The following is the result of my decision to publish my thesis on the World Wide Web. I apologize in advance for the many inconsistencies, omitted references, and coding errors which may be evident here. This page should be considered as under constuction; in particular I have not included any inline graphics, and not all cited graphics have links yet. I have not spent any significant amount of time updating or pursuing many important aspects of this research in the last six years, due to my preoccupation with gainful employment during that time. Much more interesting work could be done on this topic. Some additional information can be found here: Milankovitch Cycles in Paleoclimate.
Please feel free to send me any questions, comments, or suggestions you may have. I am planning on adding more links and additional graphics, etc., as time permits. If you are interested in receiving data or software, drop me a line. I understand that a great deal of Bureau of Mines and USGS data has been put on line by Jack Dyni of the USGS in Denver, but I do not have an address. Please let me know of any other information or resources that you are aware of.
The Green River Formation (Paleocene-Eocene) of the Uinta Basin, northeastern Utah, and the Piceance Creek Basin, northwestern Colorado, may exhibit cyclicity in lacustrine facies. Cyclicity is described in outcrops and in oil-shale yield curves by various authors. Large-scale (> 1 m) cyclicity in the Green River Formation and other rocks has been attributed to variations in the Earth's orbital geometry over time periods of roughly 20,000 to 100,000 years, referred to as "Milankovitch" cyclicity after the Serbian mathematician Milutin Milankovitch, who postulated an orbital forcing mechanism for Pleistocene glaciations in the early 1900’s. No sedimentologic data from the Green River Formation have previously been mathematically tested for long-period cyclicity. Documentation of orbitally-induced (Milankovitch) cyclicity in stratigraphic sequences is useful in establishing depositional rates and in documenting climatic control of biologic, sedimentologic, and hydrologic factors in lacustrine deposition. Detailed analyses may allow close correlation of terrestrial stratigraphic sequences with global climatic fluctuations and marine stratigraphy, and may possibly elucidate characteristics of the Earth's climate system and orbital behavior in the past.
Sonic log and oil shale yield data from three oil wells and four coreholes that penetrate the Parachute Creek Member of the Green River Formation in the Uinta and Piceance Creek basins were analyzed for cyclicity. Three methods of spectral estimation were used to test for the existence and significance of sedimentary cycles, and to determine the spatial and implied temporal frequencies of variation in rock properties. A method of accounting for variable depositional rates, based on depositional modelling and varve thicknesses, was developed and tested. It is useful in improving recognition of orbitally-induced climatic fluctuations recorded in Green River Formation lacustrine strata.
The consistent appearance of three dominant low-frequency spectral peaks with frequency ratios close to those of modern orbital parameters implies that the Lake Uinta depositional system was strongly influenced by orbitally-induced climatic fluctuations. The spatial frequencies of spectral peaks interpreted to represent the Milankovitch orbital frequencies of roughly 20, 40, and 100 kyr imply mean depositional rates of 20 to 55 cm/kyr in studied sections, which correlates well with depositional rates calculated from varve thicknesses in the Green River Formation. The depositional model used in this study assumes that the depositional rate fluctuated by a factor of approximately 10 in proportion to hydrocarbon content in the most hydrocarbon-rich sections, and is found in some cases to improve agreement of calculated spectra with cyclicity predicted by the Milankovitch orbital theory of climatic change.
Numeric data from several stratigraphic sections in the Green River Formation were analyzed to detect significant long period cyclicity, and if possible, to correlate cyclicity to orbitally forced climatic fluctuations. This is accomplished by treating data from well logs and core analyses as time series, and by using the statistical methods of spectral analysis to determine the significance and spatial extent of cyclicity. Results from these analyses were compared to independently determined values for depositional rates in the studied intervals and similar rocks. Such a comparison shows that the cyclicity in stratigraphic sequences within the Green River Formation can be attributed with moderate confidence to variations in Earth orbital parameters of precession, obliquity, and eccentricity. A time interval correction technique, devised to compensate for the effects of variable depositional rates documented by previous studies, was applied to test the validity of a depositional model. Synthetic data were used to test spectral estimation methods on data with known strengths and frequencies of cyclical components, and to test the effects of various amounts of noise in the dependent variable (lithologic properties) and varying amounts of time distortion in the independent variable (time or depth) on the results of spectral analysis.
Variations in the geometry of the Earth's orbit have been postulated to affect the behavior of the atmosphere and oceans since the late nineteenth century (Croll, 1875). The astronomical theory of climatic change was more fully developed by Milankovitch (1941), who theorized that the orbitally-derived changes in the relative amount of solar radiation received by high latitude continental areas in the northern hemisphere during summer months was the driving force behind major Pleistocene glaciations. Solar radiation varies by several percent over periods of thousands of years due to changes in Earth's orbit (Berger, 1978). Refinement in the understanding of the precise nature of orbital mechanics and resultant effects on climate during the Quaternary was achieved by Berger (1976, 1977b) and other workers (Imbrie and Imbrie, 1980; Imbrie, Hays, Martinson, McIntyre, Mix, Morley, Psias, Prell, and Shackleton, 1984).
The Earth's climate over the last several million years has been shown to be strongly controlled by orbital variations through studies of oxygen isotope and CO2 data from marine sediments and ice cores (Hays, Imbrie, and Shackleton, 1976; Imbrie and Imbrie, 1980; Imbrie and others, 1984). Significant effects of orbital variations have been documented in rocks formed in depositional systems sensitive to climatic conditions, such as deep ocean basins, lacustrine basins, and glaciers (Bradley, 1929; Van Houten, 1964; Duff, Hallam and Walton, 1967; Hays and others, 1976; Fischer, 1982; Imbrie and others, 1984; Herbert and Fischer 1986, Kerr, 1987).
The present values for the periods of precession, obliquity and eccentricity have been shown to fit various Cenozoic and Mesozoic cyclic sedimentary sequences reasonably well. Milankovitch cyclicity was documented in Cretaceous black shale sequences from Italy (Park and Herbert, 1987) and in Triassic lacustrine deposits in New York (Van Houten, 1964; Olsen, 1984). Sedimentary cyclicity in late Tertiary and Pleistocene marine sequences has been shown to be controlled by large-scale climatic variations, in turn caused by variations in the Earth's orbital geometry over time. The biostratigraphic and isotopic age control in the data from Plio-Pleistocene marine sediment and ice cores is much better than for older stratigraphic sequences, and has allowed very detailed correlation of changes in orbital parameters with changes in sea surface temperature through the d18O proxy indicator and with atmospheric CO2 (Emiliani, 1963; Prentice and Matthews, 1988). These orbital variations now have well-documented frequencies and effects (Berger, 1976, 1977a, 1977b, 1978; Hays and others, 1976; Imbrie and Imbrie, 1980; Imbrie and others, 1984). Milankovitch cyclicity has been the subject of considerable research in the last two decades. Much modern research into cyclicity has been focused on documentation of climatic fluctuations, represented by variations in d18O in Plio-Pleistocene deep-sea deposits and ice cores. Milankovitch cyclicity is now widely accepted to be a principal cause of glacial-interglacial changes during the Pleistocene (Imbrie and others, 1984). One of the goals of current research into Milankovitch cyclicity is to extend recognition of cyclicity in older depositional sequences, which may allow more accurate resolution of time intervals represented by sedimentary sequences than is possible with other means (House, 1985). It may be possible to provide a better understanding of climatic and depositional relationships through geologic time. This type of research has been termed "cyclostratigraphy" in the recent Cretaceous Resources, Events and Rhythms conference (Fischer and others, 1988). Such an approach may provide a means of quantifying depositional response to climate (Perlmutter and Matthews, 1989), and has become increasingly feasible in the last twenty years due to the large amounts of data generated by energy exploration and the rapid development of instrumentation and digital processing technology (Fischer, 1986).
The Green River Formation (Paleocene-Eocene), of northeast Utah, northwest Colorado, and southwest Wyoming, was deposited in a large lacustrine depositional system in a series of ponded intermontane basins. These developed during a period of transition from the Cretaceous Sevier orogeny to the Laramide tectonism of latest Cretaceous to late Eocene time (McDonald, 1972; Dickinson and others, 1988; Franczyk and others, 1989). The sections studied in this paper are from the Uinta Basin in northeast Utah and the Piceance Creek Basin in northwest Colorado (Figure 1). Estimates of the lacustrine system's duration range from about 13.3 Myr (Picard, 1963) to about 23 Myr (Dickinson and others, 1988). Later estimates take into account more complete subsurface data and improved radiometric age control. The principal kerogen-rich "oil shale" units of the Parachute Creek Member of the Green River Formation were deposited during an interval of 2.3 million years during the middle Eocene (Picard, 1963), when the lacustrine system was relatively stable, at or near its highstand, and conditions were favorable for the accumulation of algal kerogen.
The Green River Formation is one of the more thoroughly studied formations in the world due to the tremendous hydrocarbon reserves contained in the kerogenous Parachute Creek Member. The latest edition of the USGS bibliography on the formation contains over 1,300 references (Smith, 1986). Numerous US Bureau of Mines and Geologic Survey studies have documented the hydrocarbon reserves present in the Green River Formation (Stanfield, Rose, McAuley, and Tesch, 1957; Stanfield, Smith, Smith, and Robb, 1960; Stanfield, Smith, and Trudell, 1967).
