HYDROCARBON FLUX VARIATIONS IN NATURAL AND ANTHROPOGENIC SEEPS
Jones, Victor T., and Burtell, S.G.

Table of Contents

Abstract
Introduction
Migration Model
Leakage from Underground Storage Reservoirs
Leakage from an Underground Salt Dome
Leakage from a Mined Propane Storage Cavern
Barometric Relationship to Gas Flux
Migration Timing Related to Pressure Pulse
Injection of Helium Tracer into Cavern
Leakage from an UGC Reactor
Injection of Helium Tracer into UGC Reactor
"Hydrocarbon Spots"
Deep Mobile Gases/Relation to Earthquakes
"Helium Spots"
Helium-Hydrogen Associations/Duchesne Fracture Zone
Earthquake Predictions Based on Gas Flux
Arrowhead Hot Springs
Geology
Continuous Gas Monitoring
Methane
Propane
Propane/Methane x 1000 Compositional Ratio
Helium
Stable Carbon Isotopes
Helium Isotopes
Mercury Associations
Origin of Arrowhead Hot Springs Gases
References Cited


LIST OF FIGURES

Figure 1 Marine sniffer anomaly over a geochemical brightspot illustrating association with deep seismic faults.
Figure 2 Frequency distribution of the sum of methane homologs in deposits of different types.
Figure 3 Compositional separation similar to percent methane from near-surface soil gas data sets collected by Gulf Research.
Figure 4 Surface geochemistry - direct detection of hydrocarbon seeps over Pineview Field.
Figure 5 Ryckman Creek fields,. (Utah - Wyoming Overthrust Belt).
Figure 6 East Texas major faults are mapped at the surface permitting verification of the direct physical relationship between faults and magnitudes of anomalies.
Figure 7 Green Canyon area in the Gulf of Mexico demonstrating the deflection and control of seepage pathways by growth faults.
Figure 8 Seep technology -(John Hunt AAPG 1981) 70% reserves related to visible seeps.
Figure 9 Surface seeps along a thrust fault. Infantas field in Colombia was observed to have hissing gas seeps when first discovered.
Figure 10 Content of Hydrocarbons from drilling mud as a function of tectonic activity.
Figure 11 Distribution of hydrocarbons in near-surface environment, as a function of pressure; Kirgizya Area.
Figure 12 Link (1952). Migration pathways for macroseeps.
Figure 13 Hypothetical fracture zone model from Patrick Draw.
Figure 14 A gas bubble emerging from one of the larger mud craters.
Figure 15 Gas vents in the Cimarron River deposits, reflect fracture orientation in underlying Flowerpot Formation, Geotimes Magazine.
Figure 16 Leakage of ethane/propane from a salt dome storage well.
Figure 17 Distribution of ethane/propane product from salt dome.
Figure 18 Nitrogen front rapidly passed through the thirty foot aquifer in this area.
Figure 19 Lithologic cross section, based upon the color of augered soil samples and semi-log profile of original propane values, from ten foot depth data over propane storage cavern.
Figure 20 Rainfall and barometric pumping of hydrocarbons over propane storage cavern.
Figure 21 Propane recharge above propane storage cavern - (October 1981) Majority of leakage came from around the central shaft.
Figure 22 Helium recharge above propane storage cavern - (October 1982).
Figure 23 Regional geology of the southwest flank of the Rawlins Uplift, Carbon County, Wyoming showing location of the GR & DC, U.C.G. Facility.
Figure 24 Sandstone joint orientation summary showing probable product gas migration paths from the North Knobs, Wyoming UCG report.
Figure 25 Geologic cross section at Rawlins No. 1. Configuration of nearly vertical beds which dip too steeply to do anything except gasify the coal in-place.
Figure 26 Coalbed methane site. Looking west downdip.
Figure 27 Coalbed methane site. Looking north along the strike of the beds.
Figure 28 Surface Geologic Map. 122,-eighteen foot deep permanent sites were installed over general area of the retort.
Figure 29 Coalburn survey - methane flux - plots of selected methane magnitudes showing typical changes noted in time.
Figure 30 Coalburn survey - propane flux from selected sites.
Figure 31 UCG near surface geochemical survey. Methane variations over time.
Figure 32 UCG Near surface geochemical survey. preburn methane (ppm).
Figure 33 UCG Near surface geochemical survey pressure -methane (ppm).
Figure 34 UCG Near surface geochemical survey -burn A -methane (ppm).
Figure 35A UCG Near surface geochemical survey -burn B - methane (ppm).
Figure 35B UCG Near surface geochemical survey -one year after startup -methane (ppm)
Figure 36 Mercury in earthquake zone - (Fursov 1968) - first detection of earthquake.
Figure 37 Dependence of gas seepage on earthquakes - Mukhto Oil Field - NW Sakhalin.
Figure 38 Gas concentrations in the near surface rocks before and after earthquake.
Figure 39 Gas content in the subsoil rocks (vol. percent) measured in observation wells before and after earthquake of October 5, 1975. 1%-methane content 2%-content of heavy hydrocarbons.
Figure 40 Variations in the content of methane in air, as a result of seismic shock to the ground.
Figure 41 Comparison of the variations in methane content in air with the displacement of the crust in a seismically active region.
Figure 42 San Andreas Fault detail - Helium/hydrogen measured by crew.
Figure 43 White line on Highway 46 was offset by the 1966 Parkfield earthquake. Offset was only two inches during the first day (June 22), but had lengthened to five inches by August 4.
Figure 44 Geochemical traverses which run across and along strike with the Duchesne fault.
Figure 45 Duchesne Line 2 - Profile run during July 1976 which extended along strike and follows the fault zone for a considerable distance.
Figure 46 Duchesne Fault detail - More detailed traverse conducted perpendicular to strike in December 1976 - (10) closely spaced sites.
Figure 47 Duchesne Fault detail - Helium/hydrogen anomaly on close-spaced detail.
Figure 48 Photograph of faulting in bluff east of site 7 in fault detail area.
Figure 49 San Andreas Fault -Earthquake epicenters from Caltech database on 11/23/81.
Figure 50 Caltech radon monitoring sites (with deployment sequences) plus Arrowhead Hot Springs.
Figure 51 Pacoima Dam and its fractured rock walls - mass loaded earthquake site.
Figure 52 Close up view of Pacoima Dam and its fractured rock wall which includes geochemical hut.
Figure 53 Pacoima helium/hydrogen data in ppm- 75 ppm H2 anomaly just before Imperial Valley earthquake.
Figure 54 Photograph of Caltech station well showing radon and carbon dioxide measurement.
Figure 55 Methane bubbles rising to the surface in Arrowhead Springs concreted hot well.
Figure 56 Harmon Craig Arrowhead Springs - Helium plot (1975-81) - Increase in gases before the Big Bear earthquake.
Figure 57 Arrowhead Springs (initial data collected in cylinders) hot springs gases contain methane, ethane through butanes, helium, hydrogen. Very mature methane isotope.
Figure 58 Arrowhead Hot Springs - John Welhan at Scripps analyzed Gulf Research sample collected on 5/82 confirming methane value in percent of residual gas.
Figure 59 Geomorphology of the San Bernadino mountains and its relationship to major faulting.
Figure 60 Location of Arrowhead and Waterman Hot Springs on San Bernadino north quadrangle with respect to major faults.
Figure 61 Hot springs in the field area are divided into two canyons approximately 1/2 mile apart on the Campus Crusade for Christ International property. (Methane and temperatures).
Figure 62 Plexiglas sample collection bucket was installed above a stainless steel collection pan placed at the bottom of the concreted hot spring.
Figure 63 Arrowhead Springs methane data from December 1981 to December 1982.
Figure 64 Arrowhead Springs propane data from December 1981 to December 1982.
Figure 65 Arrowhead Springs propane/methane X 1000 from December 1981 to December 1982.
Figure 66 Arrowhead Springs helium data from December 1981 to December 1982.
Figure 67 Contoured mercury magnitude shows two distinct and well controlled zones of anomalous values defining hot wells.
Figure 68 Anomalous Arrowhead mercury data - Line A9.
Figure 69 Anomalous Arrowhead mercury data - Line A2.
Figure 70 Background Arrowhead mercury data - Line A5.


ABSTRACT

The methodologies for conducting surface geochemical surveys and the hydrocarbon flux rates with which hydrocarbons migrate to the surface will be addressed with numerous examples from both natural and manmade seeps. Manmade seeps are available for the study from underground gas storage reservoirs, leaky well casings and underground coal gasification reactors.

