Kenneth R. Sundberg
Phillips Petroleum Co.
Bartlesville, Okla.
Geochemical exploration presumes that oil or gas reservoirs leak petroleum to the surface, and that these seeping hydrocarbons can be related to possible reservoirs in the subsurface. As an exploration technique, surface geochemistry assumes neither that every reservoir actively leaks and will be expressed geochemically nor that every geochemical anomaly is associated with a commercial reservoir. It does assume that seepage is common enough to be useful.
Over the last several years, Phillips Petroleum Co. has executed geochemical exploration projects in its worldwide exploration program (Fig. 1). Some uses, in North America, are prospect evaluation. Others, in Egypt for instance, were part of work commitments associated with opening and operating new exploration concessions. Some are part of technical agreements with international partners and research organizations.
In fact, the use of seeps in hydrocarbon exploration is widely accepted and practiced throughout the industry. Independents and majors all use a variety of techniques aimed at seep detection and characterization. Particular methods vary, but the general objectives of the various surveyors are about the same:
- locate hydrocarbon seeps,
- map the seeps to relate them to subsurface prospects,
- characterize the petroleum type seen in a play's seeps,
- refine economic evaluations before entering new plays, and
- aid explorationists in making lease relinquishments.
In the following, we will illustrate geochemical applications to these problems.
SCIENTIFIC OBJECTIVES
MICROSEEPAGE-SURVEY METHODS
Scientifically, our survey objectives are to detect and map seeps and to relate them to prospects.
Assuming hydrocarbons seeping from a reservoir, through a caprock, find their way to the surface, some analytical method is used to detect them, and they are plotted as linear profiles or mapped in two dimensions. At Phillips we have conducted research in and actively used four methods:
- Light hydrocarbon survey - detecting hydrocarbons themselves. Methane to pentane are analytical objectives, and methods include headspace analyses and fixed phase measurements like the Horvitz (1985) acid digestion.
- Measurements of helium and oil associated gases, which can be present in elevated concentrations over petroleum reservoirs that are charged with helium or other associated gas. This charge is common enough to make the method useful.
- Measurements of microbial blooms of light gas consuming microorganisms - an indirect effect. These organisms exploit light hydrocarbon gases as a food source.
- Detection of general secondary seepage effects via remote sensing--primarily spectral shifts associated with biogeochemical stress.
Other methods are used from time-to-time, such as fluorescence, magnetic susceptibility, gamma-ray data, and Curie point pyrolysis of seeping gases adsorbed on an activated surface. However, the numbered methods were specific research objectives, and they will be the major focus of this report.
Strong and weak seepage features may appear near a seep. The near surface controls on seepage expression may be quite strong. Abrupt changes in the environment can be important. Consequently, the detection and location of a seep often is a statistical exercise. We have released a small collection of papers on our experience with the various methods (Hughes and Holba, 1987a, b; Beghtel, et al., 1987; Garcia et al., 1988; Henry, 1989; Sundberg, 1990; Sundberg and Tiedemann, 1990; Walters and Sundberg, 1992; Ackerman and Boatwright, 1993).
MACROSEEPAGE-CHARACTERIZATION
Macroseeps, where petroleum seepage is intense enough to allow a macroscopic quantity of hydrocarbon to be obtained, offer a unique opportunity to characterize one's exploration target prior to drilling.
Analytical techniques include fluorescence, gas chromatography (GC), gas chromatography/mass spectrometry (GC/MS), and pyrolysis/GC.
The seeps are typically biodegraded and water washed. This severely limits the usefulness of standard chromatographic methods. it also degrades GC/MS capability. Only the most resistant chemical fossils survive in the seep environment. Nevertheless, statistical relations among biomarker ratios obtained from crude oils and well cuttings extracts sometimes can be applied to seep samples. The benefits obtained range from prospect upgrades to opening new plays and condemning areas as noneconomic (Ackerman and Boatwright, 1993).
INTERPRETATION METHODS
STATISTICS-SEEP PRESENCE
Surface geochemical interpretation to locate seeps in principle is a relatively simple undertaking.