The Uinta Basin occupies a large portion of northeastern Utah (Figure 2). Localized deposition in the Uinta Basin began in the late Cretaceous at about the Campanian-Maastrictian boundary (73 Ma), and lacustrine strata were deposited beginning in the middle Late Paleocene to the middle Late Eocene, from approximately 61 to 38 Ma (Dickinson and others, 1988). The Uinta Basin was bounded by the waning Sevier orogenic belt to the west, the emergent Uinta Mountains to the north, and the Uncompahgre Uplift to the south and east (Hintze, 1980; 1988). The Uinta Basin is an asymmetric depositional and structural basin, with the depositional axis running approximately east-west, nearer to the northern edge of the basin, which is bounded on the north by the Uinta Mountain uplift (Figure 2). Uplift of the Uinta Mountains was active during Green River Formation deposition (Hintze, 1988; Franczyk and others, 1989). The greatest thickness of oil shale occurs near the Utah-Colorado border in T. 10 S., R. 24 E., roughly 40 km (25 miles) east-southeast of Ouray, Utah (Smith, Trudell, and Robb, 1972).
The Piceance Creek Basin is a structural and stratigraphic basin located in northwestern Colorado. Lacustrine deposition began later in the Piceance Creek basin than in the Uinta Basin (Picard, pers. comm.). It is considerably smaller than the Uinta Basin, and is more symmetrical. The depocenter is in the north-central portion of the basin, and its major axis trends approximately northwest-southeast (Figure 3). The Piceance Creek Basin contains considerably thicker and higher-grade sections of oil shale than does the Uinta Basin. Sections of oil shale yielding 25 gallons per ton (10% recoverable hydrocarbons) are up to 1900 ft (580 m) thick in the center of the basin (Stanfield and others, 1960; Donnell, 1961). The Piceance Creek Basin was bounded by the White River uplift to the northeast, the Uncompaghre uplift to southwest, and the eastern extension of the Uinta arch to the north (Figure3). The Douglas arch was a subdued positive feature during Green River deposition. Deposition was initially separated from that in the Uinta Basin by the Douglas Arch, but became continuous with that in the Uinta Basin during most of Green River Formation deposition (Dickinson and others, 1988).
The Green River Formation includes open lacustrine, marginal lacustrine, deltaic, and fluvial facies. Lithofacies include shale, calcareous shale, mud supported carbonate, grain supported carbonate, siltstone and sandstone (Fouch, 1981). The marginal lacustrine facies within the formation fluctuated laterally in response to changes in lacustrine base level, while sediment supply and tectonic subsidence remained fairly constant (Picard and High, 1972). Variations in sediments deposited in the open lacustrine environment near the depositional center of the basins primarily reflect changes in the chemistry of lake water and biologic productivity (Picard, 1963; Picard and High, 1972). The lake chemistry and hydrologic balance fluctuated due to changes in amount of runoff and evaporation, both of which were controlled by regional or global climatic changes. The Lake Uinta depositional system was large and relatively shallow, with a maximum width on the order of 100 km and depths of up to a maximum reported 300 m in the Piceance Creek Basin (Johnson, 1981), but commonly reported to be much shallower overall (Hintze, 1972; Picard and High, 1972; Picard, 1985). Some authors have proposed a playa-lake model for deposition of Green River Formation rocks (Eugster and Hardie, 1975; Lundell and Surdam, 1975). This model postulates that the paleolakes Uinta and Gosiute (in Wyoming) were subject to periods of desiccation fairly often, and that oil shale deposition could have taken place in very shallow (<10 m) water. Surdam and Stanley (1979) give convincing evidence of cyclic deposition of kerogen-rich and kerogen-poor carbonates, and attribute them to repeated lake level fall and dessication. While it is apparent that the playa-lake model is appropriate for rocks in Wyoming and the uppermost deposits in the the Uinta and Piceance Creek basins, it is now generally agreed that the major oil shale units in the Piceance and Uinta basins were deposited in fairly deep, permanently stratified lake (Desborough, 1978; Picard, 1985; Johnson, 1989).
Several authors discuss cyclicity in the Green River Formation (Gilbert, 1895; Bradley, 1929; Picard and High, 1972; Fischer, 1986; Fischer and others, 1988). Cyclicity in the Green River Formation was initially postulated by G. K. Gilbert (Gilbert, 1895), who also initially proposed an astronomical cause for certain cyclic sequences in the Cretaceous of Colorado, and commented on the cyclic nature of Green River Formation rocks. W. H. Bradley (1929) describes sedimentary cycles with three different periods in the Green River Formation. He found what he interpreted to be annual varves, 11-year average sunspot cycles, and long-period cycles averaging 21,630 years, and postulated an orbitally-generated climatic cause for the long-period cycles (Bradley, 1929). Picard and High (1972) document cyclic intertonguing of fluvial, marginal lacustrine and open lacustrine facies in the Raven's Ridge area. They attributed the cyclicity to climatically-induced fluctuations of several m in the water depth of Lake Uinta. Other authors have postulated that all three types of orbital variations are recorded in cyclic intervals in the formation (Fischer, 1986; Fischer and others, 1988).
Varves in the Green River Formation are described by Bradley (1929). They are interpreted to represent annual layering, indicating deposition in quiet, anoxic bottom waters. Varve sequences within the Parachute Creek Member have been correlated over distances of up to 40 km in the eastern Uinta basin, indicating relatively homogeneous depositional conditions over large areas (Smith and others, 1972). Varves have different mean thicknesses, depending on the rock types in which they are found. For fine-grained sandstone, varves range from 0.6 to 9.6 mm and have an average thickness of 1.16 mm; for "marlstone and related rocks," 0.016 to 0.37 mm, average 0.167 mm; and for oil shale 0.014 to 0.153 mm, average 0.05 mm (Bradley, 1929). Bradley (1929) calibrated the larger scale cyclic packages in the Parachute Creek Member by comparing average varve thickness to the thickness of the larger cyclic intervals. This indicates average depositional rates ranging from 5 to 116 cm/kyr (Figure 4). Some authors have questioned whether the fine laminations in the Green River Formation actually represent annual layering (Eugster and Hardie, 1975; Lundell and Surdam, 1975; Anderson and Dean, 1988; Buchheim and Biaggi, 1988). Buchheim and Biaggi (1988) show that the number of laminations between two ash layers in the Green River Formation in the Fossil Buttes Basin in Wyoming is not constant; they report twice as many laminations in basinward sections as in more marginal locations. Davis (1964) reports that there are typically two major annual algal blooms in lacustrine settings, suggesting that varves may be biannual phenomena in some cases. Therefore, sedimentation rates determined from varve counts should be viewed with some caution.
Anderson (1986) describes varves as an indicator of the type of depositional environment susceptible to longer-period cyclicity. The varve represents the sedimentary response to the high frequency, high amplitude annual cycle, and contains the rock types that typify the end members of the larger scale cyclicity. Bradley gives a weighted average depositional rate of 18 cm/kyr after compaction for Green River Formation rocks. The rocks examined by this study were deposited in open lacustrine, quiet water conditions. Their accumulation rate probably lies within the lower half of the range given by Bradley (1929) for oil shale, marlstone, and shale. The variability in varve thickness for oil shales and related rocks implies that the depositional rate varied by a factor of approximately 10, which is an important consideration for the depositional model described below. Depositional rates reported in modern lacustrine settings range from 100 to 400 cm/kyr of uncompacted sediment, which agree well with depositional rates determined in this study when corrected for compaction (Olsen, 1976).
Larger-scale cycles, with thicknesses on the order of 10 ft (3 m) are also described by Bradley (1929) in outcrops on the edges of the Piceance Creek Basin. He arrives at a mean duration of 21,630 years for these cycles, which agrees well with the precessional orbital period. Based on this information, the mean sedimentation rate in sections studied by Bradley would be about 15 cm/kyr.
Fischer (1986) describes longer period cycles in the Green River Formation, based on visual inspection of oil-shale yield curves. They are manifested as bundles and superbundles of the smaller precessional cycles, and represent the short and long eccentricity periods, respectively. Cyclic packages of interbedded marginal lacustrine, fluvial, and deltaic facies are described in the Raven Ridge outcrops on the northeastern margin of the Uinta Basin (Picard and High, 1972). These imply repeated fluctuations of lake level of magnitudes on the order of 10 to 20 ft (3 to 6 m). Fine grained quartz, feldspars and clay minerals are common minor constituents of open lacustrine rocks and were likely of pelagic or eolian origin (Smith, 1972). Currents were incapable of distributing clastics throughout the lacustrine basin, and medium and coarse clastic sediments were trapped fairly near paleoshorelines (Franczyk and others, 1989). Rapid facies changes as described by Picard and High (1972) and facies relationships as depicted by Fouch (1981) support this interpretation. Clastic marginal lacustrine deposits are reported to grade into chemically precipitated open lacustrine deposits within a few km of inferred paleoshorelines (Franczyk and others, 1989).
