LIST OF FIGURES

INTRODUCTION

Surface geochemical prospecting for hydrocarbons includes a myriad of techniques ranging from the direct detection of hydrocarbons escaping from subsurface accumulations and source beds to identifying secondary responses in the soils, rocks and biota in proximity to such accumulations or source beds. In the historical sense, the observation of visible seepage of hydrocarbons is the oldest method of prospecting for petroleum. Drake's historic well near Titusville, Pennsylvania, was drilled on the basis of a seep in the adjacent creek bed. The relationship of such "macroseeps" to reservoirs was well established by Link (1952), who stated: "A look at the exploration history of the important oil areas of the world proves conclusively that oil and gas seeps gave the first clues to most oil-producing regions. Many great oil fields are the direct result of seepage drilling". In this respect, few would argue that the presence of a macroseep indicates the presence of petroleum migration or surface source beds. Microseeps, or smaller scale macroseeps, also occur because of the physical continuity necessarily imposed by nature. These are invisible seeps, usually detectable only by sensitive instruments or by the visible result of their effect on the near-surface environment. These microseeps, although perhaps not as obvious or dramatic as macroseeps, are just as valid for the exploration of undiscovered reserves. This chapter presents the conceptual and practical application of microseepage detection and interpretation in the evaluation of areas for their subsurface hydrocarbon potential.
Five factors are necessary to form a hydrocarbon reservoir. These are: (1) a source, (2) a reservoir in which the hydrocarbons can collect or concentrate, (3) a means of trapping these fluids in this reservoir (a seal), (4) a pathway to the reservoir (migration) and (5) the proper timing such that the source, reservoir, seal and migration pathway are present when required. Near-surface seepage of thermogenic hydrocarbons indicates the subsurface presence of a mature source and migration pathway. It also suggests that, if the hydrocarbons are reservoired, the seal is imperfect. This is true of both macroseepage or microseepage. Patton and Manwaring (1984) found that even in an area of extensive evaporites (Hugoton Field, Kansas), the seal was not perfect, and that microseepage could be detected in the vicinity of the Syracuse Fault.
Basically, surface geochemical prospecting is a source rock tool applied at the surface. The magnitude of a microseep from a reservoir is related to the permeability of the migration pathway (and not to the economic worth of the reservoir). A surface geochemical survey is not currently, and perhaps never will be, a stand-alone prospect tool. However, with judicious use, this technology can provide information on the maturity of source beds in a basin and the composition of subsurface hydrocarbons. In addition, detection of surface microseepage allows mapping the surface expression of the migration patterns created by the expulsion of fluids as a basin compacts and matures. When used in conjunction with geophysical and geological information, geochemical data can refine subsurface models of hydrocarbon trapping and migration configurations. It is only through careful analysis and integration with other exploration tools that one can achieve the optimum benefits from this technology.
Near-surface hydrocarbon detection techniques have been shown in both the former USSR and the United States to be capable of distinguishing basins (or large portions of basins) that are unproductive from those that are productive, and of distinguishing the type of production (oil, gas, or mixed oil and gas). This ability has been independently recognized by Jones and Drozd (1979), Mousseau and Williams (1979), Janezic (1979), Weismann (1980), Drozd et al. (1981), Jones and Drozd (1983), Richers (1984), McCrossan et al. (1971), Richers et al. (1982, 1986), Horvitz (1985) and Klusman and Voorhees (1983). Surface geochemical techniques can select which of several frontier basins has the greatest chance of containing reservoired hydrocarbons, and the expected composition (gas, oil, mixed), in addition to high-grading portions of these basins that have the highest potential. The premise that microseeps occur and that they provide useful information for exploration is no longer questionable.

ORIGINS OF LIGHT HYDROCARBON GASES

Origin of petroleum

The formation of petroleum and natural gas from organic matter through increasing depth of burial and temperature has been very well established by many geochemical studies (Tissot and Welte, 1978; Hunt, 1979). As shown in Fig. 5-1, the generation of the light hydrocarbon gases, methane (C1), ethane (C2), propane (C3) and the butanes (C4), occurs in three main stages: diagenesis (<50°C), catagenesis (50-200°C) and metamorphism (>200°C) in which only dry gas and ultimately graphite are formed. During the first stage bacteria acting under reducing conditions on organic substrates in sediments form predominantly methane. According to Hunt (1979), about 82% of the methane and practically all the heavier hydrocarbon gases are formed in the next, catagenic stage. Ethane, propane and the butanes are formed in the temperature range from 70°C to 150°C with peak generation occurring around 120°C. As shown in Fig. 5-1, a very large thermal methane peak occurs near 150°C.
In addition to time, the quantity of gaseous hydrocarbons formed varies with the type of organic source material, which can be broadly classified as sapropelic (marine) or humic (terrestrial). As shown in Fig. 5-1, considerably more C2-C4 and other oil-type hydrocarbons are generated from sapropelic sources than from humic sources. In addition to the different volumes and types of petroleum (oil versus gas) produced from the two source materials, their carbon isotope compositions are different; terrestrial organic matter is reported to have lower 13C concentrations than marine organisms (Galimov, 1968; Silverman and Epstein., 1958).
The carbon isotope concentration of 13C, as compared to 12C, is also very useful for classifying natural gases as to their source type and/or maturity. Maturity is generally proportional to the depth of generation. Schematic diagrams published by Schoell (1983a, 1983b) distinguish the major natural gas types into three end members, as shown in Fig. 5-2. Schoell suggests that most natural gases are admixtures of these three basic end members. As shown in Fig. 5-3, further classification of reservoir gas types can made from data for deuterium in methane 13C in ethane.

Origin of light hydrocarbon gases in the near-surface

The near-surface occurrence of ethane through butanes is of fundamental importance to the purpose of this chapter and to the usefulness of these gases as prospective indicators of buried natural gas and petroleum deposits. An extensive review of the literature suggests that C2-C4 hydrocarbons can be generated biogenically; however, solid proof exists only for methane and ethylene as major products of bacteria (McKenna and Kallio, 1965). A review of the literature shown in Table 5-I provides conflicting evidence for the biogenic occurrence of the C2-C4 hydrocarbons, although most of the literature suggests an abiogenic, thermocatalytic origin for these gases. Compositionally, however, these gases display large variations and do not resemble compositions characteristic of petroleum gases. All these studies are further characterized by methane:ethane ratios in excess of 1000 and a percent methane composition >99%, and are quite uncharacteristic of petrogenic gases. Some of the results reported before the invention of the gas chromatograph must be regarded with suspicion due to limitations of the analytical methods employed and possible sampling collection at locations contaminated by mixed biogenic and petrogenic gases. Russian researchers have illustrated that some of the earlier analytical methods, such as the combustion technique of Kartsev, et al (1959), can measure gases which are mistaken for hydrocarbons.

Laboratory and field evidence of biogenic C2-C4 hydrocarbons

Studies were conducted at Gulf Research & Development Company by Janezic (1979) to investigate the anaerobic microbial evolution of C1-C4 hydrocarbons upon decomposition of various organic substrates including green plant branches, grass clippings, plant roots, decayed wood and pure cellulose. These substrates were chosen to compare results with those of Davis and Squires (1954), Bukova, (1959), Smith and Ellis (1963), Kim and Douglas (1972) and Voytov et al. (1975), who used similar substrates.
The experimental scheme for anaerobic decomposition is shown in Fig. 5-4. Exactly 1.5 g of each substrate was added to a modified 1 liter Sohngen flask and autoclaved at 120°C and 15 psi to ensure sterility, after which each flask was filled to capacity with a sterile inorganic nutrient medium and pH adjusted. Next 50 ml of a heterogeneous innoculum prepared from muds from a local lake was injected into each flask as "inoculates", while 50 ml. of sterile nutrient medium was used for control samples. Headspace C1-C4 hydrocarbons were measured prior to incubation to provide baseline concentrations. Minimum detection limits were 3 ppb on a volume basis using a high sensitivity gas chromatograph equipped with a flame ionization detector. Samples were incubated at 25°C and 36°C over a five week period.
The results are summarized in Table 5-II and Fig. 5-5. Of the organic substrates fermented, green plant branches, grass clippings and plant roots evolved significant quantities of gas after a few days of incubation. Of the C1-C4 hydrocarbons determined, only methane was observed in copious quantities, with minor amounts of ethylene co-produced. Ethane, propane and butanes were not evolved, in good agreement with the work of Kim and Douglas (1972) and Voytov et al. (1975). Peak concentrations of methane and ethylene exceeded 25,000 ppm by volume and 8 ppm by volume, respectively. Ethane evolution would be masked on the chromatographic trace by these quantities of methane, but ethane above background levels was not observed after seven days of incubation. No measurable C2-C4 gases were found in the remaining 30-day incubation period. This experiment suggests that biogenically-generated gases do not initially contain any C2-C4 hydrocarbons.
Another approach was that taken by Coleman (1979), who studied both the chemical and isotopic composition of glacial till gases in Illinois. Coleman obtained the same result, finding that C2-C4 gases are not present in the glacial till gases in Illinois. Coleman also determined 14C age dates on the gases and showed that the biogenic methane varied from about 10,000 years to as much as 40,000 years in age. This is particularly significant since it suggests that no bacterial generation of C2-C4 hydrocarbons occurs, either initially in test tubes or even within the first 40,000 years in glacial till.

Distinguishing petrogenic and biogenic hydrocarbons

The results of the studies by Janezic (1979) and Coleman (1979) strongly suggest that C2-C4 hydrocarbons are not generated biogenically. Most of the previous studies cited appear to be compromised because they were conducted in natural environments in which migrated petrogenic gases might have also been present.
Even assuming that small quantities of C2-C4 gases are generated in biological environments, a methane:ethane ratio greater than 500 appears sufficient to delineate anaerobic gas production from thermocatalytic gases, since such ratios do not occur in petrogenic natural gas deposits. As shown in the test tube experiments (Table 5-II and Fig. 5-5), this value is achieved within two to three days for all substrates studied and exceeds 100,000 after seven days of incubation. Similar values are cited in the literature (Frank et al., 1970; Swinnerton and Lamontagne, 1974; Bernard et al., 1976; Sackett, 1977; Reitsema et al., 1978) as the biogenic threshold in marine geochemical prospecting (Table 5-III).

HISTORY

The first attempt to relate soil gas hydrocarbon concentrations to oil and gas deposits was made in 1929 in Germany by Laubmeyer (1933). Surveying a known oil deposit, he collected samples of soil gas from systematically located boreholes, 1 meter to 2 meters deep, after sealing them from the atmosphere for 24 to 48 hour periods. Using portable analytical equipment, he demonstrated that the samples over the deposits were enriched in methane. Soil gas investigations were initiated shortly after this time in the then Soviet Union by Sokolov (1933), who verified Laubmeyer's results (Kartsev et al., 1959), but measured both methane and heavier hydrocarbons.
Research in the area of surface prospecting was also carried on in the United States during the 1930's beginning with Teplitz and Rogers (1935), Rosaire (1938) and Horvitz (1939). These investigations entailed the collection and analysis of the soils themselves for hydrocarbon gases. The use of adsorbed gas on soils was regarded as an important improvement upon soil gas, as short-term diurnal variations in soil gas flux could be avoided by the assumption that soil would have a tendency to establish over time a metastable equilibrium with the regional flux.