Natural gas flux was monitored for approximately one year (12/8/81 to 12/1/82) on a semi-continuous basis from a natural 80 degree Centigrade spring at Arrowhead Hot Springs near San Bernardino County, California as part of an extensive "Earthquake Prediction Program" jointly operated by Gulf Research and Development Company and the Kellogg Radiation Laboratory at Caltech University (Burtell)1. The hot spring chosen for monitoring, which is located on a splay of the San Andreas fault, releases significant quantities of free gases (40 cc/minute), including helium, hydrogen, radon, and the light hydrocarbons; methane, ethane, propane, iso-butane, and normal butane.

Although sample collection intervals were variable over this time period, systematic magnitude and compositional changes were recorded for the measured gases. Hydrocarbon magnitudes nearly doubled from December 1981 to July 1982 and remained fairly stable until sample collection termination on 12/1/82. These measured changes in gas flux, which could indicate a precursory signal related to potential earthquake activity on the locked southern section of the San Andreas fault, demonstrated the rapid flux changes occurring with tectonic activity along a major basement fault.

In addition to these natural seeps, gas flux associated with pressure changes in manmade underground storage reservoirs confirm the rapid variations noted from natural seepages. For example, gas concentrations measured under a ground sheet placed over a propane storage cavern demonstrates the effects on gas flux caused by barometric and meteorological variations.

The rapidity with which natural gas can migrate through the earth were also established by measuring the gas flux in 122 eighteen foot deep permanent sites over an extended time period from July 1981 to July 1982 directly over an underground coal gasification reactor at the North Knobs GR&DC DOE UCG facility near Rawlins, Wyoming. Baseline gas concentrations were established one month before the 600 foot deep retort was pressured and fired. Monitoring of leakage gases continued daily throughout the three months during the burn and for one month after in direct response to subsurface pressure changes within the reactor. Time of migration was found to vary from 2 to 15 days, depending on location with respect to the position of the retort at depth. A final monitoring was conducted six months after all operations ceased in July 1982 in order to follow the relaxation of surface leakage.

Results indicate that an UCG reactor provides an outstanding vehicle for migration flux measurements because of the uniqueness of the gases generated within the reactor. For example, the dominant gases are methane, hydrogen, carbon dioxide, and carbon monoxide, with only traces of other hydrocarbon gases. Pressure and compositional changes within the reservoir (reactor) occur at known times in direct response to operational procedures and can be used to track changes in the migration flux rates. In addition, migration of individual gas pulses were observed to exhibit chromatographic effects as the individual gases migrated away from the source at depth. These chromatographic changes existed for only a few hours at the onset of a pressure pulse in the subsurface reactor and were observed to quickly return to the steady state conditions in which the composition of the reactor gases match those escaping at the surface.

Introduction

During early development of soil gas prospects by the Gulf Oil Company, it was clear that magnitude variation of soil gas and dissolved gas sniffer data required a very complex explanation that appeared to be strongly affected by fault and fracture systems. Migration of oil and gas from the subsurface follows a complex pathway which can only rarely be imaged, as illustrated by this seismic section. Gulf scientists recognized that deep gas migration would be likely to form shallow micro accumulations along the pathway to the shallow subsurface. These "bright spots", which are the source of the sniffer anomaly in Figure 1, must then be projected to depth to find the field associated with the surface anomaly. This example was often used by Gulf managers Bob Brodine and Ed Driver to illustrate the expected relationship of sniffer anomalies to their subsurface sources (Weismann, 1980). This was an actual Gulf discovery in the East Canyon area with the accumulation located downdip to the southwest; note the "bright spots" marching down the fault.

Dick Mousseau and Bill Glezen of Gulf Research, evaluated the Gulf of Mexico sedimentation rates and compared these to calculated diffusion rates for the migrating gases. Their conclusion was that neither macro nor microseeps could occur via diffusion since the sediments were being laid down faster than the gases could diffuse upwards. The presence of both macro and micro seepages in the Gulf of Mexico obviously meant that diffusion was not the migration mechanism. Perhaps this striking conclusion would have stopped Gulf's program; however the comparison of seep compositional information with known production proved these microseeps were real, since they could be correlated to their respective source rocks at depth by comparing the compositional information exhibited by the near-surface gases directly to their underlying sources.

Some typical reservoir gas analysis data which was instrumental in this comparison was published by Nikonov (1971), who compiled gas data from 3,500 different reservoired deposits in the United States, Europe and USSR, and grouped them into very useful subpopulations, as shown in Figure 2. As shown, gases from basins containing only dry gas (designated NG) contain less than 5% heavy homologs, whereas gases dissolved in oil pools (designated P) contain an average of 12.5 to 15% heavy homologs. The heavy homologs plotted by Nikonov include ethane, propane, butane, and pentane.

Four near-surface soil gas data sets collected by Gulf Research are shown in Figure 3, (Jones and Drozd, 1983; Drozd et al., 1981). This data, collected over the dry-gas Sacramento basin, the gas-condensate deposits in the Alberta foothills, and two oil fields (Abo Reef and Sprayberry) in the Permian basin showed that the concept put forth by Nikonov could be used to establish similar relationships between surface seepage data and their respective sources in these areas.

Given this compositional encouragement, continued data gathering and generation of additional examples further demonstrated the strong tectonic control, as shown by examples from the Pineview (Figure 4) and Ryckman Creek fields (Figure 5), located in the Utah-Wyoming Overthrust Belt (Jones and Drozd, 1983). In these two cases, many of the major faults could be mapped at the surface, permitting verification of the direct physical relationship between the faults and the magnitudes of the anomalies. Additional examples from East Texas (Figure 6) and the offshore Green Canyon area in the Gulf of Mexico (Figure 7), further demonstrated the deflection and control of seepage pathways by growth faults (Pirkle, 1985).

These examples helped in the reevaluation of our basic concepts, and proved that the conclusions reached regarding the association of macroseeps with production must also apply to microseeps (Link, 1952). As shown by Figure 8, the Infantas Field in Colombia was observed to have hissing gas seeps when first discovered. As the field was produced, these seeps disappeared. During water flood operations, the hissing seeps returned upon repressuring. In the 1980's Gulf research called upon other noted scientists, such as Dr. John Hunt from Woods Hole, to help in understanding this complex relationship. As presented in a 1981 AAPG invited paper (Figure 9), Dr. Hunt pointed out that over 70% of the reserves in the world are associated with visible macroseepages.

Gulf scientists also continued to translate and read a considerable volume of Russian literature. This not only confirmed our concepts and results, but produced additional insights to proper understanding of this technology, as shown by Figures 10 and 11 which demonstrate the relationship of seepage magnitudes on tectonic activity and subsurface pressure. This comparisons between USA and USSR literature led back to Dr. Link's classic paper, where he listed the mechanisms by which macroseeps migrate to the surface, (Figure 12). As demonstrated by the previous examples, diffusion is only one of six migration mechanisms, and obviously cannot provide the main process for migration of hydrocarbons to the surface.

Migration Model

In 1976, Dr. Martin Matthews and I were working in West Texas over the Puckett Field when we noted that the larger magnitude CO2 seeps occurred with a higher frequency directly over the deep seated fault zones which lie along the west flank of the Puckett field. These deep faults do not come to the surface, but are covered with over 14,000 feet of supposedly unfaulted sediments. Our migration model, proposed at this time, is shown in Figure 13, as illustrated by a data set collected over the Patrick Draw field as part of the joint industry GEOSAT program (Richers, et al., 1984). Numerous studies have continued to suggest the validity of this concept.

An outstanding visual example of this model was demonstrated by macroseeps in Oklahoma when a shallow gas well was overpressured in 1980 (Preston, 1980). Figures 14 and 15 show gas bubbles emerging from one of the larger mud craters and the surface expansion of the gas vents as expressed in the Cimarron River deposits. Note the fracture orientations shown in Figure 15. The Cimarron River deposits were reported to be dominated by brittle shales which were broken by vertical fractures that generally occur in orthogonal patterns. The dominant fracture set, which was intersected by a northeasterly secondary set, appeared to be oriented northwesterly. To the extent to which they were open, these fractures obviously provided migration paths for fluids. Many of the vents and bubble trains were clearly aligned in directions corresponding to the fracture orientations - a phenomenon even reported to be manifested in the unconsolidated river deposits. Attempts to plug the surface vents would be expected to merely divert the migration path to the surface to an alternate fracture system. In fact, it was reported that many of the initial vents were apparently abandoned by natural plugging, and new vents formed during the later states of activity.

Leakage from Underground Storage Reservoirs

The next examples could be captioned "preparation meets opportunity", where soil gas exploration techniques were applied over a variety of underground storage reservoirs and an underground coal gasification reactor.

Leakage from an Underground Salt Dome

In the first case, as shown in Figure 16, a leaky well in a salt dome lost its product into the cap rock at about 600 feet deep. The product migrated out into an adjacent town requiring the evacuation of people and creating both political and technical difficulties. Soil gas geochemistry was successfully used to evaluate the leakage situation and help locate the lost product, which could no longer be recovered from the storage well since it had migrated up through the cap rock and into the sands that overlie the salt dome.