Two possibilities have been considered in this study. The first is that biologic productivity, and thus oil-shale yield, was directly influenced by orbitally induced climatic fluctuations. Since this is a minor component of the total amount of sediment, it would imply that sedimentary rates did not fluctuate significantly. This is a simplistic assumption and ignores the varve evidence already discussed, but it corresponds to treating lithologic data plotted against depth as a representative time series, which is the most simple model to test. The second possibility, considered to be at least partially correct for reasons outlined above, is that variations in actual hydrocarbon content of Green River Formation rocks is at least partially due to the effects of changes in the rate of deposition of nonbiologically derived sediment components (carbonate and clastics). An inverse relationship between carbonate and kerogen content has been noted by Bradley (1929) and discussed by de Boer (1982), who explained the phenomenon as resulting from carbonate dissolution due to increased pH below the chemocline, caused by anaerobic decay of organic matter during periods of high productivity.
Rocks with relatively high kerogen content are characterized by varve thicknesses roughly an order of magnitude thinner than those with a low kerogen content (Bradley, 1929). Two possible explanations can be made for this relationship. If the production of algal kerogen was relatively constant, variation in varve thickness and implied depositional rate was due entirely to dilution effects. Alternatively, there may have been a primary biologic response to some combination of lake level, water chemistry, and insolation, as mentioned above. While the first hypothesis agrees with the relationship between varve thickness and hydrocarbon content reported elsewhere (Bradley, 1929), there is also substantial evidence to support the second hypothesis. Algal productivity has been shown to increase when lower concentrations of nutrients are available (Hutchinson, 1957). Increasing lake level is more suitable for algal growth because it provides fresher, less nutrient-rich water. Therefore, a combination of biologic and sedimentologic responses to changing lake level is likely responsible for observed characteristics of open lacustrine Green River Formation rocks.
This evidence led the author to attempt to compensate for known variations in depositional rate by assigning more time to organic-rich zones, and less time to organic-poor zones. In this scenario, one cm of rock containing roughly 30 percent kerogen (approximately the maximum found in the studied sections) represents a much greater time interval than one cm of rock with only 1 percent kerogen. This is analogous to the condensed section as used in marine sequence stratigraphy (Haq, Hardenbol, and Vail, 1987). A procedure for numerically modelling this relation is described and tested below. In this model, the average depositional rate is the same as in the constant rate model, but the depth component of sedimentologic data cannot be considered as linear in time, and depth must be converted to relative time increments in order to perform time series analysis. This assumption is probably an oversimplification of the actual relationship of kerogen content to depth and time, but it allows a relatively simple numerical conversion of depth intervals to relative time intervals. The observed lithologic cyclicity in the Green River Formation is probably due to a combination of varying sediment influx and varying biologic productivity. This correlates with an actual depth-time relationship intermediate to the two end-member possibilities outlined above. The dilution model improves the agreement of power spectra with modern orbital parameters in some cases, while in one case, application of the dilution model decreases agreement.
The stratigraphy of the Tertiary rocks of the Uinta Basin is complex due to the diversity of rock types and rapid facies changes (Picard, 1963; Ryder, Fouch and Elison 1976; Pitman, Fouch and Goldhaber, 1982). Stratigraphic relaionships are summarized in Figure 5. The rocks of the Green River Formation have been subdivided in several different ways by different workers. Early workers named three or four members on the basis of prominent marker horizons (Bradley, 1929; Picard, 1955, 1957, 1963). More recent stratigraphic studies with improved subsurface data have attempted to take intertonguing relationships of alluvial facies into account (Fouch, 1976; Ryder and others, 1976; Pitman and others, 1982). Some authors use dominant rock type as a basis for subdivision (Koesoemadinata, 1970; Ryder and others, 1976; Pitman and others, 1982). Most workers would assign the intervals studied in this paper to the Parachute Creek Member, or simply the "open lacustrine facies." In the Uinta Basin, Ryder and others (1976) and Pitman and others (1982) assign open lacustrine rocks to the Parachute Creek Member, marginal lacustrine facies to the Douglas Creek Member, and the upper hydrocarbon-poor and saline facies, representing the contractional phase of Lake Uinta, to the Evacuation Creek Member. Fluvial and alluvial rocks below the Mahogany Marker are assigned to the Paleocene-Eocene Colton and Wasatch Formations, while younger fluvial and alluvial rocks above the Green River Formation are the late Eocene-Oligocene Uinta Formation in the Uinta Basin (Cashion, 1957).
In the Piceance Creek Basin, the stratigraphy is somewhat less complicated, due to the fact that more of the basin margin facies have been removed by erosion. The Green River Formation there is generally divided into the Douglas Creek, Garden Gulch, Parachute Creek, and Evacuation Creek Members, in order of decreasing age (Bradley, 1929; Stanfield and others, 1960). The most prominent stratigraphic marker is the Mahogany Marker, a distinctive horizon that can be recognized in outcrop and in cores and well logs, and can be found throughout the central and eastern portions of the Uinta Basin and the entire Piceance Creek Basin (Stanfield and others, 1960). It is an analcitized tuff bed with a distinctive signature on geophysical logs, and it forms a prominent ledge in outcrop. The Mahogany marker has been isotopically dated at between 46 and 47 Ma (Mauger, 1977). The organic-rich zones of the Parachute Creek Member typically occur in a section beginning up to 200 ft (60 m) above the Mahogany Marker and extending downward for 200 to 500 ft (60 to 150 m) or more.
The intervals studied in this project are in the predominantly open lacustrine Parachute Creek Member of the Green River Formation. This member generally includes rocks deposited during the period of greatest lateral extent of lacustrine deposition, when Lake Uinta was at or near its highstand. Alternating zones of rich and lean oil shale have been given designations R-1, L-1, etc., from the bottom of the Parachute Creek Member upward, depending on kerogen content. In this scheme, the studied intervals fall into the Mahogany zone (R-7), the "B-groove" (L-6), R-6, and L-5 (Johnson 1988, unpub. data; Franczyk and others, 1989).
Croll (1875) originally suggested that long-term variations in three different parameters of the Earth's orbit had caused Pleistocene climatic fluctuations and glaciations. Subsequently, Milankovitch (1941) proposed a somewhat different, more quantitative model that emphasized the importance of summer insolation at high latitudes in controlling continental icesheet formation. More recently, it has been shown that Milankovitch's theory predicts Pleistocene climatic fluctuations well (Hays and others, 1976; Imbrie and Imbrie, 1980; Berger, 1980). Each component of orbital variation is actually quasiperiodic and composed of several harmonic components (Berger, 1976; Berger, 1977b). The frequencies and relative amplitudes of the various harmonic constituents are known to vary through time, but mathematical descriptions of these variations are only valid for about the last 5 million years; the accuracy of mathematical solutions for orbital parameters is lost beyond this point (Berger, 1977b). The detailed mathematics of orbital mechanics from which the following tables and figures are simplified is very complicated, but the important elements of orbital behavior can be applied to geologic problems fairly easily. Table 1 lists the modern values for the frequencies
| Precession | Obliquity | Eccentricity | ||||
| Rank | Period
(years) |
Relative
Amplitude |
Period
(years) |
Relative
Amplitude |
Period
(years) |
Relative
Amplitude |
| 1 | 23,716 | 1.0 | 41,000 | 1.0 | 412,885 | 1.0 |
| 2 | 22,428 | 0.871 | 39,730 | 0.3481 | 94,945 | 0.7909 |
| 3 | 18,976 | 0.6989 | 53,615 | 0.2555 | 123,297 | 0.6818 |
| 4 | 19,155 | 0.5269 | 40,521 | 0.1682 | 99,590 | 0.6091 |
| 5 | 19,261 | 0.1828 | 28,910 | 0.1267 | 131,248 | 0.5273 |
| Weighted
Average |
21,383 | 41,615 | 110,375 | |||
Weighted average frequencies, and frequencies and amplitudes of the five strongest harmonic components for precession, obliquity, and eccentricity.The periods and relative amplitudes of the five strongest components of each of the three orbital variations at present are given. Periods are given in years, while relative amplitudes are for comparison between harmonics for a single orbital parameter only. The weighted average period given for eccentricity excludes the long eccentricity periodicity of 412,885 years. After Berger (1977b).
The precessional component of the Earth's orbital parameters describes the relationship between the seasons and the perihelion. At present, the Earth is at perihelion (closest to the sun) during the northern hemisphere winter, while roughly 10,500 years ago the opposite was the case (Figure 6). The precessional parameter is described by a number of harmonic components, ranging from 19.0 to 23.7 kyr at present (see Table 1). The weighted average period of precession at present is 21,383 years. The precession of the equinoxes is thought to influence the degree of seasonality in the Earth's climate (Hays and others, 1976). Its effects are opposite in the northern and southern hemispheres. The current situation tends to favor hotter summers and colder winters in the southern hemisphere, and more a equable climate in the northern hemisphere.
The obliquity parameter describes the tilt of the Earth's rotational axis from the orbital plane (Figure 6). The axial tilt varies between 22 and 24.5 degrees. Obliquity variation is presently composed of components with periods ranging from 28.9 to 53.6 kyr, and currently has a weighted average period of 41.6 kyr. This parameter primarily affects the relative amounts of insolation received by high and low latitude regions of the Earth's surface. High obliquity reduces the difference in solar radiation received by high- and low-latitude regions of the Earth, while low obliquity results in more insolation at the equator, and less in the polar regions.