Basic concepts

In the years following these early studies, the basic concepts have remained largely the same, except that detection limits have been improved with technological advances. Recent work has focused on compositional ratios or signatures of the light hydrocarbon gases and their relationship to known hydrocarbon products in the investigated area (Weismann 1980; Jones and Drozd, 1983).
Emphasis has also been placed on the fundamental principles of surface seepage, and the interpretation of the data. It is the opinion of the authors that the overall acceptance of microseep technology in the West has been hindered not only by the emphasis and success of seismic methods but also because of the lack of a comprehensive and public surface geochemistry database. There are, by comparison, more publications on geochemical survey data and basic concepts in the Soviet and Russian literature. As a consequence, many of our discussions rely on experience gained in the private sector in the West, supplemented by literature published in the East.
Although the Soviet Russian literature is clearly positive about surface microseep technology, the western literature is strongly divided. Dehnam (1969) has reviewed several cases crediting geochemical prospecting with petroleum discoveries. Overall success rates range from 25% to 75%. Duchscherer (1980) reports a success rate of 25%, slightly over the industry average, of which 58% are stratigraphic traps. Sealey (1974) reported a success rate of 80% in Texas using a microbiological technique.

Methods of geochemical prospecting

Geochemical methods of prospecting are classified as direct or indirect. The direct methods involve detecting the presence of dispersed oil and gas components in the form of hydrocarbon gases or bitumens in the soils, waters or rocks in the vicinity of oil and gas accumulations. The indirect methods involve detecting any chemical, physical, or microbiological changes in the soils, waters, rocks, or vegetation associated with the oil and gas deposits. Figure 5-6 is a schematic diagram outlining most of the direct and indirect methods currently in use (Kartsev et al., 1959).
Identifying secondary responses generated by leakage of hydrocarbons at the surface has merit and has been reported by many investigators. These include the use of (1) soil microbes (Soli, 1954, 1957; Kartsev et al., 1959; Sealey, 1974); (2) reduction effects (Pirson et al., 1969; Donovan, 1974; Ferguson, 1975); (3) carbon and oxygen isotopes (Donovan et al., 1974) and many other effects as reviewed by Matthews (1985).
As an exploration tool, the identification of hydrocarbon seeps is particularly useful when coupled with remotely sensed images and photographs. Case studies by researchers in the West have shown that secondary indicators of microseepage are often present in the near-surface environment. Examples noted by Horvitz (1972), Donovan (1974), Donovan and Dalziel (1977), Matthews (1985) and Ferguson (1975) have indicated the presence of diagenetic alteration of soils above or adjacent to hydrocarbon accumulations. Work by Rock (1985), Matthews et al. (1984) and Patton and Manwaring (1984) has shown that these effects may often be reflected in the health and type of vegetation over the seep, which also alters the spectral response detected by satellite and airborne sensors. These methods of geochemical prospecting for oil and gas are reviewed in more detail in Chapter 7.
Others have noted changes in resistivity or radioactive signatures above accumulations due to the seepage and possible interaction of ascending fluids and solutions with the encapsulating medium. In some cases the actual removal or addition of soluble chemical species has been noted.
It appears therefore that the direct detection of hydrocarbon gases is not the only means of identifying areas of active microseepage, but that a myriad of other possible secondary techniques can be used either as adjuncts, or as solitary techniques in themselves, to infer the presence of hydrocarbons in the subsurface environment. Most of these utilize the detection and subsequent analysis of gaseous hydrocarbons, while other methods employ the detection and analysis of liquid hydrocarbons, non-hydrocarbon gases, the presence and relative concentration of bacteria, and even the presence (or absence) of inorganic compounds and elements. For the most part, however, methods that directly measure the hydrocarbon content of soils or soil atmospheres have met with the most acceptance.

PHYSICAL BASIS FOR MIGRATION OF HYDROCARBONS TO THE SURFACE

Basic assumptions

The fundamental assumption of near-surface hydrocarbon prospecting techniques is that thermogenic hydrocarbons generated and trapped at depth leak in varying quantities towards the surface of the Earth. That these hydrocarbons present in the near-surface environment represent the products of generation and migration from subsurface points of origin is a necessary conclusion that is universally accepted with respect to hydrocarbon macroseepage. Examples abound, such as the Santa Barbara Channel seeps, the La Brea Tar pits of Los Angeles, the Athabasca Tar Sands, etc. The same relationship has been equally well established, although less commonly accepted, for microseepage.
A further assumption is that the pattern and intensity of this leakage also provides information on preferential pathways that the leakage follows, and as such can be combined with additional geologic information to predict broad subsurface hydrocarbon fairways. In fact, in some instances it has been claimed that such data can identify areas of reservoired hydrocarbons. This last claim is often the subject of heated debate, however, commonly depending in which camp (for or against geochemistry) the explorationist resides.
The physical state of the hydrocarbons during transport is not well known. The reader is referred to Matthews (1996) for a more complete discussion. Nevertheless, most of the models proposed for the transport of these fluids from source to reservoir (aqueous transport, micellular, discrete oil-phase transport, gaseous transport, etc.) are applicable to the continued transport of hydrocarbons from these source beds and/or reservoirs to the near-surface environment. An additional constraint on land is that the last stage of transport is generally above the water table. The physics of transport can be subdivided into two categories, effusion and diffusion.

Physical transportation by effusion

Effusion transport is believed to be the dominant mode of moving hydrocarbons to the reservoir and to the near-surface environment. The sharp localized nature of many anomalies associated with microseepage and macroseepage is more consistent with an effusion model rather than a diffusion model. The experience of the authors in monitoring leakage from gas storage reservoirs and controlled experiments where subsurface gas pressures were typical of true reservoirs suggests vertical transport rates of several meters (tens of feet) per day, clearly greater than the distances of migration dictated by the diffusion mechanism alone (Jones and Thune, 1982).
The sharp and often linear nature of anomalies suggests that faults and fractures play an important part in the movement of these gases. Major linear features discernible on satellite images, as well as other remotely sensed media, from Patrick Draw, Wyoming, show such a relationship (Richers et al., 1982). The Lost River, West Virginia, Geosat study (Matthews et al., 1984) shows anomalously high soil gas values in relation to linear features on imagery. There are anomalously high gas values along faults in the San Joaquin Basin and in the Wyoming-Utah Overthrust Belt (Jones and Drozd, 1983).
The Russians have shown that the magnitude of soil gas values on faults increases dramatically shortly after an earthquake in which fault movement is involved (Zorkin et al., 1977). An extensive study, involving 105 observation wells, 3 meters to 5 meters deep, was set up over the Mulchto oilfield in northeastern Salchalin. A total of 3,700 samples were collected and analyzed over a four month period with the most active wells sampled daily. The results from this study provide impressive evidence for the tectonic relationship of this leakage gas flux (Fig. 5-7). This study leaves no doubt that faults and fractures provide the main control on the effusion of gases from the subsurface.

Physical transportation by diffusion

Diffusion, on the other hand, is a slow and widely dispersive process. Antonov et al. (1971) measured hydrocarbon diffusion coefficients for a variety of rock types from several hydrocarbon provinces in the former USSR. They discovered that the coefficients of diffusion vary over a wide range (from 10-3 to 10-8 cm2/s) depending on the particular lithology and geologic conditions.
The time required for diffusion to occur can sometimes be restrictive. Table 5-IV shows that the time required not only often exceeds the age of the hydrocarbon accumulation, but also quite often exceeds the age of the host rock. If this were the dominant process for migration, then the appearance of soil gas anomalies in the near subsurface would indicate only very shallow accumulations. If a non-steady state exists, where the hydrocarbon signal observed represents only 0.001 times the steady state signal, then these times could be reduced by a factor of 25 times that of the steady-state model. Table 5-IV shows some of the times that this scenario would require. However, diffusion can still be considered as a potential secondary process in microseepage.
Sokolov et al. (1965) calculated diffusion to be sufficient to have resulted in the dissipation of oilfields formed in the Palaeozoic, although to what extent, if any, this has occurred is not known. Furthermore, if any such fields had leakage along faults and fractures or due to erosion of the seal, diffusion might not be able to bring about accumulation before much faster effusive loss caused depletion. Diffusion of benzene into brines adjacent to accumulations has been demonstrated and used as an exploration tool by Zarella et al. (1967).
In productive basins the process of diffusion from both source rocks and reservoirs may be responsible for observed elevated background concentrations which have no apparent relationship to the known accumulations. Alternately the presence of free hydrocarbons effusing outward and upward in areas of microfractures and dispersed by groundwater flow could similarly account for this background. If diffusion were the responsible mechanism, then one might expect broad anomalous zones, with localized effusive "spikes" superimposed on the background. Starobinetz (1983) listed as typical examples of diffusion the studies of Driepro-Douetsk and Anuddria grabens.
Aside from the potential of diffusion for producing a broad dispersive background, it would also be expected to alter the composition of the gases detected in surface methods. Starobinetz (1983) notes that not only can diffusion affect composition, but two additional processes have a similar effect. These are chromatographic separation and selective adsorption.
An example of such chromatographic separation is shown in Fig. 5-8 (Sokolov, 1971 b) which shows the results of a mixture of methane and benzene injected into the bottom of a hand-bored 6-metre deep well. Samples of subsoil air were taken periodically from observation wells 1 meter to 2 meters deep, resulting in the obvious separation shown in Fig. 5-8. Indeed these processes have been cited by detractors of surface prospecting as evidence that the technique is not a valid means of searching for subsurface hydrocarbons deposits, arguing that pulses (non-steady state) of gases will have a different composition from their source because of the chromatographic separation. The example shown in Fig. 5-9, taken from an artificial underground coal gasification experiment near Rawlins, Wyoming (Jones and Thune, 1982), shows that such effects are only temporary. In this experiment, a pulse of gas travels from a retort at a depth of 180 meters (600 feet) and migrates vertically and laterally to a series of observation wells 5.5 meters (18 feet) deep. As shown in Fig. 5-9, although the first gas to be seen in high concentrations is methane, the compositional separation does not last more than a few days before equilibrium is achieved when all the migrating gases have ultimately reached the surface.
As to the second point, if selective adsorption is occurring, the volumes of material escaping over geologic time should ultimately saturate (poison) the adsorber such that no additional material can be adsorbed, or at best, material is exchanged in a steady-state. The result will be a gradual return of the signal to the original composition. This is clearly shown in a study by Zorkin et al. (1977).
There is, however, one important area where diffusion may be responsible for compositional changes; near the soil-air interface. Methane should, due to its lightness and zero net dipole moment, be preferentially lost (followed perhaps by ethane). This would possibly result in an oilier gas signal at the surface. This could be countered by the production of biogenic methane which might partially compensate for this loss.