A surface hydrocarbon survey was conducted using geochemical test holes drilled 30 feet deep to encounter the shallowest groundwater aquifer. Over 500 sample stations located on 50 to 100 foot centers were installed throughout the town and over the general storage area. The distribution of lost product is shown, along with the location of the leakage well, on Figure 17. As shown, the lost product migrated updip toward the top of the dome and 3000 feet laterally. Major lateral leakage appeared to follow a NW-SE sand channel and/or fault related feature. The surface topographic contours clearly show the top of the salt dome. The initial geochemical sampling was extended well beyond the product leakage boundaries in order to properly define the location of the leakage product.

The geochemical sampling stations were installed using PVC pipe with water well screens on the bottom, and valves on the top, allowing them to be used for both leakage gas detection, product recovery, and nitrogen injection. These stations thus served several purposes. The leakage product apparently migrated into a sand layer at about 200 feet in depth, and from there, vertically pressured the shallow sands located only 30 feet below the surface. The first two relief wells were drilled to 100 foot depths along a NW-SE lineament and flared in order to release the subsurface pressure. The first well blew out and produced more than 23 MMCFGPD. More than thirty relief wells were drilled over the anomalous area shown in Figure 17, in order to relieve the subsurface problem.

Another point of interest was the fact this was a special product, an ethane-propane mix which has a unique signature as compared to the normal hydrocarbon products originally found in the natural reservoirs associated with this salt dome. The hook-shaped seep on the north end of Figure 17 was found to consist mainly of propylene. This was traced to a earlier reported spill, as documented from the historical records, which was previously thought to have no known surface expression.

The cleanup operation was facilitated by turning the 30 foot deep geochemical monitor wells into eductor sites by installing a venturi tube on the top of the well casing. Nitrogen was run through the venturi tube to produce a small vacuum on the hole. One could then inject nitrogen into the ground in advance of the leakage area and allow that nitrogen to migrate through the ground toward the eductor sites where it escaped to the atmosphere. To aid in following this N2 flood process, Gulf converted helium/hydrogen chromatographs, typically used for exploration seep detection, to nitrogen/oxygen detectors. Typical response curves illustrating the advance of the nitrogen front and cleanup of the E/P product mix are shown in Figure 18. This figure illustrates the rapidity with which the nitrogen front passed through the 30 foot aquifer in this area. The nitrogen flood pushed the lost product laterally back onto Gulf's property through the shallow 30 foot sand after the pressure was relieved in the deep 200 foot sand, thus cleaning up and preventing any recharge to the local residences.

The time rate of cleanup response over the entire area turned out to be quite interesting for illustrating the variation in lithology and permeability. It was not uniform, and clearly demonstrated a significant influence on lateral gas migration. In some cases, the permeability was so low that 15 lbs of nitrogen pressure at 30 feet was not sufficient to push the gas 50 feet laterally over to the next soil gas station; yet, if a bucket of water were poured on the ground near the injection site, the ground would froth and bubble from the nitrogen that was escaping vertically through 30 feet of Beaumont clay. This was in spite of the fact that the nitrogen would not travel laterally to the next eductor site. It appeared that most of these recalescent hot spots had a better association with deeper vertical migration pathways rather than from lateral migration through the 30 foot sand.

To summarize, a soil gas survey was used to outline the leakage area, deep wells were drilled to relieve the product at depth, and a nitrogen flood was conducted for final surface clean up. In addition, the compositional data indicated whether the soil gas was derived from the product well, or from a previous spill or pipeline. For example, during these remediation activities, several natural gas pipeline leaks were detected within the affected area which had nothing to do with the storage product leakage. This latter application particularly points out the value of using the gas chromatograph. Often, consultants who are called in on these type of environmental jobs use small thermal conductivity detectors which are not selective to the type of gases detected. The soil gas is aspirated through the instrument and detected by a deflection on the meter. Because of the very poor sensitivity and selectivity of such instruments, a deflection of two divisions was observed to indicate a problem area in this example. However, some predetermination was required to determine what the meter was responding to when small deflections were noted. For example, nitrogen used for the flood caused the same positive deflection on the meter as low levels of the leakage product. Many of the areas that were originally depicted as "hot" based on the LEL meter had no product in them whatsoever, but were pure nitrogen. If we had not used the gas chromatograph to measure specific components, the cleanup operation would have lasted a lot longer and would have included attempts to clean up the injected nitrogen as well.

Following the cleanup operations, an advance warning system was installed to catch leakage from other storage wells before they could become a problem. Two permanent monitoring stations were installed at every storage well casing. This type of monitoring system, if properly sampled on a regular basis, will allow early detection of well leakage before it appears at the adjacent boundary of the property. Application of such a system can be very beneficial for preventive maintenance and is highly recommended.

Leakage from a Mined Propane Storage Cavern

Another excellent example is taken from a study conducted over a 200 foot deep propane storage cavern. The immediate objective was to determine the leakage rates and to test whether or not ongoing remedial efforts to repair the leaks were successful. This particular case also provided an opportunity to determine the product leakage distribution and to conduct pressure pulse tests by injecting helium into the cavern as a tracer. Although this was a fairly small facility, 455 geochemical measuring stations were installed to a depth of 20 feet using PVC pipe placed on 10 foot centers. Figure 19 shows the propane collected over the cavern plotted on a log scale. An interesting secondary observation observed on this survey was an obvious color change noted on the soil cores. The soil was observed to change from a red-brown to a green-black directly over the top of the cavern where the largest seepage anomalies occurred. These chemical changes appeared to be related to hydrocarbon seepage and were used as additional confirmational information to provide further evidence of where the gas leakage has occurred over long time above this cavern.

Barometric Relationship to Gas Flux

An underground storage cavern is also a good place to observe gas flux related to atmospheric phenomena. Plastic ground sheets about 5 feet square were installed directly over known leakage areas to measured the gas flux related to meteorological and barometric changes and in order to answer the question - what are the effects of rainfall and barometric pressure on gas seepage flux? Figure 20 shows the variation with rainfall as vertical bars. A very large seepage anomaly is shown by the dashed line at the right edge of the first bar. The rain probably displaced the gas in the ground and caused it to come up underneath the ground sheet. Since this large gas flux did not occur again after the next rainfall events, it was assumed this initial gas flux was caused by buildup of longer term gas leakage that was trapped directly under the sheet and forced out by the first influx of rainwater.

Continued barometric monitoring, as shown by the dark solid line at the top of Figure 20, produced several very small barometric changes around the 19th, 20th, 21st, and 22nd days of the month. Note the small barometric lows that are expressed in the dark curve line. Right underneath, in the light dotted curve line, are a couple of gas flux observations under ground sheets which have a direct relationship. Every time the barometer took a little dip, some gas flux popped up under the ground sheet, thus confirming that gas flux does escape into the atmosphere.

Migration Timing Related to Pressure Pulse

Several opportunities to learn from this experience were encountered during this remedial period. The cavern pressure was decreased to ambient levels for repairs and then repressured, allowing the recharge leakage rate from the cavern to be measured. Upon recharging the cavern, the time it took for the propane gas to reestablish its maximum leakage values at the surface was measured. Following the recharge of the cavern with propane to its original pressure, a value of over 90% of the original soil gas propane concentration was observed within the observation test holes. This occurred within 15 days. As shown by Figure 21, most of the leakage came from around the central shaft. However, there is a large propane background in the soil since the storage site had been known to leak for over 20 years. This propane background made it difficult to be sure the product reappearing at the surface came directly out of the reservoir during the sampling time period.

Injection of Helium Tracer into Cavern

As a second more definitive test, helium was injected into the cavern at a concentration of about 600 ppm. Results showed that in 15 days, not only had the product moved, but as shown in Figure 22, helium was also detected at the surface. Helium injection not only showed the leakage around the central cavern, but also found a leak at the end of one of the drifts that would have been missed looking only at propane. The amount of helium used for this test was not enough to damage the product for sale, and yet still gave more than adequate sample for analysis.

This data suggested that migration was quite rapid. In other similar cases, for example from an underground coal gasification reactor, we have been able to establish migration times of from 2 to 15 days at depths of up to 600 to 1000 feet for changes of gas concentrations in the reservoir or cavern to be expressed at the surface. This means migration does not follow a diffusion model, but is driven by pressure, with migration driven along fault and fracture patterns and joints.