The eccentricity parameter describes the deviation of the Earth's orbit from a circle (Figure 6). It varies from near zero to 0.06. This parameter primarily controls the strength of the effect of changes in precession. When the eccentricity is near zero, the precessional parameter has little influence, while when eccentricity is near its maximum, the effects of precessional changes on climate are amplified. The predominant components of the eccentricity parameter have periods ranging from 94.9 to 412.9 kyr. Eccentricity varies with two main periods: a long eccentricity component of 412.9 kyr, and a short eccentricity component averaging 110.7 kyr. Most of the sedimentary sequences studied here are too short to be able to record the long eccentricity cycle.
The specific mechanism(s) by which orbitally generated changes in the seasonal and latitudinal variations of insolation may have been translated into regional variations in temperature and/or precipitation during the Eocene in western North America are uncertain. The primary mechanism is thought to be caused by the north-south migration of wet and dry climatic belts due to changes in the latitude of the Earth's caloric equator and the expansion and contraction of the tropical, temperate, and polar belts in response to insolation variations (Figure 7; de Boer, 1982; Perlmutter and Matthews, 1989). The effect of this phenomenon is more pronounced in mid-latitudes than in polar or equatorial regions. Thus, the Green River Formation presents a good opportunity to study climatic response to orbital variations in a continental sedimentary sequence. The principal variable affecting Green River Formation sedimentation is lake level, which is controlled by precipitation and evaporation. These are in turn governed by average temperature and circulation patterns that reflect the global climate.
The role of temperature in causing cyclic sedimentation is unknown. If other factors were constant, large fluctuations of mean annual temperature would have an impact on evaporation, thus affecting lake level, and on biologic productivity. Temperature fluctuations could have had a significant direct effect on primary biologic productivity. Sergin (1980) believes that orbitally-induced temperature fluctuations were minimal during the Eocene, at least on a global scale, and the epoch was characterized by a non-oscillatory, ice-free climate. More recent work has suggested that significant continental ice may have existed in Antarctica as early as the middle Eocene, and possibly even in the late Cretaceous (Prentice and Matthews, 1988). It seems more likely that significant lake level fluctuations were caused by net precipitation levels, which have an inherently greater variability. Thus, it is assumed that increasing lake level was responsible for deposition of kerogen-enriched sediments, while decreasing lake level favored deposition of sediments with greater amounts of carbonate precipitates and clastic components.
Data used in this study was obtained from two sources. The first data source was public domain well logs, obtained from the offices of the State of Utah Division of Oil, Gas and Mining in Salt Lake City,Utah. The second data source was oil shale yield information, available in various publications of the US Bureau of Mines. Approximately 3000 ft (915 m) of data from eight wells and coreholes was examined in the course of this research. All of the data used in this study, except for density log measurements, were originally given in English units. Conversion to metric units is given where appropriate.
The choice of data for this study was somewhat arbitrary. There are several thousand wells and coreholes in the two basins which could have beenincluded in this investigation. Data from three wells in the Altamont-Bluebell field in the north-central portion of the Uinta Basin were chosen because of the wide choice and good quality of logs, close well spacing, and corroborating sedimentologic and stratigraphic information for some of the wells. One of these is included on a stratigraphic cross-section published by the USGS (Fouch, 1981), which gives detailed lithologic descriptions and interpreted depositional environments.
A large amount of data on oil shale yields from oil wells and oil shale coreholes in both the Uinta and Piceance Creek Basins is available in reports published by the Bureau of Mines (Stanfield and others, 1957; Stanfield and others, 1960; Stanfield and others, 1963; Stanfield and others, 1967; Smith and others, 1972). Data from three oil-shale coreholes in the Piceance Creek basin and one from the west-central Uinta basin were examined in this study.
Oil-shale data have several desirable properties. First, the data are already in numerical form, eliminating the tedious task of digitizing well logs. Second, the oil shale yield data cover large intervals which include the Mahogany zone and stratigraphically adjacent rich oil-shale beds that represent the maximum areal extent and most stable phase of Lake Uinta, the desired intervals for studying the response of the lacustrine depositional system to orbital variations.
The oil shale yield data also have some limitations. Some wells are sampled at 10 ft intervals, which provides insufficient resolution for the purposes of this study. Other wells had unacceptably large numbers of missing values, or were sampled at irregular intervals. Some of the statistical techniques employed in this research required that the independent variable (time or depth) be equally spaced. There are methods of analyzing unevenly spaced data, including cubic spline interpolation (Press, Flannery, Teukolsky, and Vetterling, 1986) and the Lomb-Scargle normalized periodogram, a new spectral estimation algorithm published in FORTRAN that does not require evenly spaced input data (Press and Teukolsky, 1988). However, interpolation can have undesirable statistical side effects (Press and Teukolsky, 1988). All of the oil shale yield data included in this study were chosen because they were evenly spaced and did not contain more than a few missing values. Occasional missing values were assigned the average of the two adjacent values.
Sonic and gamma ray logs were analyzed by entering the interval transit time values at 2 ft intervals into a spreadsheet program on a microcomputer. In some cases, this was done manually; for others, the logs were digitized on a digitizing table. Digitization of well logs is subject to problems arising from stretch and distortion common in paper log reproductions.
Sonic log information from Green River Formation open lacustrine rocks predominantly reflects variations in density, which is highly correlated with hydrocarbon content (Bardsley, 1962; Selley, 1985). High interval transit time (ITT) values on the log are interpreted as particularly organic-rich layers, whereas low ITT values are interpreted to represent more calcareous, less organic-rich zones. Known sonic log values for pure carbonate are given as 46 ms/ft. Typical values for oil shale, shale and claystone are more variable, ranging from 60 to 170 ms/ft (Bardsley, 1962; Rider, 1986; Crain, 1986). Sonic log values for this section indicate that lithology ranges from almost pure carbonate to kerogen-rich calcareous claystone. Porosity has a strong effect on sonic values, but the porosity effect in this interval is likely to be insignificant relative to the effect of lithologic variation, since porosities of even 5% would be high for the rock types found in the studied intervals. The information on a sonic log is recorded as interval transit time in microseconds per ft. Values range from about 50 to 115 µs/ft through the intervals of interest. The values were entered to the nearest whole number and are subject to a possible error of +/- 1.0 µs/ft. Estimated hydrocarbon content was calculated for sonic log information using a relationship obtained from previous studies (Stanfield and others, 1960; Bardsley, 1962). Hydrocarbon content is used to calculate relative time intervals and to compare well logs to oil shale yield data.
Gamma ray logs indicate the amount of natural radioactivity present in the formation. Minerals that account for radioactivity in the Green River Formation are potassium-bearing clays derived from pelagic and aeolian sedimentation and volcanic activity. Since these events are generally likely to be poorly correlated to orbital changes, they are not studied in detail. Only one gamma ray log was used in this study, to compare results with sonic log and oil shale assay data.
Density logs give similar information to that provided by sonic log measurements in the context of this research. Density of Green River Formation rocks is strongly correlated with hydrocarbon content (Bardsley, 1962). A density log was used in analyzing the oil shale data from WOSCO EX-1.
The oil shale yield information used in this report is taken from four Bureau of Mines publications (Stansfield and others, 1957, 1960, 1967; Smith and others, 1972) and is given as the result of Fischer assay testing of oil-shale core samples. The Fischer assay technique involves crushing and heating the rock to remove hydrocarbons in liquid form. The measurement units given are gallons per ton (GPT). The method removes most but not all of the organic carbon from the samples; data from this source is a proxy for the actual organic content of the rock. The hydrocarbon fraction of a sample is related to the Fischer assay yield reported in the data sources by the equation
HC = Yield (GPT) x 0.00417 (1)
where the HC fraction is the proportional fraction of hydrocarbons in the rock recoverable by Fischer assay methods (Stanfield and others, 1960). This relationship depends on the specific gravity of the liquid hydrocarbon extracted from the rock; I used 1.0 g/cc for this value, as is reported for petroleum extracted from Green River Formation rocks (Selley, 1985).
The purpose of time series analysis is to mathematically identify meaningful periodic components in time or spatial series (Chatfield, 1984). The methodology of time series analysis is derived from the fields of geophysics, observational astronomy, and electrical engineering. The application of time series analysis to sedimentary sequences in the hope of identifying orbital signals uses the same techniques, with some added complications.
Time series analysis requires several steps. First, any linear trend must be subtracted from the data. This leaves data composed of residuals, or deviations from the linear trend. If the data are not equally spaced, a spline interpolation routine is used to generate equally spaced data for some analytical methods. Compensating for the effects of variable depositional rates always produces an irregularly spaced independent variable, which must be made into a regularly spaced independent variable in order to apply standard methods of spectral estimation. Spline interpolation was used on all corrected data before spectral analysis, except for some intervals analyzed with relatively new methods.
Spectral analysis attempts to estimate the strengths of various frequencies in time series data. The Fourier theorem holds that any continuous, real-valued function defined over a given interval and represented by a finite number of data points can be accurately represented by the sum of a finite number of sinusoidal components (O'Neill, 1983; Davis, 1985). Thus, a time series can be viewed in either the original time domain, or in the frequency domain. In the frequency domain, the data are represented by a curve representing the amplitude (power) of various component frequencies. A curve of squared amplitudes in the frequency domain is referred to as the power spectrum of the time series. Transforming the data into the frequency domain and computing a power spectrum reveals any significant periodic components as peaks. If the depositional rate of Green River Formation rocks was fairly constant or varied systematically with lithology, and if variations in orbital parameters had a strong or significant influence on sedimentation through intermediate effects on Earth's climate, the orbital frequencies should be apparent in power spectra computed from sedimentary sections. Based on assumed sedimentation rates of from 5 to 50 cm/kyr and the known periods of orbital variations, Milankovitch cyclicity in the formation should have spatial frequencies of roughly 0.02 to 0.25 cycles per m, corresponding to cycles with thicknesses of 4 to 50 m.