HYDROCARBON RESIDENCE SITES AT SURFACE

The most important of the direct techniques shown in Fig. 5-6 involve the measurement of light hydrocarbons, methane through butane. Because of their volatility, these light hydrocarbons are generally found in the free pore space. The seepage of hydrocarbons into the near-surface environment above the water table must involves transport through both water-filled and air-filled pores. Sampling these pore gases is obviously one of the most fundamental concepts. However, gases can be bound in the sediment matrix. This latter possibility leads to the development of some disaggregation and desorption extraction techniques.
Discussion of sampling techniques must involve both "free" and "bound" gases. To facilitate this discussion the collection, measurement and analysis of light (C1-C4) hydrocarbons will be broken into two main categories each with two subcategories: (1) free gas, which can be vapor or dissolved gas; and (2) bound gas, which can be adsorbed gas or chemi-adsorbed gas.

Free gas

Gases in the free pore space can be found either in the vapor state or dissolved in water. Extensive research at Gulf Research and Development Company has demonstrated that the "free" and "dissolved" gas seeps yield comparable compositional results, both to one another and to their associated reservoirs when they are properly collected and analyzed (Teplitz and Rodgers, 1935; Jones, 1979; Janezic, 1979; Mousseau and Williams, 1979; Weismann, 1980; Drozd et al., 1981; Williams et al., 1981; Jones and Drozd, 1983; Richers, 1984; Price and Heatherington, 1984; Matthews et al., 1984; Jones et al., 1984). This documentation even extends to numerous observations over artificial underground gas generation and storage reservoirs (Jones and Thune, 1982; Jones, 1983; Pirkle and Drozd, 1984).
Sampling of vapor can be extended to any depth above the water table by analyzing the exhaust air from an air-drilled well. Complications occur because of dilution effects by the air injected for drilling and by the additional fact that the drill bit disaggregates and liberates rock or matrix gas in the process of drilling the hole.
Dissolved gases must be extracted from the aqueous system before analysis. This is usually accomplished by a simple gas-water partition into a vapor phase followed by standard headspace measurement techniques (McAuliffe, 1966). Alternatively a so-called "stripper" continuously partitions the dissolved gases into a carrier gas which is then sent to a gas chromatograph for analysis (Mousseau and Williams, 1979; Aldridge and Jones, 1987). These separations are aided by the very low solubility of the light hydrocarbon gases.
Standard mud gas logging is one variant of dissolved gas analysis conducted on deeper drill holes. A gas trap is deployed in the return mud system for extracting the dissolved and free gases. Compositional information obtained from mud logging gas is useful for predicting the composition of a potential reservoir (Pixler, 1969). These same ratios have been found to be indicative of oil versus gas potential from surface seeps observed from 12-feet deep soil gas measurements or from analysis of gases dissolved in the shallow groundwater (Jones and Drozd, 1983).

Bound gas

Bound gas, which is adsorbed on both the organic and inorganic matter contained in the sediment by means of physiochemical binding, introduces new complexities into defining the appropriate sample for analysis. The difficulty with defining this bound gas is forced by the reality that rocks and/or sediments contain gases of multiple origins. By their very nature, sediments contain both migratory (epigenetic) and indigenous (syngenetic) gases. Migratory gases (biogenic and thermogenic) have migrated to the surface from a deeper, more concentrated source.
Indigenous gas is related to biogenic, diagenetic and thermogenic generation within the rock sampled at the surface and to recycled materials which may contain some physically transported hydrocarbons tightly bound in inclusions or other interstitial sites within the sediment matrix. The nature of the bonding of the hydrocarbons to the grain surfaces leads to two categories, adsorbed and chemi-adsorbed. These form an important part of this discussion because of misnomers involved with the use of the word "adsorbed".
True adsorbed gases are by definition bound to the surfaces of sediment or rock particles. As defined by Greenland (1981) adsorption is the process by which a chemical species passes from one bulk phase to the surface of another, where it accumulates without penetrating the structure of the second phase. Because the light hydrocarbons are so labile, they do not strongly adhere to surfaces and are easily desorbed if the source of these gases is removed. The gas must be replenished by continuous migration in order to maintain the presence of adsorbed gases on the available surfaces.
Bound within the rock matrix, or within certain minerals (calcite, oxide coatings, etc.) gases are chemi-absorbed. They can be removed only by a chemical attack which completely dissolves the rock or sediment matrix. Sometimes these more tightly bound gases not only include indigenous gases, but also might integrate the signal over time, mixing the products of "dead" or "non-active" seepage with those gases actively migrating today. The non-active seeps are often coupled to the lithologies of transported, non-residual sediments (Richers et al., 1986). These last considerations provide two of the main reasons why "free" and "chemi-adsorbed" gases are often found to have no obvious spatial correlation.

Choice of free or bound gas

Any prospector would generally agree that it is desirable to measure only the gas which has migrated from depth, since this is clearly the gas signal which is related to buried reservoirs. The difficulty in doing this begins with choosing the method of sample collection, because there are few sample collection techniques which do not mix the syngenetic and epigenetic gases. Both "free" and "adsorbed" hydrocarbons can often be related to a migratory source, and thus can yield useful exploration information. The free gases appear to be dominated by the migratory gases, unless samples are taken within an outcropping source rock. In addition, the free gases also contain any biological gases which, because of their recent generation, also occur in the free state. If source rocks or recycled source rock materials are present near surface, then the "adsorbed" gases can obtain a major contribution from these sources. Exclusions are often provided by sampling in areas where calcite concretions have been deposited from carbon dioxide generated by biological oxidation of seepage hydrocarbons. This is one reason why adsorbed gas has been successful in marine offshore environments. A good example is provided by studies of the Green Canyon macroseeps (Anderson et al., 1983; Pirkle, 1985).
If one can assure that only migratory gas is measured, then the type of gas measured is unimportant. Including indigenous (syngenetic) gas results in misleading measurements. This is believed by the authors to be one of the primary causes of failure in the application of surface geochemical prospecting. Failure to collect a properly distributed data set can be equally misleading and result in an incorrect interpretation, since interpretations will always be the educated guesses of an explorationist.
Any measurement on a real-world sample is always a combination of the free and bound gas sample types. This is because the process of taking the gas sample generally requires that the sediment or rock system is disturbed by some mechanical means which creates the mixing of these sample types. Because of this unavoidable interaction, we have recognized the need to consider an intermediate sample collection technique which measures the more loosely bound gases liberated into a container containing the core sample.
Typical "headspace" sampling is potentially flawed because of the obvious losses encountered in transferring a core to a sample container. This is further compounded by the difficulty in achieving a rapid and total equilibration of the core gases into the headspace. An alternative technique for measuring the loosely sorbed gas has been proposed by Hunt and Whelan (1979), in which the headspace equilibrium is obtained mainly by mechanical disaggregation and heat. In our opinion, this disaggregated gas should more properly be called "adsorbed" gas. The truly "free" gas is always lost (or at least greatly diminished in volume) from any sample of core which is brought to the surface for collection and handled before being put into a sample container (Sokolov, 1971 b). Typical losses are shown in Table 5-V.
This mechanical disaggregation gas has been usefully applied as a bridge to relating the free and bound gas (Richers et al., 1986). Simple mechanical disaggregation always liberates a considerable volume of gas, which if handled properly has a predictably oilier composition than the associated free gas. This change in composition, created by fractionation of the lighter components, is demonstrated in later examples under case studies.

FACTORS INFLUENCING NEAR- SURFACE HYDROCARBON FLUX

The hydrocarbon flux near to the surface varies according to the supply of hydrocarbons and whether local chemical and biological conditions favor their preservation or breakdown. In addition, hydrocarbon magnitudes at any given location vary with time because of displacement by wind, rain and barometric pumping (Wyatt et al., 1995).

Microbial activity

In a very extensive review, Price (1985) suggested that surface bacterial activity can totally obliterate the gases in a microseep. That this is not typically the case has been demonstrated by extensive research over both macroseeps and microseeps (Jones, 1984). However, bacterial activity does probably contribute to the noisy appearance of soil gas seepage.

Barometric pumping

An example of gas flux related to barometric pumping has been demonstrated over an underground propane storage reservoir. This mined cavern is about 60 meters (200 feet) deep. In order to observe the gas flux related to atmospheric phenomena, plastic ground sheets about 1.5 x 1.5 meters (5 x 5 feet) were buried along their edges to contain any gas flux. The variation with rainfall is shown as vertical bars in Fig. 5-10. 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. However, the same effect is not repeated every time it rains. Around the 19th, 20th, 21st and 22nd days of the month very small barometric changes were observed. Nevertheless, small barometric lows have clearly-expressed gas flux increases. Thus falls in barometric pressure lead to a gas flux that escapes into the atmosphere. This escape occurs despite the extensive microbiological activity that has developed over this cavern.
As shown in Fig. 5-11, a propane profile collected over the top of the cavern requires a log scale to illustrate the enormous range in gas leakage flux. An interesting secondary observation taken from this example is the obvious color changes noted on the soil cores. These chemical changes are related to hydrocarbon seepage and might be used as an additional exploration tool to provide evidence of where the gas leakage has occurred around any type of storage cavern. The soil changes from red-brown to green-black directly over the top of the cavern, where the largest seepage anomalies occur.
Thus the main difficulty with atmospheric sampling is created by meteorological changes which can greatly displace and dilute the seepage emissions. Although it occurs on a different order, it has become clear that the stress fields in the earth can also influence this gas flux.

Earthquakes

The fact that earthquakes may sometimes be preceded by geochemical anomalies was discovered at about the same time in Japan (Okabe, 1956) and the then USSR (Fursov et al., 1968). Earthquake prediction studies in Russia, Japan, and China include extensive geochemical measurements. Chinese geochemical data are reported to have contributed, at least partly, to the successful prediction of several strong earthquakes (Press, 1975). In contrast, the Earthquake Hazards Reduction Program in the United States emphasizes mainly geophysical data.
Limited programs using radon for earthquake prediction began in the United States about 1975, at about the same time as 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). This study confirmed that helium is a deep basement or tectonic indicator which is commonly independent of oil and gas deposits. This is clearly illustrated in Fig. 5-12, in which helium anomalies appear to be associated with the San Andreas fault and with two other deep basement faults. The proposed deep fault on the west flank of the Lost Hills oilfield also acts as a common migration pathway for hydrocarbon gases (Fig. 5-13). This initial study, and the joint research program subsequently initiated by Gulf Research with the Cal-Tech earthquake radon program, was designed to obtain data concerning the rates of change of gas flux associated with tectonic stress in the Earth.
Numerous other examples of gas flux related to earthquakes have been reported, for example, by Kartsev et al. (1959), Fursov et al. (1968), Elinson et al. (1971), Sokolov (1971 b), Eremeev et al., (1972 Orchinnikov (1972), Zorkin (1977b), Melvin et al. (1978, 1983), Wakita et al. (1978, 1980), Barsukov et al. (1979), Borodzich (1979), Mamyrin (1979), King (1980b), Reimer (1980), Shapiro et al. (1981, 1982),) Mooney (1982) and Pirkle and Jones (1983). Particularly intriguing examples have been published by Antropov (1981) of atmospheric methane flux related to petroleum deposits (Fig.5-14) and seismic shock (Figs. 5-15 and 5-16). These measurements were made with adsorption-type gas lasers: one type makes point measurements of the sample in an adsorption tube (Iskatel-2); the other (Luch) measures the specific gas adsorption a long path length (1-100 meters).