Leakage from an UGC Reactor

Another very educational example was taken from an underground coal gasification reactor near the Rock Springs Uplift in the Rawlins, Wyoming area (Jones and Thune, 1982; Jones, 1983). The North Knobs UCG facility is located approximately eight miles west of Rawlins, Wyoming, in south-central Wyoming (Figure 23). It is situated on the southwest flank of the asymmetrical Rawlins uplift adjacent to the Washakie Basin. Throughout the area, the well exposed resistant sandstones all exhibit a remarkably consistent and well developed near-rectilinear joint pattern. The dominant (systematic) joint set strikes about 6o (N14W) from the strike of the beds (N20oW) and is well exposed throughout the area. A well developed, but subordinate, nonsystematic cross-joint set having a strike of approximately N49oW is also present in all exposures. Figure 24, modified from McCurdy (1979), schematically illustrates the rectilinear joint system which is believed to provide a mechanism for migration. Figure 25 shows the configuration of the nearly vertical beds, which dip too steeply to do anything except gasify the coal in-place. The gasification reactor lies about 600 feet vertically below the surface. Pictures looking north along strike and west downdip are shown in Figures 26 and 27, respectively. As shown in Figure 28, one hundred and twenty-two, 18 foot deep permanent sites were installed over the general area of the retort.

The objectives for this survey were to determine if gases generated during the burn leaked into the near-surface, and if so, their rates of leakage, migration paths and the composition of those gases, in order to evaluate the economic importance of such leakage and to assess their hazard potential from the standpoint of human safety and the environment.

All geochemical sites were drilled with a 3-inch diameter auger to a nominal depth of 18 feet and established as "permanent" observation sites. This was accomplished by installing a 20 foot length of 1 inch ID PVC pipe, perforated with about 30 one-quarter inch diameter holes in the lower 1-1 1/2 feet of the pipe. During installation, sufficient pea gravel was installed to provide a permeable zone for collection of soil gases leaking from the adjacent formations.

The locations of the sample sites and near-surface geological features are shown in Figure 28. Plots of selected methane magnitudes with time are shown in Figures 29-30 for sites (27, 22), (1, 2), (5, 6), (13, 20), and (17, 19), respectively. These sites were selected to represent the typical changes noted in time response of gas migration. Note sites 22 and 27 exhibit a very quick response to the retort pressuring which is almost identical even though they are about 50 feet apart. Site pairs 1 and 2, 5 and 6, 13 and 20, and 17 and 19 also show similar response to their close pairs which are several hundred feet apart. The leakage patterns which emerge over time are clearly not random, but are systematically changing in relation to subsurface controls.

During field operations it was observed that it took about 3 to 5 days after the beginning of the system air-pressure test to ignition of the coal, before any significant increases in the magnitude of the hydrocarbon gases were recognized in the near-surface. Propane leakage was related to migration from the retort of the initial products in ignition because propane was used to achieve ignition, and was not a major retort gas generated during the burn. Figure 30, illustrates this since site 22 was directly updip from the retort and clearly showed a sharp rise in propane upon pressuring and a fairly rapid decrease during the initial phases of the burn. The maximum pressure in the retort at 600 feet was 700 psi. This excess pressure, used to link the slant and vertical wells, caused a rapid change in the surface signature that occurred within two days, as shown by Figure 29.

An examination of the change in methane flux from various sites suggested the soil gas vapor data could initially be divided into at least four discrete periods for mapping, as shown in Figure 31. These selected time periods are shown by color separations separated by vertical lines in Figures 29 and 30. They are: (1) pre-pressure--July 22, 1981, through August 16, 1981; (2) pressure--August 17, 1981, through August 23, 1981; (3) burn--August 24, 1981, through November 10, 1981; (4) post-burn--November 11, 1981, through December 12, 1981 (end of field survey).

Evaluation of these data allowed the first estimates of how long gas remains in the surface sediments, and on what magnitude of fluxes might occur under various pressure conditions. In monitoring the free gases, several permeability tests were conducted in which the gases within the sampling test holes were pumped down to ambient levels to see how quickly they would recharge. The maximum gas values were observed at sites 22 and 27, which are directly updip of the outcrop of the coals being gasified. This maximum value occurred in the sandstone directly over the coal with 100,000 ppm methane showing up at the surface. Figure 29 also illustrates the location and change in shape of the methane magnitudes observed at the surface at four selected pairs of sites over the time windows selected for mapping. In contrast to sites 22 and 27, sites 85 and 86 had about 1200 ppm natural levels of hydrocarbon that had nothing to do with the active retort. They appeared to represent a natural seep which existed previously in the area. The residual gases were pumped out of these two sites and apparently did not further recharge during the retort process. The seep located to the northeast at sites 1 and 2, right above, and along the baseline, provided another anomaly that was not influenced by the 1981 gasification process. These two sites were charged by a previous (two years earlier) UGC reactor that lies directly downdip from these sites. Both the vertical and slant hole product wells for both the 1979 and 1981 retorts are plotted on Figure 28 for reference.

By selectively averaging the gas flux measurements from within each of the four time windows for each site, one can construct a series of contour maps which illustrate the most significant changes that occurred over the four time windows previously outlined. As shown in more detail in Figures 32-35, the leakage patterns changed with time in direct response to pressure variations in the subsurface retort (Jones and Thune, 1982; Jones, 1983).

Injection of Helium Tracer into UGC Reactor

Helium was occasionally injected during the stable part of the burn in order to estimate the transit time for gases passing through the reactor. This helium appeared in the surface seep gases at sites 22 and 27 within two days of its injection. It disappeared just as quickly.

"Hydrocarbon Spots"

As shown in Figures 31-35, three major methane anomalies were observed along the strike of the bedding plane. This indicated the leakage gases did not just migrate updip along the bedding plane, and then migrate laterally along the strike of the beds to fill the surface sediments with gas. Instead, the leakage gases came up almost simultaneously within three localized areas, or "hydrocarbon spots". This concept suggests a migration pattern with gases moving along a complex mixture of bedding planes and fracture avenues at depth. The location of the "hydrocarbon spots" at the surface are then controlled by these complex pathways, and as such are somewhat predictable from geologic mapping. Once these "hydrocarbon spots" are determined at a site, they will provide the pathways for all future pressure relief from depth and can be used to establish a permanent monitoring system. In this extreme case, the vertical migration zones were obvious, during both the charging and discharging periods, as the surface gases were depleted.

Gulf Research suggested the Department of Energy maintain this site for future research on soil gas analysis, since one could conduct various depth probe measurements to check the influence of joints and soil types on both vertical and lateral gas migration, and further investigate dissipation of the retort leakage gases with time. This site would also have been particularly useful for a research study because of the uniqueness of the gases generated; for example, carbon dioxide, carbon monoxide, hydrogen, and methane are the dominant gases. Because these gases were unique to the coal gasification process, it was known where and when they were generated. The maximum concentration of gas in the bedding plane leak was about 50,000 (5%) to 100,000 ppm (10%) at the peak generation of the retort, and fall off as the reactor pressure was reduced and finally filled with water at the conclusion of operations. The concentrations at sites 22 and 27 decreased until they were below 10,000 ppm a year later (Jones, 1983). This example provided adequate flux information for modeling, providing an excellent resource for further study.

Deep Mobile Gases/Relation to Earthquakes

Examinations of the published literature, coupled with extensive field applications by Gulf Research scientists have indicated that large volumes of diverse gases continually escape from the earth's crust into the atmosphere. Areas of especially high activity appear to be related to zones of deep tectonic fracturing and the accompanying jointing in which mineralization is sometimes located. Typical deep gases are CO2, N2, CH4, H2, He, Ar, Rn, Hg, SO2, COS and H2S. The major components are CO2, N2, CH4, and H2; with the remainder of this list generally found as minor or trace components. The isotope ratios of hydrogen, carbon, oxygen and uranium have considerable potential for helping to define the sources of these gases. According to published literature, the magnitudes of deep gas anomalies are governed strongly by tectonic and magmatic activity, thus stronger patterns are encountered in seismically active areas of late orogenic activity. Accordingly, the weaker patterns are observed in platform and shield areas that are relatively quiescent; that is in consolidated blocks of the earth's crust.

That earthquakes may possibly be preceded or accompanied by deep mobile gases appear to have first been observed in the USSR in 1966. In a study of the Tashkent earthquake zone, Fursov et al. (1968), found air aspirated from boreholes over faults contained as much as 15 times more mercury than air not located over fault zones. This work points out that faults can be the channel ways through which mercury vapor migrates, but also indicates tectonic activity can release mercury not necessarily related to economic mineral deposits. Fursov's classic profile is shown in Figure 36. Studies of this type were also undertaken in China at about the same time and in Japan in 1973.

In other areas, the Soviets have shown the magnitude of soil gas values on faults increase dramatically shortly after an earthquake in which fault movement was involved (Zorkin, et al., 1977). An extensive study, involving 105 observation wells 3 to 5 meters deep, was set up over the Mukhto Oil Field in northeastern Sakhalin Island. Thirty-seven hundred samples were collected and analyzed over a four month period with the most active wells sampled daily. Figures 37, 38 and 39 from this study provide impressive evidence for the tectonic relationship of this leakage gas flux. This study leave no doubt that faults and fractures provide the main control on the effusion of gases from the subsurface.