The units returned by power spectra calculations are units of variance, which are the squares of the original data units. The actual values of the spectral estimates are relatively unimportant, as they vary dramatically depending on the type of data being analyzed. To calculate the significance of spectral estimates or compare spectral estimates computed from different variables, it is necessary to normalize the spectral curve by dividing spectral estimates by the variance of the original data. The variance associated with a particular frequency interval is estimated by the area under the power spectrum curve in that interval.
Two different methods of spectral analysis were used to analyze the Green River Formation sedimentary data. The first is the maximum entropy method (MEM), which was used extensively in the course of this research. A second method of spectral analysis, called the Lomb-Scargle normalized periodogram (LSNP) method, was also used. The use of several different spectral estimation techniques eliminates the possibility that results from any one method are artifacts of the method used, programming errors or other problems.
The maximum entropy method of spectral estimation was well suited to this study. I used a published FORTRAN program for maximum entropy spectral estimation (Press and others, 1986). It has two advantages over the standard FFT method of spectral estimation in the context of this study. The first is that the detail or resolution of the spectral estimate can be controlled during computation; either relatively smoothed or highly detailed spectra can be computed by the MEM algorithm, whereas the spectra obtained by the FFT method must be smoothed in a separate, additional step. The second benefit of the maximum entropy method is that, unlike the FFT method, the MEM can calculate a spectral estimate for any frequency of interest, allowing computation of more dense power spectra. This is particularly useful for investigation of constrained data sets like those from the Green River Formation, where orbital variations may have periods on the same order as the length of the studied interval (Kay and Marple, 1981). For example, if strong periodicity in the data occurs at 1.5 cycles per data length, the MEM will be able to provide a much better estimate of the power at that frequency than the FFT method, because it can give an spectral estimate for exactly 1.5 cycles per data length, whereas the FFT method would only be able to provide estimates for 1 and 2 cycles per data length.
In general, maximum entropy spectral estimates were computed at a frequency interval of 0.25 cycles per data length, four times as densely as possible with the FFT routine. The results were then converted to cycles per m. The only significant drawback of the MEM spectral estimator is that calculating the significance of a given spectral value is not possible. This problem arises from the nonlinear nature of the MEM method (Haykin and Kessler, 1983). An approximate significance test can be made by comparing MEM results to those from the LSNP method, or to the results obtained from Monte Carlo simulations, which involve analyzing random number sequences with similar statistical properties.
The second spectral estimator used is the Lomb-Scargle Normalized Periodogram (LSNP) method (Press and Teukolsky, 1988) mentioned above. It has been included in the spectral estimation methods used in this work for several reasons. First, it weights the data on a per point rather than on a per interval basis, so it can accept unevenly spaced data points directly as input, eliminating the need to interpolate time-corrected or irregular data. Such interpolation techniques can produce undesirable statistical side effects by convolving the data with the interpolation window (Press and Teukolsky, 1988). Second, the FORTRAN algorithm provided for LSNP spectral estimation by Press and Teukolsky (1988) allows the calculation of a probability estimate of the null hypothesis for the largest values in the spectrum. In this case, the null hypothesis represents the case of spectral peaks arising by chance from random data. Significance estimates are made using the assumption that spectral values have an exponential probability distribution with unit mean when normalized by the total variance of the original data set, as is shown by Press and Teukolsky (1988). An estimate of the probability p that a particular spectral peak could have arisen by chance is therefore given approximately by p = M*(e-z), where z is the normalized spectral value, and M is a parameter that depends on the nature of the data set being investigated. For approximately evenly spaced data, M can be set to equal N, the number of data points (Press and Teukolsky, 1988). Using this test, the probability that many of the spectral peaks obtained in this study arose by chance from random data is very small (p < 0.001). Calculations of p for selected peaks appear in the figures. Third, the LSNP seems to provide a very similar spectral estimate to the FFT, but is considerably more flexible and powerful.
In order to attempt to account for the variation in depositional rate with hydrocarbon content, I developed the relative time interval (RTI). This concept can be used to convert spatial data with an assumed relationship between some sedimentary parameter and depositional rate into more nearly temporally linear data. In order to remove the effects of variable depositional rates, I followed a simple procedure. First, the average hydrocarbon content of a section is calculated by equation (1). Then, for each point in the data set, a relative time interval is calculated by dividing the hydrocarbon content at each point by the mean hydrocarbon content in the section. As an example, a hypothetical section will be considered. The mean hydrocarbon content for this section is 11%. For the mean value, the time interval is assigned a value of 1.0. For every data point in the section, the relative fraction of hydrocarbons is calculated, and divided by the mean. Finally, the resulting numbers are normalized so that the highest and lowest time intervals have a ratio of 10:1. This procedure is illustrated graphically in Figure 8, which shows how hydrocarbon content is related to sedimentation rate and relative time interval. The line segment relating hydrocarbon content and relative time intervals must by definition pass through the point describing the average hydrocarbon content and a relative time interval of 1.0, and the y-coordinates of the endpoints must have a ratio equal to the assumed rate variation, in this case 10. A time interval is then assigned to each data point.
A cumulative sum of the relative time intervals represents a new, synthetic time axis of arbitrary units with a more nearly constant time interval, if the model assumptions are correct. This is then substituted for the original spatial variable in the data. The technique compresses low yield intervals and expands high yield zones in time. Since the resulting independent variable is no longer equally spaced after the time interval correction procedure, cubic spline interpolation is used to resample the time corrected curve at equal intervals for analysis by methods that require constant interval data. Finally, the time corrected data are analyzed, and the results compared to those from analysis uncorrected data.
An important point is that time-corrected spectra, while plotted on the same axis as the uncorrected spectra, are not in the same frequency domain. Uncorrected spectra are plotted against cycles per m, whereas corrected spectra are actually plotted against cycles per relative time interval, where the relative time interval is defined as the average amount of time needed to deposit one m of sediment.
Because the relative time intervals are arbitrarily defined so that the data have the same length in relative time units as in spatial units, the two types of spectra are directly comparable in cycles per data length. The use of relative time intervals is a convenient way of assigning time to depth intervals without making any estimate of the depositional rate, which is known only approximately beforehand.
Data from a well in the S ½ of the NE ¼ of Section 11, T. 2 S., R. 4 W., Uinta County, Utah, were analyzed. The well was drilled in 1972 in the Altamont-Bluebell field area. The rocks in this interval are thinly laminated to very thinly bedded mud-supported carbonate and calcareous claystone with high organic content (Fouch, 1981). Sonic log data were recorded at 2 ft intervals for a 512 ft (156 m) interval, from 8590 to 9100 ft log depth, beginning about 660 ft (201 m) below the Mahogany marker at 7930 ft (2417 m) (Fouch, 1981). Raw sonic log data are shown in Figure 14. Gamma ray data for a smaller 220 ft (67 m) interval from 8830 to 9050 ft (2691 to 2760 m) log depth were also recorded.
The sonic data were converted to hydrocarbon content using a relationship derived from measurements by Bardsley (1962) in a study of oil shale in the extreme eastern end of the Uinta basin. They indicate a hydrocarbon content ranging from 2% to 14% in the interval. Hydrocarbon content versus relative time intervals is shown in Figure 15.
A relative time axis was calculated using the relation described above. Relative time intervals range from 0.3 for the organic-poor carbonates to 3.0 for organic-rich intervals. Power spectra of the calculated hydrocarbon content were calculated from the uncorrected data and from data corrected for a 10-fold sedimentation rate variation inverse to the calculated hydrocarbon content of the rock, as previously described. The results of maximum entropy spectral analysis of the data described above are shown in Figure 16, and the LSNP spectra are shown in Figure 17. These results clearly indicate that Milankovitch-band cyclicity is present in the open lacustrine facies of the interval in this area. Furthermore, the results show that the time interval correction created by assuming a 10-fold sedimentation rate variation dramatically improves agreement of calculated spectra with those expected from orbital forcing. The variance associated with the inferred 100 kyr eccentricity variation increases by a factor of three, while the total variance associated with the low frequency portion (0.01 to 0.25 cycles per m) of the spectrum approximately doubles. Both the increased peak heights and the improved correspondence with orbital parameters indicate that a 10-fold variation in depositional rates is a viable model for the depositional system represented by this section.
The sedimentation rate indicated by correlation of the three major low frequency peaks with known orbital parameters is 55 cm/kyr. This rate is the highest found in any of the studied sections. The well penetrates some of the thickest sections found in this interval and is close to the boundary of the Uinta Mountain thrust block, thought to have been active during Green River Formation deposition. Therefore, relatively high depositional rates caused by rapid subsidence in response to thrust loading and high sediment supply are not unexpected.