SAMPLING AND MEASUREMENT METHODS

There are a variety of sample collection and hydrocarbon analysis methods used in geochemical surveys for oil and gas deposits. In the case of free gas, samples are collected either in atmosphere or, more usually, within the soil. For the bound case soil or rock is collected and the gas is liberated by one of several methods. In practice, however, it is rarely possible to determine solely free gas or solely bound gas.

Atmospheric techniques

The detection of hydrocarbons above the ground surface offer obvious advantages: continuous sampling, no permit requirements, access over rough and hostile environments, large areas covered rapidly. A drawback is that diffusive and convecting mixing in the atmosphere decreases the signal strength with distance from the sediment or soil surface. Nevertheless, the capability of detecting gases in the atmosphere has seen significant developments over the past 10-15 years. Research has resulted in the development of approaches based on microwave energy, infrared lasers and adsorbed hydrocarbons on aerosols carried into the atmosphere by thermals.
The microwave approach has been developed by Owen (1972), Gournay et al. (1979) and Thompson (1981). Although Thompson (1981) has stated that "conclusive proof does not exist that the gases being detected by the sensor are low molecular weight hydrocarbons and nothing else", he has published numerous positive case studies relating the response of one of these instruments to soil gas probe anomalies (Burson and Thompson 1985). Additional technical difficulties result from the fact that microwave adsorption energy levels represent rotational energy in the molecule. Deactivation of rotational energy by collisions can occur rapidly at atmospheric pressure, causing the molecule excited by the microwave energy to lose its adsorbed energy in a non-emission mode, thus reducing the signal-to-noise ratio. This coupled with the low concentrations of hydrocarbons in the atmosphere has meant that the technique has not been extensively tested as an exploration tool.
Remote monitoring of the gas composition of the atmosphere with laser sources has been actively pursued for over a decade, with systems actually built and used for nitrogen dioxide, sulphur dioxide, ozone, carbon dioxide, ethylene, ammonia, hydrazine, hydrogen fluoride and methane. A small mobile laser system capable of measuring methane and ethane in the atmosphere has been developed (at Stanford Research Institute for the Gas Research Institute) for detection of natural gas pipeline leaks (Van de Laan et al., 1985). Another laser technique, based on established physical principles, is LIDAR, which stands for light detection and ranging. The technique uses light from a tuneable infrared CO2 laser to selectively detect methane and heavier gases by adsorption. The technology was reviewed by Grant and Menzies (1983). Briefly, laser light is pulsed into the atmosphere and aerosols, liquid droplets and gaseous molecules scatter or adsorb the light in different ways. Some portion of the scattered light returns to its point of origin, where a telescope-like receiver channels it to a photodetector, which produces an electrical signal proportional to the optical radiation received by the telescope. The length of time between transmission and reception indicates from what distance the light was scattered and the intensity of the electrical signal indicates the concentration of the particles or molecules being monitored. The development of an airborne or truck-mounted system capable of range resolving the location and concentrations of an atmospheric gas cloud will provide an extremely efficient and cost-effective exploration tool for detecting both macroseeps and microseeps in frontier regions.
The third atmospheric technique analyzes the residual liquid and/or condensate hydrocarbon traces on aerosols carried into the atmosphere by thermals (Barringer, 1981). The aerosols are created by gas bubbles which exsolve into the atmosphere from the sea in areas where microseeps create gas bubbles which reach the sea surface. The aerosols are concentrated from large volumes of air and collected by an airborne cyclone sampler carried aboard an aircraft which is flown at 30 meters (100 feet) above the sea surface. Hydrocarbons adsorbed on the aerosols are measured by an Flame Ionization detector which yields a total hydrocarbon signal. This system is claimed to produce direct vertical anomalies over reservoirs at depth. This technology appears reasonable for detection of seepage which is large enough to produce free gas bubbles, but for feeble seepage (i.e., below water solubility levels) the effectiveness would seem to be reduced by dispersion due to underwater currents.