One measures one or two seepage-indicating parameters and plots or maps them over areas of interest. If more than one method is used, one looks for coincident indicators. However, often the geochemical indicators are not very strong; then confidence is weak, and complications become evident. A statistical view is appropriate.
In a statistical approach, one implicitly recognizes that: (1) hydrocarbon seepage is a basically weak phenomenon, (2) many nonseepage-related phenomena can interfere with it, (3) most of these phenomena are either uncontrollable, unknown, or both, and (4) sampling and interpretation procedures should attempt to deal with these problems.
One answer to these problems is to oversample enough to allow statistical averages (signal average) to help smooth over the unknowns in the data. To get enough samples to benefit from averaging, try to hit every target of interest along a reconnaissance line with four or five samples, or try to hit every target with four or five cells or nodes in a gridded survey.
A smoothed value graphed at some point along a line, or at some point on a gridded surface presented as a contour map, is not a prediction of the value of some geochemical indicator one would expect to observe if one sampled that point. Rather, it is a statistical summary of the distribution of values of the geochemical indicators in an area around that point. If that distribution is anomalous, then a geochemical anomaly exists at that point.
By way of illustration, Fig. 2a shows histograms for two Microbial Oil Survey Technique (MOST) tests of areas around new field wildcat wells (Beghtel et al., 1987). The frequency is the number of samples from the sample suite (64 in these cases) with counts in each interval along the horizontal axis.
Fig. 2b shows the survey sample pattern used in the tests. Wildcat wells eventually completed as producers had MOST histograms typical of the lower graph in Fig. 2a. This seepage prone distribution is multimodal and has an identifiable population of samples with an unusual or high number of microbe counts. Dry holes show histogram profiles typical of the upper graph. They contain an anomalous population of positive indicators set against a bland background of relatively low values.
In test applications, distinctions of this type have been used to highgrade new field wildcat prospects. In one double blind study in Kansas, the geochemical prediction success rate more than doubled the independent, post survey production results of 34% commercial completions. Table 1 summarizes the results of such predictions on a collection of new field wildcat wells surveyed and interpreted prior to their being drilled.
Fig. 2c illustrates these results in map form. The success rate demonstrates geochemical survey effectiveness in highgrading prospects, and it also provides a validation of statistical procedures in geochemical interpretation. Although a statistical analysis of these data shows the results cannot practically be due to random chance, see Beghtel et al., one must exercise care in certain environments, like evaporate basins. This problem is discussed in detail elsewhere.
STATISTICS OF LINE PLOTS, GRIDDED DATA
The relatively dense sampling along the lines in Fig. 2b provides the sampling redundancy that allows a statistical interpretation to be made. One simple procedure is to plot the data as a moving average. In the wildcat study, features were generally on the scale of 1/4 section, so with a sampling density of 16 points/mile, a five point moving average would be appropriate. We have often used the formula
Averagej = (1 Cntj-2 + 2 Cntj-1 + 3 Cntj + 2 Cntj+1 + Cntj+2)/9
to calculate the weighted average at point j. Here Cntj is the microbe count observed at point j. This function mimics the histogram interpretation to some extent. It emphasizes samples close to a given point; high samples will be highly weighted when they are nearby, but they will still receive emphasis if they are some distance away. The number of points in the formula and the exact weights can be adjusted for special circumstances in any given survey.
Interpretation of gridded data is similar. Indeed, one could construct histograms for the grid cells in a suitably cellularized area and interpret them visually. However, typical least squares gridding algorithms - like those in many programs -really automate the procedure.
First, the programs use a search radius to gather control point data surrounding a grid node. This is analogous to collecting samples from around a well as in Figs. 2a and 2b.
Second, the programs use these data to calculate a least squares estimate of the control point data at the grid node. For highly statistical data, like a geochemical survey, this estimate is really a mathematical restatement of the arguments made to differentiate between the wildcat and dry hole histograms above. This is because least squares procedures favor outliers in making their estimates. This tendency is annoying in some circumstances, but it makes least squares estimates useful in geochemical mapping.