The data studied here are from a well drilled in Township 2 South, Range 4 West, section 21, Duchesne County, Utah, and represents a section generally correlative with 2 South 4 West 11. Fouch (1981) shows open lacustrine rocks in the interval to be thinning southward from 2 S 4 W 11, where depositional rates must have been unusually high. A 512 ft section, represented by 256 sonic values at 2 ft intervals, was used in the study. The depth of the studied interval ranges from 8400 to 8910 ft. Sonic log values range from 52 to 105 microseconds per ft, corresponding to hydrocarboncontent values of 0 to 11%, assuming that rocks involved are typical oil shales (Bardsley, 1962). The original data and time corrected hydrocarbon data are shown in Figure 18. Maximum entropy spectra using 60 and 100 poles and LSNP spectra were computed for both the original data and the time corrected data. Results are shown in Figure 19 and Figure 20. In this case, the maximum entropy and LSNP spectral estimates seemed to respond differently to the time correction procedure for reasons that are unclear. The disparity may be attributable to undesirable effects of interpolation on MEM spectra, which the LSNP algorithm is not subject to (Press and Teukolsky, 1988). The large peak representing a periodicity of about 50 m is tentatively correlated with the eccentricity variation, but other spectral peaks do not appear to have any relation to orbital parameters. Upon application of time interval correction, however, the LSNP spectrum is more easily correlated with Milankovitch parameters. Periods represented by spectral peaks are calculated using a mean sedimentation rate of 38 cm/kyr. The associated false alarm probabilities are also shown.
MEM spectra show approximately the same periodicities and effects of time interval corrections, except for the discrepancy noted above. The time corrected spectrum picks up the periodicity in the 8 m range tentatively associated with the precessional variation, but does not show the same effect in the 15 m range as does the LSNP spectrum, where time correction yielded a substantial improvement in resolution of the obliquity signal. Overall, MEM spectra of this sequence are less convincing than the LSNP spectra, and than MEM spectra from other locations.
Data from a Uinta Basin well drilled in Township 2 South, Range 4 West, Section 29 were analyzed. Sonic log information from a 512 ft section, from 8176 to 8688 ft, was recorded at two ft intervals. Hydrocarbon fractions were calculated from the sonic log data and used for spectral calculations. Original sonic log data vs. depth and calculated hydrocarbon data vs. relative time are shown in Figure 21.
Figure 22 shows the MEM spectra calculated from the raw data and from data corrected for an approximate tenfold rate variation, based on the calculated hydrocarbon content of the section. LSNP spectra calculated from uncorrected data and corrected data are shown in Figure 23. Spectral analysis of the data from this section reveals strong cyclicity with a period of approximately 50 m and moderate cyclicity with periods in the 10 to 12 m range. Assuming an average sedimentation rate of 40 cm/kyr, it is possible to match the observed cyclicity to Milankovitch climatic influence with a moderate degree of certainty. The time correction technique significantly improves agreement of the second spectrum with expected Milankovitch forcing.
The strong 50 m cyclicity peak and two higher frequencies are shown, along with their period in kyr calculated using a sedimentation rate of 40 cm/kyr, and the false alarm probability calculated for each peak. The time corrected spectrum shows better agreement with Milankovitch periodicities, particularly in the frequency interval associated with the obliquity parameter at the assumed sedimentation rate, and provides fairly convincing evidence for climatic forcing in this section.
The data analyzed are from the Burbank No. 1 corehole, drilled 174 ft south and 390 ft west of the north 1/4 corner of Sec. 22, T. 5 S., R. 96 W., Garfield Co., Colorado (Stanfield and others, 1960). The yield data cover depths from 585 to 760 ft (186 to 232 m). Fischer assay yield values range from 0.9 to 67.2 gallons per ton, representing 0.4% to 28% hydrocarbons. Yield curves are plotted against depth and relative time intervals in Figure 24.
Results from maximum entropy spectral analysis of the original and time corrected data are shown in Figure 25. At a depositional rate of 40 cm/kyr, the low frequency peak correlates well with the eccentricity variation, but the other main peak represents too short a period to represent the precessional variation. Application of a time interval correction based on a 10-fold rate variation improves the agreement of spectra with Milankovitch parameters. The inferred precessional response is greater and is found at a frequency closer to the value predicted by Milankovitch theory. Smaller spectral peaks may represent components of the obliquity response, but are below a level considered to be significant. LSNP spectra show a similar response, and also give significance levels(Figure 26).
The data studied here are from a 215 ft (65.5 m) section from General Petroleum Corp. 32-36, a corehole drilled in section 36, T. 4 S., R. 97 W., Garfield County, Colorado, in the southern Piceance Creek Basin. Sample depths range from 772 to 986 ft (235.3 to 300.7 m). Reported oil shale yields are from 2.7 to 67.2 gallons per ton, or roughly 1% to 28% hydrocarbons by weight. Yield data vs. depth and relative time are shown in Figure 27.
Spectral analysis of the data from this section produced some of the best results of the study. Spectral peaks representing all three of the main orbital parameters are identifiable. Time interval corrections improve correlation of spectral peaks with Milankovitch orbital parameters. Figure 28 shows the 80-pole MEM spectrum computed from original and time corrected data. LSNP spectra are shown in Figure 29. The ratio of the periods of the inferred eccentricity and obliquity signals calculated from the 100-pole MEM time-corrected spectrum differs from the value derived from modern weighted averages (Table 1) by less than 0.2%. The periods shown in the figures are computed using a depositional rate of 40 cm/kyr, considerably higher than the rate thought to typify oil shale deposition.
The data in this section come from the WOSCO EX-1 corehole, the subject of detailed analysis published in US Bureau of Mines Report of Investigations No. 7693 (Smith and others, 1972). The corehole was drilled in 1969 in the southeast corner of Section 36, T. 9 S., R. 20 E., to investigate oil shale yields in the area. The report contains oil shale yields, lithologic descriptions, and X-ray diffraction data for a number of predominant minerals. Additional data were obtained from the density log of a nearby oil well, Belco Petroleum Corp.'s Natural Buttes 43-36, drilled about a half-mile west of the WOSCO corehole. Rocks in the studied interval are extremely continuous, with correlation of individual varve sequences across a 25 mile (40 km) distance reported by Smith and others (1972).
The yield data analyzed comprise a 128 ft (39 m) section beginning at a depth of 2090 ft (637 m). The density log from an adjacent hole was digitized for a 367 ft interval starting at 2050 ft (635 m). No density log was available for the core hole itself. The zone corresponding to the WOSCO EX-1 oil shale data was determined to be from 2143 to 2262 ft (653 to 689 m) on the density log from Natural Buttes 43-36. Correlation of the two curves was accomplished by visual inspection. The data are shown in Figure 33. Linear regression of hydrocarbon fraction residuals from WOSCO EX-1 and density residuals from Natural Buttes 43-36B results in an r value of 0.752, indicating that these two parameters are fairly strongly related. The Mahogany marker is at a depth of 2290 ft (698 m) in the WOSCO EX-1 corehole (Smith and others, 1972). The results of spectral analysis are fairly good for this interval.
Maximum entropy spectral estimates were computed for data from WOSCO EX-1 yield data and Natural Buttes 43-36 sonic data. The maximum entropy spectrum derived from a 110 ft (46 m) assay interval is shown in Figure 34. The short eccentricity cycle and two possible precessional components can be tentatively identified in the spectrum.
The spectra computed from the density data from the longer 367 ft (112 m) section data (Figure 35) are interesting. While this is a fairly long data set, depositional conditions are inferred to be very stable in this area. The higher sensitivity spectral curve may be able to resolve the shorteccentricity cycle, the obliquity cycle, and the precessional cycle. The best fit sedimentation rate indicated by both spectra is about 23 cm/kyr. This rate is considerably lower than that obtained from other intervals, and may reflect the position of the well, which is well to the south of the depositional axis in the Uinta Basin.
The results of this study and previous work on the Green River Formation indicate that Lake Uinta base level was strongly influenced by orbitally generated climatic fluctuations. Base level fluctuations were translated into cyclic sedimentary sequences by sedimentologic and biologic response. During periods of increasing lake level, clastic sediments were trapped at the margins of the lacustrine system, further from the depositional center of the basin. This resulted in decreased clastic sedimentation and chemical precipitation in open lacustrine sediments deposited near the basin center, lower depositional rates, and a relative enrichment of kerogen in open lacustrine sediments. Also, rising lacustrine base level created conditions favorable for high levels of algal productivity. The combination of these two factors led to a dramatic variation in the organic content of deposited layers. Conversely, during relatively hotter and/or drier conditions characterized by decreasing lake level, clastic sediments were carried further into the center of the basin, and increased evaporation would cause increased precipitation of carbonate minerals. Thus, increased clastic and chemical deposition relative to accumulation of organic matter, and higher depositional rates in open lacustrine rocks, are expected during periods of lacustrine contraction. This relationship is reflected in the varve thicknesses reported for Green River Formation rocks.
Examination of power spectra computed from various intervals leads to the conclusion that the three spectral peaks commonly found in the low frequency portion of the spectrum, with periods of 5 to 50 m, can be correlated with the effects of variations in eccentricity, obliquity and precession with a fair degree of confidence. In many cases, one or more of the spectral peaks expected from the Milankovitch theory do not occur, or the spectral peaks that do occur are not easily correlatable to orbital periodicities.