Soil gas

The hydrocarbon gases migrating through soil pore spaces are not dissipated and diluted to the same extent as those in the atmosphere. There are, however, problems posed by the very low levels of hydrocarbon gases and by the diurnal "breathing" of many near-surface soils. In order to overcome these problems, soil gas techniques which integrate the hydrocarbon signal were introduced by Pirson (1946), Horvitz (1950), Kartsev et al. (1959), Karim (1964), Heemstra et al. (1979), Hickey (1983), Hickey et al. (1983) and Klusman and Voorhees (1983).
Karim (1964) published data on laboratory adsorption studies for light hydrocarbons using activated charcoal, molecular sieve (diatomaceous earth) and silica gel. As shown in Table 5-VI, these procedures greatly increase the concentrations available for analysis, but selective adsorption severely affects the relative compositions of the individual gases. The lightest gases are obviously not as effectively trapped by adsorption techniques as are the heavier, less volatile components. This is particularly true for methane and ethane. The adsorption capacity of the substrates are also strongly reduced by moisture content, which may vary from site to site, particularly since the sampling is conducted in the ground where moisture content varies more rapidly than in the atmosphere.
Klusman and Vorrhees (1983) introduced a variation of this technique which uses sample collection on charcoal wire over extended collection times, followed by analysis using a quadrupole mass spectrometer. The advantages cited are lower field expenses, increased field mobility, improved signal-to-noise ration and negation of barometric and other meteorological factors. Major drawbacks are that the most mobile light gas are not collected by the charcoal wire, so that the samples comprise mainly the intermediate to heavier molecular weight components, which include butane through gasoline and diesel. Multivariate statistical techniques are required to interpret the large number of mass peaks recorded, which includes both parent and multiple daughters. In some cases qualitative information based on fragment patterns of the adsorbed compounds is possible (Fig. 5-17). However, different molecular species and their fragment patterns overlap; for example, propane and carbon dioxide have identical masses (44) and thus cannot be separated. The exploration value of these data lies in the demonstrated presence of reservoir-type hydrocarbons at the surface and the composition noted in the lighter to heavier fragment patterns.
The difficulty in interpreting this particular type of data is further compounded by its application in the upper soil zone where the most active plant and microbiological activity takes place. Many organic and inorganic compounds(humic acids CO2, N2O, NO2, etc.) are produced in this zone, all of which are rapidly adsorbed by activated charcoal. These compounds are present in macro concentrations (parts per thousand to percent) and produce fragment patterns which overlap the much lower concentrations of hydrocarbons, which are generally in the ppm range.
Another consideration in using adsorbers is the residence time required for the collector in the soil medium. Care must be taken to ensure that the entire survey area is sampled for the same time interval. Also, each region has its own unique flux rate which will affect the results. In a region with a low flux, the collectors should be left buried in the soil for a longer period of time than collectors in a region of higher flux. An orientation survey should always be designed to establish the proper length of time required to obtain valid data prior to conducting a large scale survey.
Although the concept and approach of this technique are excellent, it does not integrate the flux of hydrocarbons heavier than butanes during the one to two weeks for which the collectors are left in the soil. Hydrocarbons heavier than butanes are liquids, and do not migrate more than a few centimeters during the short collection period. It may be equally effective to place a soil sample in a jar with the collection wire; the collection efficiency could probably even be increased by heating the sample jar.
Direct sampling of free soil gas requires that a sampling probe be inserted into the ground to collect a soil gas sample. The deeper the penetration, the more difficult and expensive the procedure becomes, eventually requiring that analysis be conducted on drilling fluids or rock samples recovered from a hole. Deeper holes almost always encounter water, which also influences the collection of free gases, forcing one to analyze the gas content of some type of recycled water or mud system which is used to drill the hole.
Although sampling from holes of any depth is possible, for simplicity two free soil gas techniques will be discussed and compared (as case studies): shallow probes (Matthews et al., 1984) which penetrate to 1.2 meters (4 feet); and auger holes (Jones and Drozd, 1983) which are 3.5 meters (12 feet) deep. These methods differ mainly in terms of resulting soil gas sample. The shallow-probe samples are influenced more by closer proximity to the atmosphere and the soil/air interface, where the boundary conditions change.
Numerous sample collection methods have been devised for extracting near-surface soil gas samples. Any suitable mechanical device having a small internal volume can be used to collect the sample. Because the probe sampling port must be forced into the soil, some soil grains are shattered by the necessary mechanical force; many laboratory studies have shown that gas is almost always liberated by this process (Collins, 1983). If the probe volume is very small relative to the dimensions of the sample hole, then the magnitude of the collected sample will be dominated by the gas liberated by crushing. In such cases the volume of available gas will rapidly deplete as the soil gas is aspirated from the hole. This effect can be reduced by collecting a larger volume of soil gas, thereby incorporating a large portion of the natural free soil gas into the sample measured, as compared to that gas liberated by forcing the probe into the ground.
One method of collecting gases with a shallow probe system that has proven to be simple and relatively reliable was developed by Burtell (1988). This probe system consists of separate devices for sampling and for creating the probe hole. The device used to make the hole is a pounder bar 1.2 meter (4-feet) long and 1.3 centimeter (1/2 inch) in diameter, with a sliding hammer which is used to pound the bar into and out of the ground. The soil gas probe consists of a short hollow tube, tightly enclosed by a concentric sealing tube of the same diameter as the pounder bar, which is inserted into the ground through the hole made by the pounder. A hand pump or syringe is used to evacuate the residual atmospheric gases from the hollow probe before the soil gas sample is collected. The soil gas sample is collected in a 125 ml glass serum bottle with an aluminum crimp top securing a butyl rubber stopper. The sample bottle is evacuated just before the sample is collected in order to reduce the possibility of contamination and to eliminate atmospheric dilution effects. A sample of the soil gas is drawn into the evacuated bottle. Additional soil gas is then pumped under pressure into the sample container.
Probe sampling using this or any similar portable design can be used in a variety of geologic terrains within the limits of surface geologic features. Since an effective soil gas survey measures gas concentrations which have migrated into the soils, it is important that sample locations be placed in areas with at least one meter of residual soil. Alluvial and glacial deposits can also be sampled in most areas, provided there is not active, high volume, sediment deposition (which would require a deeper sampling method). Water-saturated soils and mud should be avoided because the wet sediments clog the sampler and if the open pore spaces normally present in the soil are reduced by water, then the amounts of free soil gas are much lower than in non-saturated soils.
Shallow probe techniques are prone to near-surface lithologic, meteorological and barometric effects. This means that one must be careful in interpreting background values since the absence of an anomaly in a prospective or producing area may be related to lithology, rainfall, meltwater or barometric pumping. Areas containing anomalous high gas contents, on the other hand, are almost always real seeps, since active flux is necessary to overcome these dilution effects.
Shallow probes have been used successfully at Lost River in Hardy County, West Virginia, Patrick Draw in Sweetwater County, Wyoming (Matthews et al., 1984; Richers et al., 1982), Arrowhead Hot Springs in San Bernardino County, California (Burtell, 1988) and on a large number of surveys conducted throughout the industry. Limited tests by Williams (1985) in the west Texas Permian Basin suggest that shallow probes are difficult to use in this area because of impermeable deposits of caliche and thick salt and anhydrite beds at a depth of about 300 meters. An example of a halo-type anomaly reported by Williams (1985) is included in his thesis.
Despite these limitations, shallow probe sampling is still worthy of consideration because of the low sampling cost and ease of access in rugged areas with limited roads. With this method, small crews of only one or two persons can obtain large numbers of samples at minimal expense. In addition obtaining a permit (if required) is usually relatively simple because permitting authorities tend to classify such surveys as causing minimal environmental impact. The mobility of the soil gas probe sampling technique opens up large areas to geochemical exploration that are otherwise difficult to explore.
Another means of obtaining free soil gas data is from auger holes drilled to 3.5 meters (12 feet). These holes generally yield higher hydrocarbon concentrations than shallow probes. A fairly extensive research program at Gulf Research and Development Company established a database for geochemical exploration using auger holes comprising more than 21,000 analyzes covering 16,000 line km (10,000 line miles) (Jones and Drozd, 1983). The locations of some of the research surveys are shown by black dots on a map of the major US basins (Fig. 5-18).
An important aspect of this technique is the data contain compositional information which not only can be tied to known fields but also is capable of predicting the oil versus gas potential of an unknown area before drilling. This predictive capability has proven to be applicable to several other techniques as well.
A diagrammatic representation of the soil gas sampling procedure used by Gulf Research and Development Company is shown in Fig. 5-19. Soil gas measurements are made in an auger hole, at least 4 meters (13 feet) deep and typically 8.9 cm (3.5 inches) in diameter. A probe jacketed with an inflatable rubber packer is placed in the hole. When inflated, the packer effectively isolates the bottom of the hole from the atmosphere, so that the sealed base of the hole effectively serves as the sample container for the liberated gases. Soil gases are then either pumped into evacuated steel bombs or glass bottles for later analysis, or pumped directly into an on-site dual-column gas chromatograph for determination of the light hydrocarbons, helium and hydrogen. A 1 meter alumina-packed column coupled to a Flame Ionization detector (FID) is used to determine the hydrocarbon content and a 3 meter molecular sieve column coupled to a Thermal Conductivity detector is used for the hydrogen and helium determinations. Carbon dioxide is analyzed continuously using infrared adsorption techniques.
The auger hole technique yields excellent compositional information, even though the magnitudes are influenced slightly by the mechanical disaggregation associated with the drilling process. Compositional results for auger holes are sufficiently important to warrant further discussion here. An empirically-determined range of soil gas data is shown in Table 5-VII and a small selection of auger hole survey results is shown in Table 5-VIII. The geochemical distinction between gas-type basins and oil-type basins was first noted from surveys in the Sacramento and San Joaquin basins in California. Initial compositional data were gathered in these two basins in three separate years with excellent repeatability (Table 5-VIII). Additional surveys conducted in southwest Texas supported the differences noted in California. Final confirmation on the oil versus gas predictions was obtained when numerous surveys were carried out in all three types of productive areas: gas, gas-condensate and oil. Soil gas data from the Sacramento dry-gas, Alberta gas-condensate, and Permian basin oil areas were used to establish statistically valid populations based on histograms that demonstrate a close association with reservoir gases and gas shows in drilling fluids.
Some typical percentages of methane and relative amounts of ethane through butanes in different types of deposits are given in Table 5-IX. These data, taken from Katz and Williams (1952), show clearly that methane decreases in the trend from a dry-gas deposit to a typical low-pressure undersaturated oil deposit containing only dissolved gas but no gas cap. A better demonstration of this relationship comes from the study by Nikonov (1971), who compiled gas-analysis data from 3,500 different reservoirs in the United States, Europe and the then USSR, and grouped them into the populations shown in Fig. 5-20a. 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 include ethane, propane, butane and pentane.
Three of the near-surface data sets from Table 5-VIII are particularly convincing because the soil gas measurements were made in basins that contained only one type of production. As shown by Figure 5-20b, they are the dry-gas production of the Sacramento basin (more than 450 sites), the gas-condensate production in the Alberta foothills (more than 650 sites), and the oil production of the Permian basin (more than 450 sites). Figures 5-20c, 5-20d and 5-20e show methane content (%C1), the methane:ethane ratio (C1/C2), and the propane:methane ratio (1000 x C3/C1), respectively from the soil gas populations over these three basins. These data clearly demonstrate that the chemical compositions of the soil gases from these three different areas form separate populations that appear to reflect the differences which exist in the subsurface reservoirs in these three basins. This correlation is particularly striking when compared with the data of Nikonov (1971), shown in Fig. 5-20a.
The use of hydrocarbon compositions in soil gas prospecting requires enough data to allow statistically valid and separate populations to be defined, so that a particular geochemical anomaly can be related to a geologic or geophysical objective or province. A percentage composition based on only two or three sites having 85% or 95% methane is not sufficient to define a population. As shown in Fig. 5-20a, considerable overlap exists among the intermediate gas-condensate and oil-type and gas-type deposits. In basins having mixed production, prediction of a reservoir gas-to-oil ratio (GOR) is clearly impossible.
Where seeps contain gases from more than one reservoir, their compositions may not match those of any of the underlying reservoirs. Mixing of a shallow oil and a deep gas will generally yield an oily but intermediate-type composition. Without some knowledge of the reservoir possibilities, this type of signature cannot be recognized. Nevertheless, the intermediate nature of the seep will indicate some liquid potential at depth. Thus, dry-gas basins can be distinguished from basins that have at least some liquid oil or condensate potential. As suggested by Bernard (1982), the presence of fairly large ethane-propane-butane anomalies strongly suggests an oil-related source.
Pixler (1969) found that the gases observed during drilling could distinguish the type of production associated with the hydrocarbon show during mud-logging and published the graph shown in Figure 5-20f. Pixler's data were obtained by monitoring the C1-C5 hydrocarbons collected by steam-still reflux gas sampling during routine mud logging. Individual ratios of the C2-C5 light hydrocarbons with respect to methane provided discrete distributions that reflect the true natural variations of formation hydrocarbons from oil and gas deposits. Ratios below approximately 2 or above 200 indicated to Pixler that the deposits were non-commercial. The upper range for these ratios for dry-gas deposits has been enlarged by Verbanac and Dunia (1982), who studied more than 250 wells from 10 oil and gas fields. Their data, shown in Figure 5-20h suggest the following upper limits for dry-gas reservoir ratios: C1/C2 <350, C1/C3 <900, C1/C4 <1,500, C1/C5 <4,500. These ratios clearly aid in defining the transition between thermogenic and biogenic gases. Another empirical rule suggested by Pixler is that the slope of the lines defined by these ratios must increase to the right; if they do not, the reservoir will be water-wet and therefore non-productive. Verbanac and Dunia (1982) suggested that a negative slope connecting individual ratios may result from fractured reservoir zones of limited permeability.
Auger hole soil gas data for the surveys over the three basins described above are plotted on a Pixler-type diagram of reservoir gases in Fig. 5-20g. Direct comparison of these two independent data sets is very striking and proves the concept of migration of reservoired hydrocarbons to the surface. It is important to note that amounts of migrated gases almost always decrease in the following order: methane > ethane > propane > butane. Thus, in a Pixler-type diagram, soil gas data, like reservoir data, generally plot as line segments of positive slope for the soil gases to represent a typical migrated seep gas. Exceptions to this order have been noted where surface source rocks were drilled, which thus far have yielded ratios with lighter gases depleted in relation to heavier gases. According to Leythaeuser et al. (1980), this would be expected if gases in the boundary layer very near the surface followed a diffusion model. Thus, compositional changes related to diffusion might be expected at or very near a boundary layer where the hydrocarbon gas concentration approaches zero. This behavior has been observed when comparing soil gas probe data measured at very shallow depths (0.3 to 0.6 m, 1 to 2 ft) with the corresponding data from 4 meter (13 feet) auger holes. The shallow probe data are always "oilier", indicating preferential loss of methane and implying diffusion from the 4 meters (13 feet) level to the surface. If diffusion were the dominant migration mechanism, a chromatographic effect would be expected for gas that migrated through the Earth. The fact that the compositions of the soil gas data from auger holes match the underlying reservoirs confirms that the major migration mechanism to the near-surface must be via faults and fractures, rather than by diffusion.
The percent-methane compositions from the auger hole surveys conducted over the Sacramento and San Joaquin basins are plotted in Fig. 5-21. There is a decrease from 98% methane in the north of the Sacramento basin to 90% in the south part, whilst the soil gas over the San Joaquin basin has 82% methane. These data imply that a soil gas grid would have defined local differences regionally. Furthermore these geochemical data are repeatable (Table 5-X); the percent-methane values on Fig. 5-21 were all determined at least two or three times over a three-year period and found to be repeatable. Compositional data have remained repeatable throughout our experience with soil gas surveys.

