the section extends across Upheaval Dome to Horsethief Mesa (brown for very late times). While the velocity model is very simple, topographic variations are handled exactly.


Fig. 9c: Cross-sectional image showing the difference in refraction arrival time between a model containing a salt uplift, and a model without any salt uplift. Maximum time advance due to the modeled salt uplift (light brown) would be about 30 ms at the surface, and is fairly constant on the side of the structure away from the source. The amount of advance is controlled by the width of the salt uplift, while its distance of onset is controlled by the depth of the top of the uplift.

Forward-Modeling Results

For each of the trial models we show below, the velocity model in cross section follows a time-distance plot. The yellow symbols show data picks, with Picnic Area (AR0) source picks as squares or circles, the Wilhite Trail Head (TH) source picks as diamonds, and the Horsethief Point (WH) source picks as triangles. Each plot shows the computed travel times from two models: a background model (thinner orange line) meant to match the near-distance picks; and the refractor model (thicker blue line) displayed immediately below each time-distance plot.

The background model is identical for all the plots. It assigns a velocity of 4000 m/s to the water-saturated sandstones, and all other rocks, at the canyon-bottom level and below (orange on the model sections). Above, where groundwater has been sapped by springs in the canyons, the model gives the Chinle, Wingate, and Kayenta formations a 3500 m/s velocity (yellow on the model sections). Other than the water table, the background model includes no refractors or deeper structure.

The top of each model section is at an elevation of 1800 m above sea level, and all depths are given relative to that. 1800 m is approximately the mesa top elevation of the Island in the Sky.



Fig. 10: This simple model adds a horizontal 5000 m/s Paradox formation refractor at 1300 m depth (elevation 500 m), and a 500 m wide Paradox diapir (uncolored) that approaches the central depression of Upheaval Dome, to the background model. (A viewable Acrobat PDF file is available.)

Clearly the background model without the Paradox (orange) does an excellent job matching the picks, to offsets of about 4 km. Note how it establishes that the variations in AR0 picks at 2-3 km offset are completely due to topographic variations (A). Introducing the Paradox refractor helps explain some of the longer-offset picks (B), but the shallow high-velocity dome makes the computed times (blue) noticeably early at near offsets on the north side of Upheaval Dome (A).



Fig. 11: In an attempt to explain the very early times observed from medium offsets from source TH and from long offsets of source WH this model adds a shallow, high-velocity 5400 m/s refractor at 800 m depth (uncolored). This depth corresponds to the top of the high velocity rocks seen above the Paradox in the well logs. (A viewable Acrobat PDF file is available.)

This trial shows, however, that such a shallow refractor may only exist in a limited area, for it is too early for the near-offset times from source WH, and too early for the far-offset times from source TH, both recorded below Buck Mesa at the north end of the model.



Fig. 12: This model adjusts the shallow high-velocity reflector to form a ramp (uncolored), the north side of which is required to explain the less advanced picks from Buck Mesa. Yet pick times from TH are still late, and times from WH are still early. (A viewable Acrobat PDF file is available.)



Fig. 13: This model attempts by design to provide a steep, high-velocity ramp (uncolored) for the very early arrivals from WH, while leaving the area below Upheaval Dome free of shallow refractors that would advance the times from sources AR0 and TH. (A viewable Acrobat PDF file is available.)

This model of a "ring salt anticline" or other high-velocity feature below Syncline Valley does the best at explaining the very early arivals at longer offsets from the Horsethief Point (WH) source location. This fast body, however, still unacceptably advances the times to Buck Mesa from the Picnic Area (AR0) and Wilhite Trail Head (TH) source points. It also shows that no salt uplift or high-velocity refractors are needed to explain data picks south of Syncline Valley from those two sources.

The extensional mechanics of the cratering process may assist the development of the ring salt anticline. The excavation of the crater creates a mass deficit into which salt may flow, except below the central peak. The gravitational origin of the slump creating the central uplift will remove the gravitational imbalance at the center of the crater. Near the edges of the crater, however, salt will flow to lift the crater floor into balance.



Fig. 14: This model combines a relatively low-velocity 4700 m/s salt at 1.3 km depth (uncolored) with an overlying bulge of cemented formations (red) that both would match the velocities encountered in the Buck Mesa #1 well (at the 0 km offset in this profile). The 4700 m/s salt diapir (uncolored) has a lower velocity than the cemented zone (red), but a 400 m/s higher velocity than the water-saturated sandstone (orange). (A viewable Acrobat PDF file is available.)

This model is a good compromise, save for predicted arrivals that are much too early at Buck Mesa from the Wilhite Trail Head (TH) source (yellow diamonds). The diapir is not in agreement with the picks, as the model is too early between 2 and 3 km offset for the Picnic Area source (AR0; yellow circles). The mis-match is more prominent from the more distant Wilhite Trail Head source (TH; yellow diamonds), where the diapir would cause 0.1 s of advancement that does not agree with the best 3 out of 5 picks from the central depression.

Preliminary Conclusions