There are many possible explanations for these discrepancies, including deviations from the modelled relationship between sedimentation rate and organic content due to the position of the section in the basin, local tectonic activity, climatic fluctuations unrelated to orbital forcing, and probably other factors. However, the ability to detect significant cyclicity in the Milankovitch band strongly suggests that climate was a significant factor in controlling sedimentation during Green River Formation deposition.
The eccentricity variation is often the strongest component of orbital cyclicity seen in the studied sections. This result supports earlier studies of Milankovitch cyclicity, which report that the eccentricity variation is a major driving mechanism of climatic oscillation, and show that a nonlinear climate model must be invoked to produce 100 kyr power in geologic records (Hays and others, 1976, Prentice and Matthews, 1988). Precessional variations are usually the second strongest cyclic component in studied sections, but are occasionally the strongest peak in the spectrum. Several spectra show the characteristic splitting of the precessional peak into roughly 24 kyr and 19 kyr components. Obliquity variations are usually the most poorly represented of the orbital variations.
Shown in Figure 36 is a highly speculative interpretation of cyclicity present in the data from 2 S, 4 W, Sec. 11, the first interval included in the results section. To create this figure, sine waves of approximately 0.1, 0.05, and 0.02 cycles per m, frequencies commonly represented in power spectra, were visually fitted to the curve representing sonic log response. It should be noted that the three orbital parameters are represented in different portions of the log, and do not occur throughout the interval. There is no evidence to indicate how the phase of the implied orbital parameter is related to the depositional response.
The methods used to investigate the nature of cyclic deposition in the Green River Formation assume that depositional rates are nearly constant or vary systematically with lithology for fairly large stratigraphic intervals. While this assumption is almost certainly violated over very short intervals of 5,000 yr or less, the assumption allows time series analysis of sedimentologic data, which gives positive results in this study. Changes in overall tectonic subsidence rates are thought to be too gradual to affect depositional rates significantly over time spans represented by studied intervals.
The depositional rates inferred by examination of spectral estimates are of the same order as depositional rates estimated by varve calibration (Bradley, 1929) and by gross biostratigraphic age interval calculations (Picard, 1963). Further support is given by recent published estimates of the total lacustrine thickness in the Uinta Basin and its chronostratigraphic boundaries (Berggren, Kent, Flynn and Van Couvering, 1985; Dickinson and others, 1988) which indicate a gross depositional rate of 20 cm/kyr. However, rates determined by this study are often higher than depositional rates determined by varve calibration. Bradley (1929) and subsequent authors (Picard, 1963) give depositional rates of roughly 18 cm/kyr. If the correlation of Green River Formation spectra with Milankovitch climatic forcing is correct, it allows an accurate determination of average depositional rates for specific intervals. In the Uinta basin, depositional rates vary from a maximum of 55 cm/kyr at 2 S, 4 W, Sec. 11 to 20 cm/kyr at WOSCO EX-1 (Figure 2). These results support the paleogeographic and tectonic interpretations of previous studies, which place the depositional axis close to the Uinta uplift, where depositional rates would be greatest (Cashion and Dixon, 1976).
The reasons for the discrepancies for the sedimentation rates calculated by Bradley and those reported here are unclear. The following possibilities can be proposed: 1) subsidence rates decreased significantly by the time of deposition of the upper part of the Parachute Creek sections where Bradley measured his cycles; 2) cycles interpreted as precessional by Bradley may represent shorter time periods; 3) results of this research are in error.
A possible explanation for this discrepancy is that blue-green algae typically undergo two blooms annually, one in the spring, and one in the late summer or early fall (Davis, 1964). His findings on algal productivity in Lake Erie are summarized in Figure 37. If similar algal productivity trends occurred during Green River deposition, it could have caused early workers to mistake biannual laminations as annual, thereby underestimating depositional rates by a factor of two. This does not, however, explain why Bradley's precessional cycles described in outcrops in the PCB yield a sedimentation rate of 15 cm/kyr, which supports an interpretation of varves as annual layers. A further complication is introduced by Buchheim and Biaggi (1988), who show that the number of laminations between two synchronous ash beds approximately doubles between offshore and nearshore sections, examined in the Green River Formation in the Fossil Butte Basin in southwest Wyoming. This indicates that laminations may not signify annual or basinwide events, but depend on local conditions in the water column.
The feasibility of modelling and accounting for fluctuating depositional conditions through the use of the relative time interval has been found useful in resolving Milankovitch cyclicity in the Green River Formation, and could likely be applied successfully to other lacustrine and marine rocks. In most cases, agreement of computed power spectra with frequencies expected from Milankovitch theory can be improved by manipulation of sedimentologic data based on independent assumptions on the nature of the depositional system. A more sophisticated analysis of multiple sedimentologic variables might be able to separately assign variations in clastic, chemical and biologic sedimentation to the separate effects of variations in temperature and precipitation. Such methods are especially well suited to digital well log data.
Analyses of synthetic computer-generated data with varying components of sinusoidal variation and random noise indicate that approximately one-half of the variation in sedimentologic properties studied here is attributable to Milankovitch climatic forcing. The effects of covariant time distortion in the independent axis were found to be much less severe than random time distortion of equal magnitude. While input frequencies were usually still identifiable in synthetic data with covariant time distortion, random time distortion completely obscured input frequencies.
A notable departure of results from analysis of synthetic data from those from actual data is that the spectral peak associated with the 100 kyr eccentricity variation is always much larger in spectra calculated from real data than in those calculated from synthetic data. This is unexpected since some workers have stated that the 100 Kyr eccentricity variation primarily serves to modulate the shorter period obliquity and precessional variations and should not cause a strong spectral peak (Hays and others 1976; Imbrie and Imbrie, 1980; R. K. Matthews, pers. comm.). The fact that 100 kyr power is prevalent in spectra of climatic and geologic spectra indicates that the climate system has a non-linear response to orbital forcing (Hays and others, 1976).
Variation in the biologic, chemical, and clastic components of Green River Formation deposits have been shown to be partially controlled by orbitally induced climatic fluctuations. Paleolake level is an important intermediate mechanism in this phenomenon. Lake level fluctuations may correspond to third- and fourth-order global eustatic sea level curves for the Early to Middle Eocene, to the degree to which paleolake levels in Lake Uinta reflect large-scale variations in climate. Major highstands of Lake Uinta represent periods of increased pluvial conditions and increased water storage on continents. A rough calculation shows that an increase in water depth of 5 m over the combined area of the Uinta and Piceance Creek Basins, approximately 80,000 km2, would have a negligible effect on sea level, on the order of 1 mm. The entire contents of Lakes Uinta and Piceance, assuming an average depth of 50 m, would therefore equal a sea level increment of roughly 1 cm. However, if climatic conditions responsible for lacustrine highstands existed over a large percentage of continental area in the northern hemisphere, enough water could have been stored in mountain glaciers, aquifers and other large lakes to have a noticeable effect on sea level.
The conclusions that can be drawn from this research include the following: 1) Early to middle Eocene paleoclimate affecting Green River Formation deposition did have a significant oscillatory component, and the oscillations were orbitally induced. 2) Time series analysis, in conjunction with other information, provides a means of recognizing orbital cyclicity and determining depositional rates in the Green River Formation. 3) Fine laminations in Green River Formation rocks may not be strictly annual varves, or earlier studies of depositional rates based on varve counts used intervals with significantly lower sedimentation rates. Overall depositional rates determined by spectral analysis are significantly higher than those reported using varve methods. 4) Variable sedimentation rate models can be effectively used to remove time distortion from measured sections. A depositional model which attempts to account for likely variations in sedimentation rates may be useful in recognition of orbital influence on depositional systems.
Anderson, R. Y., 1986, The varve microcosm: Propagator of cyclic bedding: Paleoceanography, v. 1, p. 373 - 382.
Anderson, R. Y., and Dean, W. E., 1988, Lacustrine varve formation through time: Paleogeography, Paleoclimatology, Paleoecology, v. 62, p. 215 - 235.
Bardsley, S. R., 1962, Log Analysis of Oil Shale: unpublished Master's Thesis, Univ. of Utah.
Berger, A. L., 1976, Obliquity and precession for the last 5,000,000 years: Astron. and Astrophys., v. 51, p. 127 -135.
Berger, A. L., 1977a, Support for the astronomical theory of climatic change: Nature, v. 269, p. 44 - 45.
Berger, A. L., 1977b, Long term variations of the Earth's orbital elements: Celestial Mechanics, v. 15, p. 53 - 74.
Berger, A. L., 1978, Long term variations of caloric solar radiation resulting from the Earth's orbital elements: Quat. Res., v. 9, p. 139 - 167.
Berger, A. L., 1980, The Milankovitch astronomical theory of paleoclimates: a modern review: Vistas in Astronomy, v. 24, p. 103 - 122.
Berggren, W. A., Kent, D. V., Flynn, J. J., and Van Couvering, J. A., 1985, Cenozoic geochronology: Geol. Soc. Am. Bull., v. 96, p. 1407 - 1418.
Bradley, W. H., 1929, The varves and climate of the Green River epoch: USGS Prof. Paper 158.
Buchheim, H. P., and Biaggi, R., 1988, Laminae counts within a synchronous oil shale unit: a challenge to the "varve" concept (abs): in 100th Geol. Soc. Am. Program with Abstracts, p. A317.