Dissolved gas

In offshore prospecting "sniffers" have been used to detect anomalous hydrocarbon concentrations in bottom waters. An extensive review of the literature was published by Philp and Crisp (1982). Some of the most significant results reported by Williams et al. (1981) are highlighted here.
Gulf Research and Development Company designed and operated several marine seep detectors which were employed aboard various research vessels, such as the RV Hollis Hedberg and its predecessor the RV Gulfrex. These ships circumnavigated the globe and conducted extensive detailed surveying in areas such as the Gulf of Mexico (Mousseau , 1979). The RV Hollis Hedberg system employed three separate water inlets which, whilst the ship was underway at normal seismic survey speeds, continuously supplied sample streams from the near surface, intermediate depths to 135 meters (450 feet) and a deep towed sample inlet at a depth of nearly 180 meters (600 feet). Each sample stream is analyzed for seven hydrocarbon gases once every three minutes with a sensitivity that depends upon the hydrocarbon and, for example, is about 5 x 10-11 liters of propane at STP per liter of seawater. By using multiple depth inlets, surface contamination can be demonstrated to have no effect on seeps observed by the deep inlet. At sea "sniffer" geochemical data from a deep tow inlet were superimposed to scale on a seismic section to aid the explorationist in making real time evaluations of hydrocarbon potential of structurally significant areas.
As for surface soil gases, a powerful confirmation of the validity of marine geochemical data can be shown by the very close agreement between the composition of component hydrocarbons in production gases and the composition of seep anomaly gases in the same areas. Figure 5-22 shows the well database used for this confirmation in the Gulf of Mexico (Rice, 1980). For each of the 32 fields shown on this figure, the USGS has published the composition of gases produced from predominantly gas fields, oil fields and combined oil and gas fields or condensate fields.
A crossplot of the compositions of gases from all field types is shown in Fig. 5-23 (Williams et al., 1981). The underlying color code on this figure was chosen to distinguish oil, oil-condensate, gas-condensate and gas production using the Rice well analysis data as a standard.. The log of the ratio of ethane to propane and butane is plotted against the log of the ratio of methane to ethane plus propane. A distinctive compositional clustering of gas anomalies signifies different kinds of production: oil anomalies occur near the origin and become gassier as the points move up and to the right in Fig. 5-23. A comparison of 146 sniffer geochemical anomalies from the same part of the Gulf of Mexico is plotted in Fig. 5-24b for direct comparison with the Rice well data shown in Figures 5-23 and 5-24a. As shown, the overall distribution is very similar to the well data. Figures 5-24d and 5-24e illustrate the contrast in composition of dissolved hydrocarbon anomalies from a gas area and an oil area in the Gulf of Mexico. This type of regional separation was found to be typical of surveys conducted throughout the world.
The fact that production and surface anomaly gases correspond both onshore and offshore is significant. It proves that the observational techniques are valid despite the great variation in these surface environments.

Headspace gas

A headspace sampling technique is commonly employed for the analysis of canned samples from drilling returns and from shallow sediments. In this technique a controlled volume of sediment is placed in a can or jar filled with a measured volume of degassed brine. The can is sealed and a measured volume of brine is displaced with nitrogen to create a known volume headspace. The can is then allowed to come to equilibrium. The concentration of light gases can then be measured by syringe injection of a headspace sample into a gas chromatograph equipped with an Flame Ionization detector.
In order to maintain reproducibility it is important to measure all volumes accurately. In a typical operation using 500 ml (one pint) cans, the procedure is to place 300 ml of degassed salt water brine into the 500 ml can and add sediment until the can is filled to the brim, giving 200 ml of sediment and 300 ml of brine. The can is sealed and then zero-grade nitrogen is injected through a prepared septum to displace 100 ml of brine and leaving the can with a 2:2:1 mixture of 200 ml brine, 200 ml sediment, and 100 ml headspace.
Experiments have shown that a fairly long time is required for the adsorbed sediment gases to completely equilibrate with the headspace. This equilibrium time is shortened by heating and shaking the cans before analysis. A generally accepted procedure is to heat the cans for about 12 hours at 60oC to 70oC, followed by shaking in a paint mixer for five minutes. After heating and shaking, the cans are allowed to stand for at least five further minutes to ensure that dissolved gases return to the headspace.
One of the drawbacks to using this technique is the need to freeze the canned samples if they cannot be analyzed within one or two weeks of their collection. Failure to follow this procedure can create problems because of the generation of biogenic gas in the cans or the bacteria oxidation of the hydrocarbon gases to carbon dioxide.
Hydrocarbon concentration values are reported in terms of ppm by volume in the nitrogen headspace or as ppm or ppb by weight, normalized to the weight of sediment. Gases concentrations reported by weight are not truly representative of the actual gas migrating from depth because some of the free gas has been allowed to escape during collection and sample preparation. Furthermore, the sorbed gas is never completely extracted into the headspace, and may not always reflect the true gas content of the soil.
The headspace sampling technique can yield useful results if sufficient numbers of samples can be collected to use statistical populations to suggest anomalous areas. One should always exercise caution, however, with respect to characterization of gas composition, since evaporation during the collection stage always occurs, resulting in the relative depletion of the lighter gases.

Disaggregation

Extensive soil gas sampling programs carried out by the petroleum exploration industry have demonstrated that the crushing and/or disaggregation of soils (including the action performed in drilling auger holes) is an important component part of the extraction of gas from the soil. This suggests that it would be advantageous to employ a soil core disaggregation technique which would closely mirror the effect of auger hole drilling. A device developed at Citco and commonly used in both industry and academia for analyzing well cuttings appears suitable for accomplishing this objective (Whelan, 1979; Hunt and Whelan, 1979; Whelan et al., 1980). In fact, Richers (1984) has demonstrated successfully that in some instances, such as at Rose Hill, Virginia, and in the Western Overthrust Belt, the results obtained by this technique are in very good agreement with data from auger holes.
The device used in this technique is a small stainless steel ball-mill containing two stainless steel or ceramic balls which crush and disaggregate the sample when the ball-mill is shaken (Fig. 5-25). This approach concentrates the loosely bound adsorbed gases into the headspace of the ball-mill. Because of the equilibrium problem mentioned above under headspace techniques, this sampler was adapted by Whelan (1979) and Whelan et al. (1980) to ensure that lithified sediments and cuttings are completely broken up during analysis. Basically, the technique is as follows.
A small (but constant) volume of sediment, soil or cuttings is placed into the mixer cell along with two ceramic or stainless steel ball-bearings, and water is added to bring the remaining headspace to 10 cc. The mill is sealed and placed in a SPEX/Mixer-Mill and agitated for about five minutes. The cell is then immersed into a hot-water bath at 90oC for three minutes. A 1 ml aliquot of gas-free water is injected into the cell through a septum-sealed side arm on the cell, and then a 1 ml aliquot of the headspace is sampled using a locking gas-tight syringe. The sample is then hand injected into a gas chromatograph equipped with a Flame Ionization detector for analysis of the disaggregated gases. It is assumed that these gases represent micopore gas, some free gas and lightly adsorbed gas on the sample medium surface.
This technique (or modifications of it) has been used in the analysis of well cuttings and deep sea cores (Hunt and Whelan, 1979), in addition to surface geochemical prospecting (Richers et al., 1986; Richers and Weatherby, 1985).
Initial tests of this method were conducted at Gulf Research and Development Company for comparison with the auger hole technique and to gain a better understanding of the relationship between free gas and adsorbed gases liberated by the drilling process. To be an effective and viable technique, the disaggregation desorption method must be able to distinguish between oily and gassy areas. An area known to be predominantly oily, Rose Hill in Lee County, Virginia, and another known to be predominantly gassy, the Gulf Research Facility in Pittsburgh, Pennsylvania, were chosen as initial test sites. Both areas had been sampled previously using the auger hole technique, allowing the new data to be compared with the established data sets (Richers, 1984).
The Rose Hill test site includes 126 soil cores of which 51 fall within 300 meters (1,000 feet) of the earlier auger holes. Despite differences in the sample locations and depths, both techniques correctly identify the area as oil-prone. Table 5-XI shows the relationship between the diagnostic gas ratios (Jones and Drozd, 1983) and the results of the two surveys (Richers, 1984). It is obvious that the ball-mill technique accurately describes the oil-prone nature of the Rose Hill oil field. However, the data of Table 5-XI suggest a slight difference the composition of the hydrocarbons detected by the two techniques. In the auger holes the soil gas is slightly drier (methane-rich) than the soil gas obtained by ball-mill disaggregation-desorption. This shift may reflect the preferential loss of methane in the from the shallow cores compared to the deeper auger holes and the use of core samples instead of free gas measurements. The other gases are essentially the same in both techniques: the iC4/nC4 ratio for the disaggregation technique is 0.34, and the auger hole technique yields a value of 0.40; the C2/C3 ratios are comparable at 1.84 for the disaggregation technique and 1.76 for the auger hole technique. In addition, the intercorrelation of the various hydrocarbon gases in the disaggregation data set is higher than that for the auger hole data. This high degree of correlation among the gases may reflect a near-equilibrium condition achieved through time for the adsorption-desorption process in soils. Hence, the signal seen by the desorption technique may effectively integrate and smooth rapid changes one might expect to see with a free-gas technique such as auger holes.
At the Gulf Research Facility in Pittsburgh, Pennsylvania, there are two producing gas wells, and 38 sites were selected to test the ability of the disaggregation technique to define gassy areas. Not only did the test yield gassier results than those obtained at Rose Hill, but also the results were again comparable to those obtained using the auger hole technique. Table 5-XII is a compilation of these results. Clearly, the two data sets reflect a more gas-prone area for Gulf Research Facility than for the Rose Hill area. Although the data set for the disaggregation technique is only half of the size of the data set from the auger holes, it still yields useful information regarding composition of the subsurface reservoirs.