A simple enhancement and convenience is available. The geochemical indicators can be converted to ranks. That is, the concentrations or other anomaly indicators are rank ordered from smallest to largest, and each sample is assigned a fractional or percentile rank. For instance, if an ethane concentration in one sample is larger than 75% of the samples in the rest of the survey, it is assigned a rank of 0.75. This presentation stresses variations about the survey mean, and it tends to highlight small anomalies that might otherwise be overlooked.
Fig. 3a shows such a presentation of survey results from Roosevelt County, N.M. The lines forming the grid are densely sampled. The gridding is done with a mapping grid cell density about the same as the natural 1 mile grids seen in the survey lines, and a smooth statistical weight function is used in the gridding process.
The MOST rank anomalies were generated with a least squares algorithm in one of the conventional mapping programs. The contour map should not be regarded as a prediction of MOST counts one could expect in future surveys. Rather they are statistical statements. The area around each mapped MOST anomaly contains a few unusually high count samples, and a frequency histogram of the samples from this area would have the appearance of the distribution in the bottom Fig. 2a. The distributions for the background areas will appear more like the distribution in the upper half of Fig. 2a.
Similar arguments can be made for other prospecting methods.
CHEMICAL PROPERTIES-PETROLEUM PROPERTIES
Since surface geochemistry analyzes emplacements of actual subsurface petroleum, it is natural to attempt to use these analyses to characterize the petroleum one would expect to find on drilling. Matthews et al. (1984) suggested that surface light gas compositions actually reflect the compositions seen in the subsurface. For example, ratios like ethane/propane at the surface are close to those seen in the subsurface.
Broad petroleum classifications, like those of Nikanov (1971), Fig. 4a, can cautiously be applied to the seeping gas, and the surface hydrocarbon data could be used to loosely type an area as oil or gas prone. Compositionally, most oil fields have a C2/C3 ratio of near 2, and most gas fields have a C2/C3 ratio greater than 3. Though subject to variations in petroleum source and maturity, this simple index is still useful. This is particularly so since the light hydrocarbons that seep to the surface often do so without substantial changes in composition.
Internally, Phillips light hydrocarbon surveys have tended to confirm this idea. Table 2 shows the light hydrocarbon concentrations seen in Horvitz acid extraction analyses of samples from the Albion Scipio Trend in Michigan and the Patrick Draw area in Wyoming.
Clearly, the surface and subsurface ratios are very similar. The ratios reflect the petroleum type one would expect from Nikanov's C2/C3 ratio curves. Patrick Draw is a condensate field, with a ratio of 2.3. Albion Scipio is an oil field with a ratio of 1.4.
When sufficient sample is available, organic geochemical GC/MS analyses can confirm and supplement these data. Fig. 5 shows a light hydrocarbon seepage profile overlain by values of extracted sample biomarker ratios. The hydrocarbon seeps appear as highs in a map of the product of the propane and butane ranks, and the maturity maxima appear as lows in the biomarker ratio ranks. Clearly, the high seepage indicators coincide with the high maturity biomarker ratios.
The coincidence indicates the light hydrocarbon anomalies are indeed expressions from the subsurface, and hydrocarbons associated with them can be geochemically distinguished from hydrocarbon components of recent sediments. The sample locations are approximately 2 km apart. The area is a frontier province, and these curves provide management with additional confidence that this particular feature and others like it are actually seepage related. Although recent sediment compounds dominate the biomarker analyses, the higher relative concentrations of petroleum related, thermal hydrocarbons do appear along with the light hydrocarbon high concentrations. Both the ethane and propane light hydrocarbons and the sterane and hopanes in the biomarker ratios are relatively petroleum specific. They allow one to see through the recent sediment haze.
Sometimes these analyses can be extended to make useful economic judgments about oil properties. Hughes and Holba (1987) developed a relation between biomarker ratios and oil bulk properties. Although commonly applied to well cuttings extracts, the method can be applied to surface sediment samples. Fig. 6a shows a distribution of predicted API gravity in the Santa Barbara Channel. Seeps in this area are very intense, and the samples are of a quality to allow detailed characterizations to be made. Engineering cores were solvent extracted, analyzed, and the GC/MS data interpreted to produce the map. Santa Barbara, being a heavy oil province, could be preferentially developed using predictions of oil quality. Fig. 6b shows the Holba-Hughes relation used to create the map in Fig. 6a.