Particularly intriguing examples (see Figures 40-41) have been published by Antropov, et al. (1981) of atmospheric methane flux related to petroleum deposits, and seismic shock. These measurements were made with adsorption type gas lasers. Two types have been described: one measures the sample in an adsorption tube (Iskatel-2), while the other (Luch) measures the specific gas adsorption along an optical path. One makes point measurements while the other averages the adsorption over a long path length (1-100 meters).

Chemical monitoring of earthquake activity was not widely practiced in the US where most efforts were geophysical until about 1975, when limited studies were initiated using radon. Gulf Research also conducted their first measurements of helium and hydrogen in 1975.

"Helium Spots"

Two recent landmark publications in Science by Wakita et al. (1978) have indicated that both helium and hydrogen are observed in anomalous quantities along faults. In 1978 helium was observed to be as large as 350 ppm within a nitrogen vent on the Matsushiro fault swarm. The authors proposed to call these anomalous areas "Helium Spots" because the helium leakage was not homogeneous throughout the fault zone. These unevenly distributed helium spots were reported to occupy areas of about 30 x 50 meters. Extensive experience in soil gas prospecting by Gulf indicates that all soil gas anomalies occur in such spots. This is obviously because the migration of gases are dominated by faults and fractures, either on the macro or micro scale.

A second paper, published in October 1980 in Science, reported 70 measurements for hydrogen in the Yamasaki fault zone (Wakita, 1980). These measurements, made in 0.5 to 1 meter deep holes, reported hydrogen anomalies ranging from 2 to 30,000 ppm H2 in the fault zone, with ambient background values of 0.5 ppm observed outside the influence of the fault. The authors postulated that hydrogen was formed by the reaction between groundwater and fresh rock surfaces formed by fault movement.

Limited programs using radon began about 1975 at about the same time Gulf Research and Development Company first made measurements on light hydrocarbons, helium, and hydrogen on the San Andreas Fault in the Cholame Valley in California (Jones and Drozd, 1983). An early published example, shown in Figure 42, confirmed helium is a deep basement, or tectonic indicator which is commonly independent of oil and gas deposits. At Cholame, where this profile is located, the fault moved in 1857, 1906, 1922 and most recently in 1966. Figure 43 shows two pictures taken immediately after the 1966 earthquake. The first on July 22nd when the offset was only two inches, and then again 12 days later on August 4th when the highway offset had increased to five inches (Iacopi, 1976). It was reported to Gulf scientists at the time these measurements were made in 1975 that nearby doors and gates at the Hurst Ranch changed their overnight fit on a daily basis, indicating the San Andreas Fault remained active at the time of the survey.

Numerous published examples of gas flux related to earthquakes have been reported, Barsukov (1979), Borodzich (1979), Elinson, et al. (1971), Eremeev (1972), Fursov (1968), Pirkle and Jones (1983), Kartsev (1959), King (1980), Mamyrin (1979), Melvin (1978, 1981), Mooney (1982), Ovchinnikov (1972), Reimer (1980), Shapiro, et al. (1981, 1982), Sokolov (1971), Wakita, et al. (1978, 1980) and Zorkin (1977), to name a few.

Helium-Hydrogen Associations/Duchesne Fracture Zone

Leakage from less active faults are demonstrated by an example taken from the Duchesne fault zone in the Uintah Basin (Pirkle and Jones, 1983). The extreme length and linearity of the Duchesne fracture zone, which can be traced for approximately 42 miles, suggests it extends to great depth, perhaps even to basement. The surface geometry also indicates the zone is essentially vertical, and this is certainly true where the zone can be viewed through some 800' of stratigraphic section in the east wall of Lake Canyon (Sec. 17, T4S, R6W). At its widest point the zone is about two miles across. The Duchesne fracture zone consists of a main line of normal faults which mark the south edge of the fracture zone proper. This zone of maximum deformation has the general aspect of a slightly asymmetric graben in that, although individual fault blocks within the zone may be up or down relative to juxtaposed blocks, the general sense of movement has been down relative to the terrain to the north and south. In places, minor normal faults are found north of the main zone for another quarter mile or so. Locally the individual surface faults may be offset slightly en-echelon. The faults are all high angle and of small displacement, maximum surface displacement of the bounding faults of the graben probably being not more than 50-75 feet. Faults, which are secondary to the main faults, appear to follow a regional trend of jointing. To the extent there has been continued activity along the Duchesne zone through the Tertiary and probably also Cretaceous, it is possible the graben at the surface reflects a single fault at depth. Thus, the en-echelon spatial relations of the individual faults comprising the Duchesne fault zone may extend to depth forming a complex structural system.

Geochemical traverses which run across the Duchesne fault are shown in Figure 44. The largest soil-gas anomalies occur where a geochemical site is located on or immediately adjacent to one of the faults of the zone. The faults may have similar en-echelon spatial relations in section as well as in plain view so that, as conduits for gas leakage, they can be expected to have abrupt discontinuities both areally and at depth.

Figure 45 shows a profile run during July 1976 which extended along strike and follows the fault zone for a considerable distance. Two major helium anomalies of 681 and 470 ppm were noted. Hydrogen seemed to be greater than 20 ppm over most of the zone, with numerous values over 150 ppm. The maximum hydrogen values range upwards of 3 to 5 thousand ppm.

A second more detailed traverse was conducted perpendicular to strike in December 1976. This more detailed traverse, consists of 10 closely spaced sites which cross the northernmost fault of the Duchesne zone that has a surface expression near the center of the strike line. As shown by Figure 46, this traverse continues south to cross a second fault, which is down to the north, as expressed in the topography, and forms a narrow graben with the first fault.

Sites 10 and 11 are north of the surface expression of the fault but lie in small, shallow but topographically well-defined washes. As shown by the outstanding helium anomaly of 372 ppm on Figure 47, they could be near or on zones of fracturing associated with the fault zone. Site 3 is definitely located on the surface expression of a fault. Sites 4, 5 and 6 are within the graben. Station 5 is on the crest of a little knoll and is located close to bedrock, while Site 6 is in the bottom of a shallow wash and is probably very close to or on a main fault, as indicated by the topography and subsidiary faults on the bluff to the south. Sites 7 and 8 were placed so as to lie above the secondary normal faults of slight displacement. These faults can be seen in the bluff east of these stations (Figure 48). Of particular interest is the large helium anomaly of 227 ppm at Site 7, which certainly indicates the possibility of a deep fault at this station. It should be noted that all the sites between 2 and 7 were observed to have considerable helium (50-100 ppm) associated with them.

Earthquake Predictions Based on Gas Flux

After having studied both microseeps and manmade macroseeps, it was apparent the next step was to acquire gas flux data from an active fault. As shown by Figure 49, the level of seismic activity associated with the San Andreas Fault in California in 1981 makes this geologic province an obvious choice. In order to accomplish this objective, Gulf Research approved a corporate level research project entitled "Gas Flux Related to Earth Motions". Evaluation of the known earthquake prediction programs at the USGS and various universities revealed that the Kellogg Radiation Laboratory at Caltech in Pasadena had the only computerized system which could provide automatic data collection of a series of geochemical variables.

Caltech scientists, under the direction of Drs. Tom Tombrello, Mark Shapiro, and John Melvin, had established a network of automated Radon-Thoron monitors operated by a microcomputer so data could be collected almost continuously (every eight hours), stored onsite in the computer memory and then transmitted back to the central laboratory at Caltech over regular telephone lines on command from a remote computer (Shapiro et al., 1981). Since Gulf's research objective was to map short-term flux changes, this computer link was an essential requirement. As shown in Figure 50, Caltech's operating stations were located at Fort Tejon, Lake Hughs, Pasadena, Santa Anita, Stone Canyon Reservoir, Big Dalton Canyon north of Glendora, Lyle Creek, and Sky Forest in the San Bernardino Mountains, and at Pacoima Dam, where Gulf paid the drilling costs in order to be included in the Caltech program.

Pacoima Dam was chosen for a site by Caltech scientists because they anticipated microseismic activity might be generated by mass loading and unloading within this very steep and fractured valley. Initial measurements made by Gulf at Pacoima Dam indicated that hydrocarbons were very low, so Gulf choose to set up this station with a helium-hydrogen gas chromatograph. A view of the dam and its fractured rock walls is shown in Figure 51 and a close-up view, which includes the geochemical hut, is shown in Figure 52. Shortly after the gas chromatograph was functioning, a strong hydrogen peak of 75 ppm was measured in April of 1981, just before a 5.6 magnitude earthquake hit Westmoreland, California. This H2 anomaly lasted nearly three weeks and peaked sharply at about 75 ppm, as shown in Figure 53. Whether a coincidence or not, this classic response provided considerable encouragement to the joint Gulf/Caltech program.