Acid extraction

A technique which measures only the most tightly bound gas was originally developed by Horvitz (1939, 1945, 1950, 1954, 1957, 1965, 1969, 1972, 1980, 1981). In this technique the sample is subjected to acid digestion under vacuum at an elevated temperature of about 80oC. Further developments by Debnam (1969) and Horvitz (1972, 1981) involved corrections for lithology to reduce the effect of acid-soluble minerals biasing the data. Debnam (1969) noted that soil samples could be dried, pulverized and sieved without affecting their hydrocarbon content. He also noted that sieving sand samples to <200 mesh gave analytical values comparable with those produced by shale samples from the same location. Horvitz developed a wet-sieving technique to concentrate the analysis on only the clay-fraction of the sediment.
McCrossan et al. (1971) evaluated the acid-extraction technique in the western part of Alberta. This extensive survey of over 4561 samples covering 15 townships concluded that the distribution of anomalous points was random and was strongly biased by samples rich in carbonate minerals. Adequate corrections for amounts of acid-soluble material were not successful and it was concluded that this method could not be used in areas covered by glacial till.
As early as 1940, Sanderson had discussed a number of factors that affected adsorption of hydrocarbon gases by soils. He noted that the ability of the soil to adsorb any gas depends upon the type of gas, the characteristics of the soil and the conditions under which the soil is exposed to the gas. Adsorption will depend upon the type and surface area of particles and their chemical composition. The surface reactivity will be modified considerably by the presence of previously adsorbed molecules, such as carbon dioxide, water and mineral ions. The condition of adsorption is complicated by temperature and pressure and length of exposure time in addition to concentration and species of gases present. Adsorbed gas data can, at best, be only approximations of the original mixture of migrated gases. Another possible problem lies in the quantitative desorption of the gases from the mineral components of the soil.
Sanderson (1940) observed up to six-fold differences in the ability of soils to adsorb hydrocarbons in his laboratory. He also noted that the adsorptive characteristics of the colloidal soil systems would vary slowly with moisture content, time and season. Of particular significance was his observation that the adsorptive capacity for hydrocarbons on wet soil was only a small fraction of that for dry soil. A further complication is created by near-surface biological activity that creates wide variations in the content of carbon dioxide, nitrous oxide and other biological gases. Overcoming all these problems is probably impossible; however, it will suffice if the gases are liberated in proportion to the amounts present so that the analytical results bear some relationship to one another, and allowing identification of potentially prospective areas.
Various other approaches have been devised in attempts to overcome this problem. Bays (US patent no. 2,165,440) suggested correcting for the sorptive power of the soils and McDermott (US patent no. 3,120,428) suggested correcting for the surface area. An alternate technique proposed by Thompson (1971) used ethylenediaminetetracetic acid (EDTA) at about pH 7 and slightly heated in order to decompose the carbonate minerals under conditions that do not release such large quantities of carbon dioxide. Thompson reports that a comparison on duplicate samples shows that the EDTA technique consistently releases from 94% to 99.5% of the hydrocarbon gases released by the standard strong-acid treatment. A further refinement of this method by Thompson et al. (1974) separates a critical carbonate mineral before analysis. This critical mineral was almost always found to be dolomite, but occasionally is other carbonate minerals, such as iron or calcium carbonate. The ratio of hydrocarbons per unit of critical mineral is then plotted to form a geochemical prospecting map. This technique was reported to highlight a salt dome in the Gulf of Mexico on which a major oil discovery was made after the survey was conducted.
Poll (1975) addressed this problem of lithologic corrections by dividing data according to desorption efficiencies based on their physiochemical properties. The first step is to prepare a detailed lithological description of the samples. This involves a differentiation on sediment lithology, sample coherence, structure, cementation and mineral types, including carbonate and sand percentages. This information is used as shown in Fig. 5-26 to classify the samples into homogeneous sets for each of which the average, or background, gas content is computed. The gas content in each group is assumed to be distributed according to a Laplace-Gauss law. Each subset is then assumed to have a uniform efficiency of desorption and its own background and anomaly threshold. As shown in Fig. 5-27, for calcareous sediments these are very high, due to the effectiveness of the acid attack. The mean normal standard can be computed for each set yielding dimensionless values that can be added together for mapping, regardless of the sediment type. This technique has been applied by Poll (1975) in the Gippsland Basin and by Devine (1977) and Devine and Sears (1985, in the Cooper Basin in Australia. Reasonably positive results were reported in all three cases.
The acid-extraction technique relies on the ability of soil and minerals to retain hydrocarbons that migrate past them through the soil pore system. It is therefore not subject to the fluctuation involved in the soil air system but hopefully represents some averaged or integrated signal over time. As noted above, the samples must be corrected for lithologic effects by only making comparisons within a given lithology or by specifically analyzing certain minerals. Corrections must always be applied because adsorption occurs in both the fine-grained fractions and in carbonates, which often release disproportionately large amounts of hydrocarbons.

Fluorescence

As an extension of the light hydrocarbon gas analysis, UV fluorescence spectroscopy can be used to measure the oil potential of near-surface sediments by analyzing their aromatic hydrocarbons. The method is highly sensitive and selective method for the analysis of oil components, particularly those containing one or more aromatic functional groups. Using spectroscopic scanning, complex molecular aggregates, such as those found in crude oils, can be rapidly characterized and quantified on the basis of their combined intensity wavelength distribution or "fingerprint".
The fluorescence spectra of nine crude oils of different gravity are shown in Fig. 5-27 (Purvis et al., 1977). These two-dimensional fluorograms were produced by exciting at 265 nm and scanning from 250 nm toward the red end of the spectrum. The accepted procedure for illustrating the change in the emission spectrum associated with different gravity crude oils is to measure the intensity of fluorescence at two wavelengths: 320 nm for light aromatic compounds; and 365 nm for the heavier, multiple-ring aromatic compounds. The intensity of the fluorescence emission is proportional to the quantity of aromatics in the extracted sample. The standard field method employs a rapid wet extraction process which dissolves loosely bound trace aromatics into hexane. This extract generally favors the heavy oil fraction, which is in hydrophobic association with the sediment.
A second phase of in-depth, total scanning fluorometric analysis is often performed on selected anomalous samples identified by the field fluorescence or by adsorbed and interstitial gas data. These samples undergo freeze drying followed by a thorough cyclic extraction in hexane to optimize recovery of associated sedimentary aromatics (Brooks et al., 1986). The oil type is then determined by total scanning fluorescence which employs step-wise scanning of excitation and emission wavelengths to produce a three-dimensional fingerprint fluorogram (Fig. 5-28).

SAMPLING STRATEGY

Spatial patterns of near-surface hydrocarbon composition and concentration are prime factors when interpreting the survey results. Results from a poorly-designed or an uncontrolled survey can be difficult or impossible to interpret, and can lead to a completely erroneous assessment of the hydrocarbon potential of an area. An improperly-spaced grid with sample spacing in excess of target size can result in only the most cursory assessment of potential, with anomalous areas appearing as localized single-point anomalies (Matthews, 1996a).
The distribution of sample sites in a geochemical survey is largely governed by the purpose and budget of the survey. For regional surveys a sampling density of one sample per 2 km2 to 5 km2 seems adequate. Such a density still allows for the discrimination of regional ambient backgrounds from secondary backgrounds. Detailed diagnostic work requires a close-spaced grid, sometimes with a sample interval of only a few tens of meters.
Regional sampling is generally performed using a modified grid because a regular grid, on which samples are taken at the intersections of a straight lines, does not minimize cost or maximize information. We recommend that sample positions be chosen within grid cells according to ease of access (minimum cost) and along zones of known or inferred fracturing and faulting (maximum information). Satellite imagery, aerial photography, seismic data and other data are useful when attempting to site samples on or near fractures and faults. The analytical results from a regional survey should yield some indication of compositional and/or magnitude "sweet-spots", either as isolated data points or small clusters. If the objective is merely to evaluate whether a basin has a source section, and general trends of where it is mature and focused to the surface, a regional study may be all that is required. A more detailed follow-up survey, however, is recommended if the objective is to highlight the zones of higher hydrocarbon potential.
One method commonly employed for detailed surveys is to sample seismic shot holes, further providing a means to easily tie the geochemistry to subsurface structure. Because seismic lines are not normally placed on a close-spaced grid, infill sampling between seismic lines is usually recommended. It should be emphasized that in order to define a target adequately, approximately 70% of the data should be collected in presumed background areas beyond the immediate target area. An embarrassingly large number of surveys have been performed in which sample locations do not extended more than one or two sites beyond the anomaly. The result of this misplaced desire to save money is often an ambiguous survey interpretation.
The selection of a technique that is inappropriate for the surface geologic conditions in part of the survey area can also lead to erroneous results. An example is the use of the acid extraction technique on glacial till or acid soils, which normally yield low results. Without regard to the particular constraints on such data, one could easily overlook a favorable area. In this case another technique such as free gas would be more representative.

DATA INTERPRETATION

There are many ways to analyze hydrocarbon gas data with no one particular method being correct or incorrect. Common sense and a deterministic approach to sound geologic models are the best guidelines. Integration with other data such as structure, lithology, soil types and hydrogeology, to name a few, can be most fruitful.