EXPLORATION APPLICATIONS
PROSPECT EVALUATION
Mt. Pearl field, eastern Colorado. Mt. Pearl oil field was developed from about 1982-83 on. Prior to the major development period, Phillips ran a Microbial Oil Survey Technique study of the area. Results of the study, its contemporaneous production, and the subsequently developed production are seen in Fig. 7.
The general Mt. Pearl trend is clear in the data, and several other leads are suggested. The line indicating the channel course was developed from seismic data. The anomalies were mapped in 1983, prior to the major field development.
While the data are very suggestive, it is important to note a few problems:
- The field would not be visible in the data were the survey confined to the seismic lines. The off-seismic data are essential to the interpretation, and 2D presentation is very important.
- The seismic tie is a geochemical low, and the geochemical low is tied by two geochemical lines. Many factors can influence geochemistry at a single point. It is important to survey the general area around any target.
- In this area, local evaporates are a problem, and it could well be the geochemical signal at the seismic tie is blocked.
Nevertheless, the Mt. Pearl work does indicate potential for geochemical data to screen prospects and boost exploration productivity.
MODEL VALIDATION
New Mexico Fusselman. Fig. 3a shows an exploration model of an area of Peterson field in eastern New Mexico.
Basically, the play is thought to be controlled by porosity developed on the Fusselman top where it was exposed in a major unconformity. Reservoirs are found along the flanks of a basement high where uplift of the pinchout against the basement high creates the reservoir potential. Reservoirs should develop in a general ring around the high. The geometry of the basement feature will control their structure.
A second play might be in the Abo Reef trend. This trend might occur above the structure as seen in the illustration of Fig. 3b.
A microbe survey generally indicates hydrocarbon seepage along the flanks of the basement structure, Fig. 3a. This is in general agreement with the unconformity controlled porosity and would downgrade the Abo Reef trend as a primary target in this area. As noted in the foregoing, the gridding is set up to mimic the statistical interpretations used in our published wildcat studies.
PLAY DEFINITION
Western Desert of Egypt. Light hydrocarbon data were collected over a large concession in the South Umbarka area of the Western Desert of Egypt. This survey included control lines over production and surveys of areas regarded as prospective and rank wildcat areas. Fig. 8a shows the anomalous propane distribution in the area and some large basement structures.
Clearly, the seepage is confined to a relatively well defined trend through the west central and north parts of the concession. Seepage seems to be fault controlled, both in the areas of established production and in the areas of relatively low seepage intensity. The major trend follows the western edge of a somewhat conjectural structural/stratigraphic trend called the Fagur swell.
Fig. 8b shows a generalized cross section from southeast to northwest through the study area. Multiple sources are present, and the depositional system would place likely reservoir candidates in a broad trend up the west central part of the block and across through the Umbarka area and on to the east. It is fluvial in the southeast trending to marine toward the north and northwest. Exploration in the east, central, and far western areas of the block confirmed this general trend.
RELINQUISHMENTS
In concessions like South Umbarka, operators typically must relinquish large portions of the tract after a specified time.
Guides to these relinquishments are typically projections made from the geological, geophysical, and normally modest production data. Geochemical maps are a simple adjunct to these considerations.
Fig. 8a shows large areas that are relatively barren of hydrocarbon seepage. The general southeast of the South Umbarka is geochemically quite barren. Though a concession manager might not base relinquishment decisions on geochemical data, some comfort could be taken from data like those in Fig. 8a. Phillips first relinquishment in this concession was consistent with these data.
PETROLEUM TYPE IN CAMEROON
When macroseeps are located, by accident or by reconnaissance survey, a much richer suit of geochemical analyses can be brought to bear on them.
Phillips explorationists located several such seeps in Cameroon, and these seeps were subjected to a full organic geochemical characterization (Ackerman and Boatwright, 1993). Fig. 9a locates these seeps.
This area normally has been classified as a gas prone area, and it has been largely neglected by explorationists. However, the seep geochemistry has suggested there may indeed be oil prone source rocks in this area, and this observation was part of a Phillips decision to acquire acreage in Cameroon and to begin exploration there.