Previous experience in using carbon dioxide to map soil gas anomalies encouraged Gulf to introduce instruments for measuring carbon dioxide at several of the established Caltech stations, such as that shown at Lake Hughs in Figure 54. Within a short time, results from the Lake Hughes station showed the presence of correlated radon and carbon dioxide anomalies. It appeared carbon dioxide reached saturation levels in the water and then served as a carrier for the radon (Shapiro et al., 1982). Although not an earth shaking result (no pun intended), each improvement in measuring and relating the natural gases emanating from these stations increased the possibility of producing interpretable data.

Arrowhead Hot Springs

Earth gas monitoring studies for possible fluid phase precursors to earthquakes by Scripps Institute of Oceanography also began in 1975 with sampling at Arrowhead Hot Springs (Figure 55). Grab samples of spring gases were collected at one month intervals at the concreted hot spring and analyzed for dissolved radon, helium, nitrogen, temperature, and conductivity. Beginning in 1977, methane was also measured in each sample. Results of these compiled data reflected a variety of short term variations in measured gas content for comparison with seismic events along the San Andreas fault in southern California. The most significant correlation identified was a large increase in measured gases (radon, helium, nitrogen, and methane) in 1979 before the Big Bear earthquake with a magnitude of 4.8 (Figure 56) (Craig et al., 1980). This significant increase has been interpreted as the result of an increase in the deep gas component dissolving into hot springs waters. The success of the Scripps' grab sampling program suggested this location would provide even more valuable data for earthquake prediction studies with onsite computer controlled continuous monitoring of gases.

Preliminary gas monitoring by the Gulf/Caltech team at Arrowhead Hot Springs began in early 1981 by collecting gas bubbles with a funnel and gas cylinder. Samples were analyzed for methane, ethane, propane, iso-butane, normal butane, ethylene, propylene, helium and hydrogen by gas chromatography. The initial results, shown in Figure 57, indicated the hot springs gases contained 3,918 ppm of methane, 17.5 ppm of ethane and 1780 ppm of helium. The overall high magnitude of measured gases observed upon continued sampling suggested the location was ideal for continuous gas emission monitoring and inclusion into the Caltech/Gulf Research earth gas research programs. A variety of sample collection and analysis methods were employed during sampling at Arrowhead Hot Springs, including a cross check analysis by Dr. John Welham of Scripps as shown in Figure 58. Based on these early results, it was clear Arrowhead was the type of active seepage site Gulf scientists wanted to sample and to direct the Caltech earthquake prediction program toward.

Geology

Arrowhead Hot Springs is located at the base of the San Bernardino Mountains in the Transverse Range Province of Southern California (Hadley and Kanamori, 1977; Miller, 1979). The Transverse Ranges are a unique east-west structural and geomorphic belt which crosses the San Andreas fault. The ranges are bound by the Coast Ranges to the north, Peninsular Ranges to the south and the Mojave Desert to the east and northeast (Figure 59). Despite apparent strike-slip movement along the San Andreas fault, the Transverse Ranges seem to be continuous across the bend of the fault in Southern California. The San Bernardino Mountains are located northeast of and bounded by the San Andreas fault zone and are the last range east of the San Andreas fault.

The hot springs are actually located in two canyons in the foothills of the San Bernardino Mountains on a splay of the San Andreas fault. The splay seems to be related to the bifurcation of the fault into the northern and southern segments which continue southeast toward the Salton Trough (Figure 60). It is apparent the spreading and subsistence in the trough is moving northward along the San Andreas fault. Basement rocks of the San Bernardino Mountains, which were uplifted in Late Quaternary time, can be divided into two structural blocks separated by the north branch of the San Andreas fault.

Outcrops in the vicinity of the field area range in age from Precambrian granite, to Paleozoic meta-sedimentary limestones and schists and Mesozoic intrusive rocks. This suite of formations reflects the complex history of the San Bernardino structural block. The Precambrian formations seem to be related to similar units in both the San Gabriel Mountains and the Mojave Desert region.

The San Andreas fault cuts through the base of the San Bernardino Mountains as two distinct branches which continue to the southeast, and a third branch which is truncated in the vicinity of Arrowhead Hot Springs. The San Andreas, in this area shows, extensive vertical displacement which has been active through recent times, as can be seen from terraces in the alluvial formations in the canyons of interest. The fault splays, which cross the area, do not show obvious recent movement.

The faults crossing the field area were identified on the surface and from copies of low level areal photographs where they were not covered by recent alluvial sequences, but which are present in the vicinity of the two canyons. The faults are most easily identified by linear drainage and contacts of metamorphosed Paleozoic carbonates with Precambrian gneiss formations. Small amounts of vertical displacement of one to four feet can be seen in the outcrop which continue until covered by alluvium. The three mapped faults are only inferred for a short distance past the hot springs area although they may not be easily identified in the rough topography of this area.

The hot springs in the field area are divided into two distinct groups (Figure 61) and are located in two canyons about 1/2 mile apart on the Campus Crusade for Christ International property. The main group of springs to the east are located in and around Penyugal Canyon and can be divided into three groups by location as east, west and in the canyon. The area has been developed as a bathing spa and recreational area since the late 1800's, and most of the original hot spring locations have been altered from their original character by the construction of baths and collection pools. In the case of the springs to the west in Waterman Canyon, four caves were dug into the side of the alluvial fill covering fractured bedrock from which steam and hot water was flowing.

The caves are, at this time, bulldozed over to keep out trespassers and still show signs of warm ground and surface steam. The best descriptions of the area, before substantial development obliterated many of the surface geologic features associated with the springs, were as an assessment of the geothermal potential of hot spring areas adjacent to Southern Sierra Power Company transmission lines, United States Geologic Survey Water Supply Paper #338 (1915). The temperatures of the steam caves were described as being dependent on the location with respect to the fault which lies directly to the east. The temperature was observed to decrease regularly as the distance from the fault increased.

Field sample collection progressed in stages from simply filling an evacuated one liter sample cylinder through an inverted funnel, to collecting and analyzing the gases continuously. To aid this process, a Plexiglass sample collection bucket was installed above a stainless steel collection pan placed at the bottom of the spring, as shown in Figure 62. Samples collected from the sample collection bucket had a natural flow rate of 40 to 50 cc/minute, and provided an integrated sample over time.

Gas chromatographic analysis of gas samples were completed using a Gulf Research designed, dual gas chromatograph with two columns and two detectors. Light methane through butane hydrocarbons were analyzed using a three foot alumina column and Gulf designed FID detector. Helium-hydrogen was analyzed using an eleven foot mol sieve 5A column coupled to a thermal conductivity detector. Both columns were set-up with a timer controlled backflush that prevented hydrocarbons heavier than butane from entering the alumina column and any components heavier than helium from entering the hydrogen-helium mol sieve column. This feature greatly reduced contamination of the GC and significantly increased the sensitivity and reliability of the C1-C4/He-H2 analysis. Samples could be analyzed by either flow-through or hand-injecting methods.

Continuous Gas Monitoring

As a preview to continuous monitoring, an onsite laboratory trailer was parked at the concreted hot spring and run continuously during daytime hours. The continuous analyses, at six minute intervals, provided a large body of data available for interpreting short term fluctuations in the gas emission magnitude and compositions with time. Real time sample analysis was completed on March 16, 1982, April 22, 1982, and from October 19, 1982 through December 2, 1982. These numerous measurements, along with multiple analyses from single days, have been compiled into graphs for evaluation (Burtell, 1989).

In order to conduct these continuous monitoring tests Gulf/Caltech scientists had to design and build the first computer-controlled gas chromatograph system. Caltech scientist, Dr. John Melvin, designed the hardware and software to equip one of Gulf's dual C1-C4/He-H2 gas chromatographs for remote, computer controlled operation. The system was controlled by an LSI 1123+ Digital computer with TU-58 tape drives for remote non-volatile storage of data (Melvin et al., 1981).

Methane

As shown by the initial grab samples (tabular data, Figure 57) collected in one liter cylinders between 12/8/81 and 1/17/82 there was a steady rise in methane from 3918 to 4507 ppm during the first year. More continuous measurements, which overlap this data, continued to show a steady, but variable overall increase of methane content in the hot spring gases which ranged as high as 6000 ppm by 12/2/82. These data indicated gases migrating into the spring have a proportionately larger methane content throughout the sampling period.