Preferential pathway model

The lack of a model explaining the mechanisms and constraints of hydrocarbon leakage is often an obstacle to the acceptance of surface geochemical prospecting (although a similar lack of understanding of the migration of hydrocarbons from source beds to reservoir has not precluded the acceptance that migration occurs). Assimilation of the data, however, suggests that much can be explained by a relatively simple model. The conclusion that effusion is the dominant mode of migration enables us to use the visual patterns associated with macroseeps as a basis for our microseepage model.
Link (1952) and Levorsen (1967) have summarized the geologic conditions and controls on macroseepage. There is no reason to expect that these controls should not apply as well to microseepage; the only real difference should be a matter of scale. In addition to seepage directly from exposed source beds, controls on surface seepage include: (1) the surface exposure of reservoir beds or porous carrier facies; (2) porosity associated with unconformities; and (3) surface expressions of faults and fracture systems that are pervasive to depth. These controls may be summarized as the focusing of migration along preferred permeability pathways. Horizontal migration along the pathway is dominated by grain or bed permeability (including old erosion surfaces and other unconformities), whilst vertical migration is controlled by cross-stratigraphic discontinuities.
Horizontal pathways deflect the surface location of the anomaly laterally away from its subsurface origin. Thus if an anomaly is associated with the surface expression of a porous formation, one should suspect a down-dip source (or down-groundwater gradient source). The same conclusions can be inferred for anomalies associated with unconformities, low angle faults, and listric faults.
Vertical pathways are dominated by the intersection of high angle faults and fractures with reservoir and carrier beds. In this case the surface expression of the source of the hydrocarbons will lie directly above, or only slightly displaced from the source. The presence of multiple, stacked porous zones also often results in a surface geochemical expression that is approximately vertically over its subsurface origin.
The role of faults and fractures is particularly important for microseepage and some further comment is in order. The close association of near-surface geochemical anomalies with faults and fractures has been pointed out by, amongst others, Horvitz (1939), Sokolov (1971 b), Richers et al. (1982), Jones and Drozd (1983) and Matthews et al. (1984). McCrossan et al. (1971) point to the close association of high concentrations of hydrocarbons in the surface environment with photolineaments. McDermott (1940) suggests that the permeability of shale is dominated by microfractures and that these fractures are preferentially normal to the bedding plane. This potentially important role of microfractures is emphasized by Rosaire (1938), who correctly points out that the failure to observe displacement does not eliminate the existence of a fault or fracture.
The high permeability of fractures causes them to preferentially focus fluid flow. The effectiveness of fractures as mass transport systems for fluids is evident from a casual examination of mineralization in fractured rocks and leakage of groundwater at fracture outcrops. Similarly, these fractures act as preferential hydrocarbon pathways, focusing their flow from source beds to surface.
Faults and interconnected fracture systems have a significant effect on the magnitude and, less commonly, composition of the near-surface gases. The effect on magnitude is generally to increase concentrations in fractured areas, whilst the effect on composition theoretically should be preferential loss of lighter gases compared to heavier gases. In practice, gas compositions on faults are often lighter or heavier than those at neighboring sites. This is believed to be controlled primarily by the depth of the fault and the composition of the subsurface gases it conducts. Thus deep, basement-related faults are often gassy because they tap deep over-mature sediments. Shallower faults are often oily because large molecules migrate more easily than the lighter compounds.
The increase in magnitude in fracture systems can often be abrupt and localized. It commonly spans several orders of magnitude, going from nil to macroseep levels in the extreme cases. In an area where there is no significant source of subsurface hydrocarbons, there are no high magnitude soil gas signals, even on faults and fractures.
In a hydrocarbon-bearing environment, however, overall high variance in the data is more often the case, but the anomaly-to-background ratio is smaller in non-producing areas than in producing areas. Some of these anomalous zones are associated with preferential leakage directly from a source bed, while others are from reservoirs. Since some faults and fractures are sealed locally along their lengths, high magnitude signals do not occur everywhere along their length. Thus, we often observe "hydrocarbon spots", similar to the "helium spots" discussed by Wakita (1978). Naturally, those faults penetrating only source beds will show a signal that reflects the source beds, whereas those penetrating a reservoir or both reservoir and source beds will exhibit a larger anomalous signal. It is not known, however, if one can truly distinguish between the two types in all instances, although extremely high magnitudes are felt to be more diagnostic of reservoirs, as seepage volumes are expected to be larger from reservoirs than from a source bed (Hunt, 1981).
The expectation that all samples in a leaking fracture zone are higher than those outside the zone is simplistic, and is not always realized in practice. A faults or fracture is rarely one discrete plane, but zones of broken or disrupted strata, separated by relatively unaffected competent strata. It is analogous to a fractured pipe: certain portions of the conduit are solid, whereas the fractured section is composed of both intact fragments and cracks. Fluids flowing through the pipe are going to leak in the fractured areas of the pipe but not in the solid-walled portions. Similarly, even in the fractured zones, the fragmented areas will leak only through the fractures, not through the fragments of pipe between the fractures. Extrapolating this model up to geologic scales, sampling outside the fracture zone is expected to give values that are typical of the background of the area. Within the fractured sample zone, sample sites may intersect discrete fractures or encounter coherent blocks between the fractures (Fig. 5-29a). The intensity of fracturing, and hence the probability of the fractures interconnecting, increases toward the center of the fracture zone, as shown in Fig. 5-29b. Therefore, samples taken near the center would be expected to be a mixture of high values (intersecting fractures that connect), median values (intersecting fractures that do not connect) and low values (not intersecting fractures). Further from the center of the fracture zone, the maximum values fall until they merge with those typical for the background of the area. This distribution of free soil gas magnitudes as a function of distance from the center of the fracture zone is shown in Fig. 5-29c (Richers et al., 1986). Disaggregation data from Patrick Draw exhibit a similar pattern, although the increase near the center of the fracture zone is not as great; acid extraction data from this example show no obvious relationship, clearly suggesting that different analysis techniques are extracting gases from different sources.
The following examples illustrate the means of interpreting what are often referred to as direct anomalies using preferential pathway models. These direct anomalies may be either vertically over their subsurface source, or laterally displaced by varying amounts (Sokolov, 1971 b; Pirson, 1969; Laubmeyer, 1933). What is generally not realized is that most areas contain microfractures to the extent that they allow gases to escape vertically.
Using a coal-burn experiment in the central Wyoming coal region, Jones and Thune (1982) showed that a definite vertical migration component could be identified. In that experiment, gases formed during combustion appeared both in soil gases directly above the retort and up-dip along the bedding planes of the strata involved in the burn. Thus, vertical signals from a known subsurface origin were shown to exhibit cross stratigraphic migration, presumably due to the presence of fractures in the system. A second horizontally displaced component also migrated along the bedding planes at the same time.
An example of the use of direct anomalies and the preferential pathway model is shown in Fig. 5-30, which shows an idealized subsurface cross-section through the Lost River field in West Virginia along with a propane profile (Matthews et al., 1984). From this profile and with some knowledge of the geology, it can be seen that a large anomaly is probably caused by updip leakage of the fractured Devonian Oriskany reservoir at depth. This outcrop anomaly is due to updip leakage along the bedding plane of the reservoir facies. A smaller but significant anomaly is related to leakage from a fault which strikes along and to the east of the crest of the producing anticline. Blind drilling on the outcrop anomaly would have resulted in a dry hole, whereas drilling just west of the fault anomaly would have encountered the producing structure. Appropriate geological modeling identifies the location at which to drill.
An alternative to the direct anomaly interpretation method relies on identifying one of two types of halo: (1) local lows, source background areas surrounded by highs; or (2) extremely low areas, surrounded by moderate areas of concentration. These halos are consistent with the initial results obtained with soil gas analysis techniques (Rosaire, 1938; Horvitz, 1939, 1945, 1954, 1985; McDermott, 1940; Rosaire, et al., 1940), which indicated that adsorbed and occluded hydrocarbons occur in greater quantities around the edges of production, whereas relatively lower values are found directly above production. Halo anomalies have been recognized in many regions of the former USSR (Kartsev et al., 1959). Horvitz (1969, 1980) has emphasized that although other hydrocarbon distribution patterns are recognized, including direct anomalies, the halo pattern continues to be the most common type found in conjunction with important oil and gas accumulations.
Numerous explanations have been put forth as to why halos form around hydrocarbon accumulations. Most of these link the phenomena to the impedance effect of a diagenic mineralization zone overlying the main part of the petroleum accumulation. Such a zone would tend to reduce the ability of gases to seep vertically, except along well pronounced fracture systems. Hence, most transport would be deflected around the edges of the occluded zone. The occluded zone could form by any number of diagenic processes. Rosaire (1940) suggested that the greater solubility of carbon dioxide in petroleum, as compared to water, results in the conversion of bicarbonates to less soluble carbonates over an accumulation. An initial chimney effect would result in a greater supply of bicarbonate being present above an accumulation resulting in the cementation. Rosaire (1940) also proposed the reduction of sulphates to sulphides over an accumulation. Fenn (1940) reintroduced another process which was first introduced by Mills and Wells (1919). This model is based on the evaporation of ground moisture as the result of gas expansion which results in the subsequent precipitation of minerals at shallow depths. The origin of the blocked central portion over an accumulation implies that gas-induced evaporation occurs more effectively over an accumulation than along its margins. This model is consistent with results on the variations in unusual chemical and isotopic compositions of carbonate-cemented surface rocks over oil and gas fields (Donovan and Danziel, 1977; Donovan, 1974). Stroganov (1969) has confirmed that the deeper distribution of hydrocarbons only rarely yields a halo pattern, suggesting the halos have a near-surface origin. Matthews (1985) suggested that diagenetic blockage related to hydrocarbon emplacement may originate at intermediate depths and then be exhumed by erosional processes.
Although direct anomalies and halos have conflicting explanations, both appear to be valid. Indeed, the controversy is significant only if it is assumed that lateral displacement has not occurred during subsurface leakage. This is certainly is a valid assumption in some, but definitely not all, cases. If the halo pattern is interpreted as a subset of several preferential pathways, one can assume that at least one major flowpath could become blocked by diagenetic cement, resulting in a bias of leakage, with a false halo forming as the gases are diverted around this blockage in an area that previously yielded a direct anomaly. In one study the occurrence of halos was suggested by adsorbed soil gas samples, whilst direct anomalies were observed using free soil gas samples (Richers et al., 1986). One must speculate that these techniques measure different aspects of the leakage phenomena. For this reason, it is felt prudent to always collect both types of samples whenever economically feasible. In addition one would be well advised to incorporate geological and geophysical data into the model.
A significant portion of near-surface hydrocarbon survey results appear to be compatible with the mechanisms of macroseepage, particularly leakage occurring along preferential pathways. Those anomalies seemingly not coincident with known faults, fractures, unconformities, bedding planes or other obvious pathways may lie on pathways unrecognized due to limited or incorrect mapping. Alternatively, some occurrences may represent processes not completely understood, or processes not validly extrapolated from macroseepage to microseepage.
The preferential pathway model summarizes the movement of hydrocarbon fluids through the subsurface to their final destination as a surface seep, either directly or by way of an intermediate trap. It is certainly not definitive nor complete, but illustrates some of the challenges confronting the petroleum geologist in his quest for new resources.

Geochemical populations

An alternative to modeling hydrocarbon gas migration as a basis for data interpretation is to decompose data into geochemical populations. On this basis surface geochemical data can be interpreted with respect to both composition and magnitude.
The goal of compositional analyzes is to be able to characterize the type or types of subsurface accumulations present and to be able to predict the location at which they occur. This can be achieved through using ratios of the various hydrocarbon constituents that are detected in the soil gas sample. In general, gas reservoirs are commonly dominated by the presence of methane, whereas oil reservoirs usually contain additional quantities of hydrocarbon gases heavier than methane (Nikonov, 1971).
There are three potential origins for gases detected in the near-surface environment: biogenic, thermogenic (or katogenic) and igneous (including mantle degassing); and irrespective of the origin, the gases tend to migrate towards the surface due to pressure and buoyancy effects. Gases from several sources may mix or undergo other compositional changes such as chromatographic separation during this migration. Thus the measured compositions may not always reflect the original subsurface composition. In most areas mixing presents little problem because gases of thermogenic origin are by far the most abundant. Furthermore, the tendency for gases of biogenic and igneous origin to be extremely dry and of a different isotopic composition from thermogenic gases enables recognition of their presence. Extreme chromatographic separation may only be recognized by careful isotopic analysis and through the close comparison of near-surface gas with known reservoir gas in the region. The presence of gas of igneous origin generally indicates the occurrence of deep, pervasive faulting, and/or the presence of igneous activity in the area. This association, as well as the extremely methane-rich character of such gases, allows for the facile distinction between gases from thermogenic and igneous sources.
Telegyna and Cherkinskaya (1971) found that the olefin content of soil gases decreased relative to saturated hydrocarbons until depths of about 300 meters. Experimentally, as illustrated in Table 5-XIII, olefins can be formed from saturated compounds in areas of low oxygen content (0.5% to 3.2 %). The presence of these olefins may be biogenic (Smith and Ellis, 1963), although Starobinetz (1976) showed a linear relationship between the concentrations of saturated and unsaturated gases derived from the thermogenic alteration of organic matter. Sokolov (1971 b), among others, suggested a relationship between the generation of unsaturated compounds and drilling activity.
Gleezen (1985) showed that there is promise in using the olefin contents of soil gases as a scaling factor to separate seep signals from ambient signals. He was able to define areas with signatures similar to those of the reservoired gases. It would appear that in some cases the presence of olefins may merely represent the breakdown of saturated hydrocarbons by some yet- undetermined process during the migration of gases to the surface and/or some activity such as biogenic degradation of the saturates in the near-surface environment (Telegyna, 1971).
Compositional information in soil gases has been related to subsurface accumulations through the application of specific ratios (Jones and Drozd, 1983). Methane-dependent ratios (Table 5-VII) are reliable unless multiple sources of gas are present in the area. An independent methane-rich source biases an oilier composition toward a drier gas composition. This can sometimes be overcome by plotting histograms of the compositional data and noting multiple populations in the data. Another set of diagnostic ratios that are not methane dependent has also been defined and further aid in properly defining the true potential of an area (Drozd et al., 1981; Williams et al., 1981). In general, the agreement between the surface compositions with reservoir compositions is the strongest evidence that surface prospecting can accurately define the potential of an area.
In addition to compositional information, soil gas data can yield useful information according to the presence or absence of anomalously high magnitudes. To understand the concept of anomal