Fig. 9b is a ternary diagram of the gas prone light alkanes (11 to 14 carbons), the waxy alkanes (25 to 28 carbons), and the aromatic indicators (benzene, toluene, and xylene). The Cameroon seeps plot toward the center of the light oil side of the diagram and trend toward the waxy side of the diagram.
Only two samples show definite gas tendencies. These are the Logbaba seep, from a long established gas province, and the N'Tota seep, found near a newly opened gas province.
Typically, seeps are badly biodegraded and water washed. The Cameroon seeps were no exception. Nevertheless, analytical data can be obtained from these materials. Pyrolysis/GC overcomes many of the sample damage problems. It thermally isolates the heaviest, most polar, most insoluble, chemically undisturbed portion of the petroleum and breaks it down into light fractions that can be analyzed via conventional GC methods. Fig. 9c compares a conventional GC and a pyrolysis/GC trace of one of these oils.
INTERNATIONAL WORK COMMITMENTS
Geochemical exploration methods are not new, and they have always enjoyed a varied degree of support within the exploration community.
Phillips typically runs some kind of geochemical survey as part of its international exploration work. We have also written these geochemical programs into a number of our work options and international teaming agreements.
The acceptance is broad enough that some of our lessors and partners expect such work to be done.
ACKNOWLEDGMENTS
The author is grateful to his colleagues. In particular he thanks D.C. Boatwright for his diligence in studying the organic geochemistry of surface samples, and he notes a debt to the earlier work of his colleagues and former coworkers Dr. A.G. Holba and Dr. W.B. Hughes. The author is also deeply grateful to Phillips Petroleum Co. for permission to submit the paper.
BIBLIOGRAPHY
Beghtel, F.W., Hitzman, D.O., and Sundberg, K.R., Microbial Oil Survey Technique (MOST) evaluation of new field wildcat wells in Kansas; APGE Bull., Vol. 3,1987, pp. 1-14.
Garcia, R., Deibis, S., and Sundberg, K.R., Light hydrocarbon survey data and accumulation in the Umbarka-South Umbarka area, Western Desert, Egypt; Proceedings 9th Petroleum Exploration Conference, Cairo, Egypt, 1988.
Horvitz, L., Geochemical exploration for petroleum, Science, Vol. 229, 1985, pp. 821-827.
Hughes, W.B., and Holba, A.G., Relationship between crude oil quality and biomarker Patterns, Advan. Org. Geochem., Vol. 13, Nos. 1-3, 1987a, pp. 15-30.
Hughes, W.B., and Holba, A.G., Relationship between crude oil quality and biomarker Patterns in samples from the Point Arguello area, California; corporate report, 1987b.
Matthews, M.D., Jones, V.T., and Richers, D.M., Remote sensing and hydrocarbon leakage; Int. Symp. Remote Sens. of Environment, Colorado Springs, Colo., 1994.
Nikanov, V.F., Distribution of methane homologs in gas and oil fields; Akad. Nauk SSR, Doklady, Vol. 206, 1971, pp. 234-236.
Richers, D.M., Reed, R.J., Horstman, K.C., Michels, G.D., Baker, R.N., Lundell, L., and Mars, R.W., Landsat and soil-gas geochemical study of Patrick Draw oil field, Sweetwater County, Wyo.; AAPG Bull., Vol. 66, No. 7, 1982, pp. 903-920.
Sundberg, K.R., Multispectral imagery (Landsat) hydrocarbon alteration signature: Definition of the signature based on studies of probable hydrocarbon microseepage in the U.S. MidContinent; APGE Bull., Vol. 6, 1990, pp. 1229.
Sundberg, K.R., and Tiedemann, H.A., Multispectral imagery (Landsat) hydrocarbon alteration signature: Prospect leads in the Haswell-Kit Carson area of Eastern Colorado; APGE Bull., Vol. 6, 1990, pp. 30-48.
Walters, J.P., and Sundberg, K.R., Soil-gas helium surveys for petroleum exploration in Kansas, APGE Bull., Vol. 8, No. 1, 1992, pp. 55-63.
Copyright 1994 Oil & Gas Journal. All Rights Reserved.