In addition, many shorter term methane fluctuations were also identified in the concreted hot spring gases. The most significant of these changes occurred between 4/4/82 and 5/21/82 when methane content increased from 5132 ppm to 6031 ppm on 4/16/82 and subsequently decreased to a level of 5241 ppm. Sporadic changes followed four months of steady increases and were followed by a relative plateau of values ranging from 4959 to 5431 ppm from 5/21/82 to 7/18/82. On 7/27/82 methane content again increased to a magnitude of 5834 ppm and on 8/12/82 the highest value recorded by this study was obtained at 6193 ppm methane. Following 8/12/82, values began to decrease to 5765 ppm on 9/11/82 and continued to fluctuate from 5204 ppm to 5798 ppm through the remainder of the study which ended on 12/1/82.

Propane

Propane magnitude data from the concreted hot spring have also been plotted to examine propane content versus time and to compare these variations with those identified by the methane results (Figure 64). Although propane exhibits a much narrower range of values and varies from a low of only 2.697 ppm on 12/8/81 to a high of 4.828 ppm on 8/3/82, the propane data clearly mimicked the gross changes recorded by methane. The net propane increase with time was almost twice the initial magnitude, as was observed for methane.

Individual high magnitude events, identified in methane, correlate well over the entire time period, with propane magnitudes. However, on a sample to sample basis, propane does exhibit some definite independence from methane. This was apparent from 1/17/82 to 2/15/82 when propane increased from 3.135 on 1/17/82 to 4.124 on 1/22/82, but methane only changed from 4507 to 4722 ppm. From 1/31/82 to 2/25/82 propane magnitudes remained relatively stable from 3.822 to 3.726, but methane decreased significantly with values of only 4464 to 4265 for the same time period. The large increase in magnitude for methane on 4/16/82 was reflected in propane magnitudes although propane (+12.1%) did not increase proportionately with methane (+17.5%).

The magnitude event noted in methane data from 7/27/82 to 8/12/82 was paralleled by propane. Strong correlation continues through 8/27/82 when both methane and propane begin to decrease. At 9/11/82 propane decreased from 4.614 ppm to 4.200 ppm, a 9% decrease, whereas methane only decreased 3% from 5949 ppm to 5765 ppm. Since ethane, iso-butane and normal butane results show a very close relationship with each other and parallel propane results, they were not plotted.

Propane/Methane x 1000 Compositional Ratio

As a further means of evaluating light hydrocarbon seepage at the Concreted Hot Spring, the composition of the migrating gases were closely examined. Compositional indicators included ratios of one gas to another and percentages of individual hydrocarbon gases as compared to the entire hydrocarbon content of the migrated gas. The propane/methane X 1000 light hydrocarbon ratio is commonly used in soil gas geochemical exploration activities to identify whether a seep originates from an oil, oil and gas or natural gas type source in the subsurface (Jones and Drozd, 1983). This ratio, which has been plotted in Figure 65, exhibits values which range from 0.69 to 0.87, indicating a mature dry gas source for the measured hydrocarbons.

Although both methane and propane generally correlate with one another, their ratio clearly exhibited variations similar to those noted over the Mukhto Oil Field in the seismically active Sakhalin Island. These well established changes are probably a result of changes in the earth's stresses and its influence on gas emission. Ethane through butane light hydrocarbon gases also followed this trend and suggested that over this time period, a larger relative proportion of mature hydrocarbons were emitted from this spring.

Helium

Helium magnitudes, over this time period, appeared to fluctuate independently of the light hydrocarbon gases, suggesting an independent source for the helium (Figure 66). Overall helium magnitudes ranged between 1700 and 2350 ppm as free gases from the spring reflected a very concentrated source of helium (over 350 times atmospheric levels). Although the initial Scripps' data suggested a positive correlation between methane and helium for the 1979 Big Bear Earthquake, this more detailed analysis showed this correlation was much more complex and that helium may be affected by different geologic and tectonic events than methane.

Stable Carbon Isotopes

As noted earlier, the composition of the measured hydrocarbon gases reflected a dry gas signature. The presence of percent level to thousands of ppm's of heavier ethane through butane hydrocarbons suggests a deeply buried sedimentary source for the hydrocarbons. Stable carbon isotope measurements made on two gas samples collected on January 9, 1982 and January 17, 1982 had consistent results of -23.7%, confirming the presence of a very mature source.

The methane and higher hydrocarbon gases measured at the site suggest a migrated product derived by normal sedimentary processes, typical of very mature oil and gas accumulations. Small blocks of sediment could have been squeezed across the fault plane. A small sedimentary block could be thrust below the San Bernardino Mountains in the vicinity of Arrowhead Hot Springs, where the additional heat could produce the measured hydrocarbons.

Helium Isotopes

Isotopic measurements of He3/He4 for Arrowhead Hot Springs were found to have an average of 0.431 ñ 1.2%, from five measurements presented by Craig et al. (1980). This value is not nearly as thermally mature as measurements made over hot spot thermal reservoirs in Iceland, Yellowstone Park, and helium spots in the Matsushiro area of Japan. Although these measurements are in the range of 7 to 18 times atmospheric concentrations, their low ratios relative to other known thermal reservoirs suggests a dilution of normal granitic helium generated by radioactive decay within the crystalline rocks of the San Bernardino Mountains, mixed with limited mantle helium from depth. The presence of only a slight amount of mantle derived helium suggests that the thermal waters are a result of either frictional heat and resultant fractures, or possibly a small intrusive body.

Mercury Associations

In order to identify the size and shape of the hot springs system and its relationship to other geologic features, a low temperature soil mercury mapping program was also completed.

Five hundred twenty-five (525) soil samples were collected at 50 foot intervals from a grid consisting of 10 N-S and 4 E-W survey lines placed 250 feet apart and analyzed for adsorbed mercury content. Contoured mercury magnitude data (Figure 67) show two distinct and well controlled zones of anomalous values which clearly define the area of thermal springs and wells. Anomalous zones are centered around the surface spring outlets in each area and on the north edge of the study area in regions of Precambrian outcrop. Anomalous zones have very sharp boundaries and tend to be located within and north of mapped E-W trending fault zones which cross the property. Profiles of three survey lines are included as Figures 68, 69 and 70 in order to illustrate the strong contrast between anomalous and background areas.

The highest magnitude mercury anomalies are located in the Penyugal Canyon portion of the study area where several samples had concentrations in excess of 250 ppb. The distribution of mapped mercury values conclusively locate and confirm the distribution of thermal springs and wells on the Arrowhead Hot Springs property. Areas of anomalous mercury do not correlate with any particular mapped geologic unit, suggesting that the measured mercury migrates from depth and accumulates in the soils, rather than forming as a residuum of weathering of surficial rock units.

The overall distribution of mercury anomalies correlates well with the surface thermal spring outlets. This indicates that the subsurface mercury source is controlled more by thermal systems, and less by the east-west trending faults which splay from the San Andreas Fault.

High concentrations of mercury in the soil are coincident with surface thermal outlets, suggesting that mercury sampling can be used to identify potential thermal areas that are not otherwise evident. In the hot spring, mercury is more highly concentrated in the gas bubbles than in the spring water. Therefore, mercury may be enriched in soil as a result of vapor phase migration. The mapped mercury anomalies record the areal extent of the mercury rich vapor which probably migrated to the surface from a subsurface thermal reservoir.

Origin of Arrowhead Hot Springs Gases

Light hydrocarbons, helium and mercury have been monitored over time at Arrowhead Hot Springs and found to exhibit significant magnitude and compositional changes. The origin of the measured helium appears to be primarily of crustal radioactive decay, with minor input from mantle sources. Methane and other light hydrocarbons have a very mature sedimentary signature. This mature organic hydrocarbon gas strongly suggests that sedimentary rocks are present below crystalline and metamorphic rocks of the San Bernardino Mountains. The possibility is strongly supported by recent geologic research which indicates low angle thrust faults of Laramide to Tertiary age in the Mojave desert in the east. Tectonic activity may have thrust the San Bernardino Mountain block over sedimentary formations which now lie below the San Bernardino Mountains. Subsequent burial and maturation of sedimentary source rock may have produced significant quantities of hydrocarbons. If this possibility can be further substantiated by additional data, exploration for an untested petroleum province with commercial potential below the San Bernardino Mountain and other portions of the Transverse range province may be warranted.

This project began as a spin-off of a joint Gulf Research-Caltech deep fault migrated gas monitoring program to help Caltech develop vapor phase earthquake prediction techniques and to improve Gulf's understanding of gas migration mechanisms. The resultant investigations have tested various assumptions and theories about deep fault and fracture system gas emanations and their expression at near- surface sampling sites. Each phase of the program was completed to help develop geochemical sampling techniques, start a data base for future programs, and as an aid for the evaluation of potential earthquake prediction sites.

This program includes only the study of a single hot spring site in an attempt to relate gas emissions to earthquakes. Light hydrocarbons, helium, radon, hydrogen, carbon dioxide, and carbon monoxide were monitored for magnitude and compositional changes for correlation with earthquake occurrences. Although no clearly earthquake related events were identified, significant gas magnitude and compositional changes were recorded. Regional stress variations were interpreted as the most probable cause for the recorded changes in gas flux at this site. Plans for additional sites and longer term monitoring were made by Gulf and Caltech scientists and are highly recommended for future research. Industry downsizings and takeovers prevented the final implementation of these research plans. No comment needs to be made regarding the value this data might have had to California and the nation if this program had continued until today.


References Cited

Antropov, P.Y., 1981, Laser gas analysis in solving geological and production problems: Internat. Geology Rev., V. 23, No. 3, pp. 314-318.

Barsukov, V.L., 1979, Geochemical methods of predicting earthquakes: Geochemistry International, V. 16, No. 2, pp. 1-13.

Borodzich, E.V., 1979, Preliminary results on helium pattern variations in seismically active zones: Geochemistry International, V. 16, No. 2, pp. 37-41.

Burtell, S. G., 1989, Geochemical investigations at Arrowhead Springs, San Bernardino and along the San Andreas Fault in Southern California: MS Thesis, University of Pittsburgh.

Craig, H., Chung, Y., Lupton, J.E., Damasceno, S., Poreda, R., 1980, Investigation of radon and helium as possible fluid phase precursors to earthquakes: Technical Report 13, United States Geologic Survey Contract #USGS 14-08-0001-18348, 36 pages.

Drozd, R.J., Pazdersky, G.J., Jones, V.T., and Weismann, T.J., 1981, Use of compositional indicators in prediction of petroleum production potential (abs.). Presented at: the 1981 American Chemical Society Meeting, Atlanta, GA., March 29-April 3.

Elinson, M.M., Pashkov, Y.N., Agababov, G.M., and Ignatiev, I.B., 1970, Measurements of gases distributed above Cu-Mo ore bodies. In: The Geochemistry of Hg, Mo and S in Hydrothermal Processes. Ed. Nauka, Moscow.

Eremeev, A.N., Sokolov, V.A., and Yanitskii, I.N., 1973, Applications of helium surveying to structural mapping and ore deposit forecasting. In: M.J. Jones (Editor), Geochemical Exploration 1972. Institution of Mining and Metallurgy, London, pp. 177-183.

Fursov, V.Z., Vol'fson, and Khvalorski, A.G. 1968, Results of a study of mercury vapor in the Tashkent Earthquake Zone: Dolk Acad. Navk USSR, Vol 179, No. 5, pp. 1213-1215.

Gunter, B.D., 1978, C1-C4 hydrocarbons in hydrothermal gasses: Geochemica et Cosmochimica Acta, Vol. 42, pp. 137-139.

Hadley, D. and Kanamori, H., 1977, Seismic structures of the Transverse Ranges, California: Geologic Society of America Bull., V. 88, p. 1469-1478.

Hunt, J.H., 1981, Surface geochemical prospecting pro or con. Invited Paper at: the AAPG Annual Convention, San Francisco, May.

Hunt, J.M., 1979, Petroleum geochemistry and geology. W.H. Freeman, San Francisco, 617 pp.

Iacopi, Robert, 1976, Earthquake Country, Lane Books, Menlo Park, California, p. 87.

Jones, V.T., and Drozd, R.J., 1983, Predictions of oil and gas potential by near-surface geochemistry: A.A.P.G., Bul., Vol. 67, No. 6, p. 932-952.

Jones, V.T., and Thune, H.W., 1982, Surface detection of retort gases from an underground coal gasification reactor in steeply dipping beds near Rawlins, Wyoming: Soc. Petrol. Engineers, SPE 10050, 24 p.

Kartsev, A.A., Tabarsaranskii, Z.A., Subbota, M.I., and Mogilevskii, G.A., 1959, Geochemical methods of prospecting and exploration for petroleum and natural gas: University of California Press, Berkeley, 349 pp.

King, Chi-Yu, 1980, Geochemical measurements pertinent to earthquake prediction: J. Geophy. Res., V. 85, No. 136, June 10, 3051 pp.

Link, W.K., 1952, Significance of oil and gas seeps in world oil exploration: AAPG Bulletin, V. 36, pp. 1505-1541.

Mamyrin, B.S., 1979, 3He/4He Ratios in earthquake forecasting: Geochemistry International, V. 16, No. 2, pp. 42-44.

Matthews, M.D.., Jones, V.T., and Richers, D.M., 1984, Remote sensing and surface hydrocarbon leakage. Presented at the International Symposium on Remote Sensing for Exploration Geology, Colorado Springs, Colorado.

McCurdy, R., 1979, Site qualification studies of the UCG-SDB site, North Knobs, Wyoming. Prepared by GS&T under DOE contract De-AC03-77-ET13108, May 23. C&M Division Report 64RK076.

Shapiro, M.H., Melvin, J.D., Tombrello, T.A., Mendenhall, M.H., Larson, P.B. and Whitcomb, J.H., 1981, Relationship of the 1979 southern California radon anomaly to a possible regional strain event: J. Geophys. Res., 86, 1725.

Melvin, J.D., Shapiro, M.H., and Copping, N.A., 1978, An automated radon-thoron monitor for earthquake prediction research: Nucl. Instrum. and Methods, 153, pp. 239-251.

Miller, K.F., 1979, Geologic map of the San Bernardino North Quadrangle, California: Department of Interior, United States Geologic Survey Open File Report 79-770.

Mooney, J.R., 1982, Sniffing for oil and earthquakes: The Orange Disk, V. 25, No. 9, pp. 26-29.

Nikonov, V.F., 1971, Distribution of methane homologs in gas and oil fields: Akademiya Nauk SSSR Doklady, Vol., 206, p. 234-246.

Ovchinnikov, L.N., Sokolov, V.A., Fridman, A.I., and Yanitskii, I.N., 1972, Gaseous geochemical methods in structural mapping and prospecting for ore deposits. In: M.J. Jones (Editor), Geochemical Exploration, 1972. Institution of Mining and Metallurgy, London, pp. 177-183.

Pirkle, R.J. and Jones, V.T., 1983, Helium and hydrogen anomalies associated with deep or active faults. Presented at: the 181st ACS National Meeting, Atlanta, GA, March 29-April 3.

Pirkle, R.J., 1985, Hydrocarbon seeps in the Gulf of Mexico. Presented at: the 189th ACS National Meeting American Chemical Society, Miami Beach, Florida. April 28-May 3, paper no. 78.

Preston, Don, 1980, Gas eruptions taper off in northwest Oklahoma: Geotimes, October, pp 18-26.

Reimer, D.M., 1980, Use of soil gas helium concentrations for earthquake prediction, limitations imposed by diagonal variation: J. Geophy. Res., V. 85, No. 136, June 10, pp. 3107-3114.

Richers, D.M., 1984, Comparison of commonly applied soil-gas techniques used in evaluating hydrocarbon potential. Presented at 187th Annual Meeting, American Chemical Society, St. Louis, MO, April.

Shapiro, M.H., Melvin, J.D., Tombrello, T.A., Fong-liang, Jiang, Gui-ru, Li, Mendenhall, M.H., and Rice, A., 1982, Correlated radon and C02 variations near the San Andreas Fault: Geophysical Research Letters, V. 9, No. 5, pp. 503-506.

Shapiro, M.H., Melvin, J.D., Tombrello, T.A., Mendenhall, M. H., Larson, P.B., and Whitcomb, J.H., 1981, Relationship of the 1979 Southern California radon anomaly to a possible regional strain event: J. Geophys. Res., V. 86, 1725 pp.

Sokolov, V.A., 1971, Geochemistry of natural gases: Niedra, Moscow.

Southern Sierra Power Company, 1925, Geothermal areas adjacent to the transmission lines of the Southern Sierra Power Company, Geothermal Resource Council Special Report #3, Geothermal Exploration in the First Quarter Century, pages 145-161.

Wakita, H., 1978, Helium spots: caused by diapiric magma from the Upper Mantle: Science, V. 200, April 28, pp. 430-432.

Wakita, H., 1980, Hydrogen release: new indicator of fault activity: Science V. 210, October 10, pp. 188-190.

Weismann, T.J., 1980, Developments in Geochemistry and Their Contribution to Hydrocarbon Exploration: Proceedings of: the 10th World Petroleum Congress, Bucharest, Romania, V. 2, pp. 369-386.

Zorkin, L.M., Zabairaevi, S.L., Karus, E.V., Kilmetov, K. Kh., 1977, Experience of geochemical prospecting in petroleum and gas deposits in the seismically active zone of Sakhalin Island, IZU (Russ.). UZSSh Uchebn. Zaved, Geol. Razved, V. 20, pp. 52-62.

1Burtell, Stephen G., 1989. Geochemical investigations at Arrowhead Springs, San Bernardino and along the San Andreas Fault in Southern California. Master's Thesis