Cashion, W. B., 1957, Stratigraphic relations and oil shale of the Green River Formation in the Eastern Uinta Basin, in: Otto G. Seal, ed., Guidebook to the Geology of the Uinta Basin, IAPG. p. 131 - 135.
Cashion, W. B., and Dixon, G. H., 1976, Isopach map and cross section of the Mahogany zone of the Green River Formation derived principally from geophysical well logs, eastern Uinta Basin, Utah and Colorado: US Geological Survey Miscellaneous Field Studies Map MF-797, scale 1:250,000.
Chatfield, C., 1984, The Analysis of Time Series. Chapman and Hall, New York. 286 p.
Cooley, J. W., and Tukey, J. W., 1965, An algorithm for machine computation of complex Fourier series: Mathematics of Computation, vol. 19, p. 297.
Crain, E. R., 1986, The Log Analysis Handbook, Volume 1. PennWell Publishing, Tulsa. 684 p.
Croll, J., 1875, Climate and time in their geological relation; a theory of secular change in the earth's climate: Edward Standford, London. 557 p.
Davis, C. C., 1964, Evidence for the eutrophication of Lake Erie from phytoplankton records: Limnology, Oceanography, vol. 9, p. 275 - 283.
Davis, J. C., 1985, Statistics and data analysis in geology, Wiley, New York. 550 p.
de Boer, P. L., 1982, Cyclicity and storage of organic matter in Middle Cretaceous pelagic sediments: in Cyclic and Event Stratification, Springer-Verlag, New York..
Desborough, G. A., 1978, A biogenic-chemical stratified lake model for the origin of oil shale of the Green River Formation: an alternative to the playa-lake model: Geol. Soc. Am. Bull., v. 89, p. 961 - 971.
Dickinson, W. R., Klute, M. A., Hayes, M. J., Janecke, S. U., Lundin, E. R., McKittrick, M. A., and Olivares, M. D., 1988, Paleogeographic and Paleotectonic setting of Laramide sedimentary basins in the central Rocky Mountain region: Geol. Soc. Am. Bull., v. 100, p. 1023 - 1039.
Donnell, J. R., 1961, Tertiary geology and oil-shale resources of the Piceance Creek Basin between the Colorado and White Rivers, Northwestern Colorado: US Geological Survey Bulletin 1082-L.
Duff, P. M., Hallam, A., and Walton, E. K., 1967, Cyclic Sedimentation: in Developments in Sedimentology, v. 10. Elsevier, New York. 288 p.
Emiliani, C., 1963, Pleistocene temperatures: J. Geol., p. 538 - 591.
Eugster, H. P., and Hardie, L. A., 1975, Sedimentation in an ancient playa-lake complex: The Wilkins Peak Member of the Green River Formation of Wyoming: Geol. Soc. Am. Bull., v. 86, p. 319 - 334.
Fischer, A. G., 1982, Long-term climatic oscillations recorded in stratigraphy: in Climate in Earth History, National Academy Press, Washington, 198 p.
Fischer, A. G., 1986, Climatic rhythms recorded in strata: Ann. Rev. Earth and Plan. Sci., v. 14, p. 356 - 371.
Fischer, A. G., Silva, I. P., Anderson, R., Berger, A. L., Bottjer, D., Cotillon, P., Dean, W. E., de Boer, P. L., Erba, E., Foucault, A., Gardner, J., Hinnov, L., Loutre, M. F., Matthews, M., Melnyk, H., Napoleone, G., Paul, C. R. C., Ricken, W., Ripepe, M., Roth, P. H., Schwarzacher, W., Smith, D., Strasser, A., Tornaghi, M. E., van Echelpoel, E., and Weedon, G., 1988, Cyclostratigraphy: Cretaceous Resources, Events, and Rthythms Workshop 3, Perugia, Italy.
Fouch, T. D., 1976, Revision of the lower part of the Tertiary system in the central and western Uinta Basin: USGS Bulletin 1045-C, 7 p.
Fouch T. D., 1981, Distribution of rock types, lithologic groups, and interpreted depositional environments for some lower Tertiary and upper Cretaceous rocks from outcrops at Willow Creek-Indian Canyon through the subsurface of Duchesne and Altamont Oil Fields, southwest to north central parts of the Uinta Basin, Utah. USGS Oil and Gas Investigation Map, Chart OC-81.
Franczyk, K. J., Pitman, J. K., Cashion, W. B., Dyni, J. R., Fouch, T. D., Johnson, R. C., Chan, M. A., Donnell, J. R., Lawton, T. F., and Remy, R. R., 1989, Evolution of resource-rich foreland and intermontane basins in eastern Utah and western Colorado: AGU Field Trip Guidebook T324.
Gentleman, W. M., and Sande, G., 1966, Fast Fourier transforms for fun and profit: AFIPS Conference Proceedings, vol. 29, p. 563 - 578.
Gilbert, G. K., 1895, Sedimentary measurement of Cretaceous time: Jour. Geol., v. 3, p. 121 - 127.
Haq, B. U., Hardenbol, J., and Vail, P. R., 1987, Chronology of fluctuating sea levels since the Triassic: Science, v. 235, p. 1156 - 1166.
Haykin, S., and Kesler, S., 1983, Prediction-error filtering and maximum-entropy spectral estimation: in Nonlinear Methods of Spectral Analysis: Topics in Applied Physics v. 34, Springer-Verlag, New York, 263 p.
Hays, J. D., Imbrie, J., and Shackleton, N. J., 1976, Variations in the Earth's orbit: pacemaker of the ice ages: Science, v. 194, p. 1121 - 1132.
Herbert, T. D. and Fischer, A. G., 1986, Milankovitch climatic origin of mid-Cretaceous black shale rhythms in central Italy: Nature, v. 321, p. 739 - 743.
Hintze, L. F., 1980, Geologic Map of Utah: Utah Geol. and Min. Sur., S. L. C., Utah, 1:500,000 scale.
Hintze, L. F., 1988, Geologic History of Utah: Brigham Young University Geology Series, Provo, Utah, 202 p.
House, M. R., 1985, A new approach to an absolute timescale from measurements of orbital cycles and sedimentary microrhythms: Nature, v. 315, p. 721 - 725.
Hutchinson, G. E., 1957, A Treatise on Limnology, Vol. 2: Introduction to lake biology and the limnoplankton. John Wiley and Sons, New York.
Imbrie, J., Hays, J., Martinson, D. G., McIntyre, A., Mix, A. C., Morley, J. J., Psias, N. G., Prell, W. L., and Shackleton, N. J., 1984, The orbital theory of Pleistocene climate: support from a revised chronology of the marlen ë18O record: in Berger and others, eds., Milankovitch and Climate. Dordrecht/Boston/Lancaster, Reidel, 510 p.
Imbrie, J., and Imbrie, J. Z., 1980, Modelling the climatic response to orbital variations: Science, v. 207, p. 943 - 953.
Johnson, R. C., 1981, Stratigraphic evidence for a deep Eocene Lake Uinta, Piceance Creek Basin, Colorado: Geology, v. 9, p. 55 - 62.
Kay, Steven M., and Marple, Stanley L. Jr., 1981, Spectrum Analysis - A Modern Perspective: Proceedings of the IEEE, vol. 69, p. 1380 - 1419.
Kerr, R. A., 1987, Milankovitch climate cycles through the ages: Science, v. 235, p. 973 - 974.
Koesoemadinata, R. P., 1970, Stratigraphy and petroleum occurrence, Green River Formation, Red Wash Field, Utah: Colorado School of Mines Quarterly, v. 65, no. 1, p. 1 - 77.
Lundell, L. L., and Surdam, R. C., 1975, Playa-lake deposition: Green River Formation, Piceance Creek Basin, Colorado: Geology, v. 3, p. 493 - 497
Matthews, M. D., and Perlmutter, M. A., 1989, Cyclostratigraphic sequences from the Fundy and Green River basins: (abs.) 101st Geol. Soc. Am. Ann. Program with Abstracts, p. A138.
Mauger, R. L., 1977, K-Ar ages of biotite from tuffs in Eocene rocks of the Green River, Washakie, and Uinta basins, Utah, Wyoming and Colorado: University of Wyoming Contributions to Geology, v. 15, no. 1, p. 17 - 41.
McDonald, R. E., 1972, Eocene and Paleocene rocks of the southern and central Rockies, in Mallory, W. W., ed., Geologic Atlas of the Rocky Mountain Region: RMAG.
Milankovitch, M., 1941, Canon of insolation and the ice-age problem: K. Serb. Akad. Geogr., Special Publ. No. 132, 484 p.
Olsen, C. R., 1976, Sedimentation Rates: in Fairbridge, R. W., and Bourgeois, eds., The Encyclopedia of Sedimentology. Stroudsburg, Pa., Dowden, Hutchinson and Ross, p. 687 - 691.
Olsen, P. E., 1984, Periodicity of lake-level cycles in the Late Triassic Lockatong Formation of the Newark Basin: in Berger and others, eds., Milankovitch and Climate. Dordrecht/Boston/Lancaster, Reidel, 510 p.
O'Neill, P. V., 1983, Advanced Engineering Mathematics. Wadsworth Publishing, Belmont, California. 1108 p.
Park, J., and Herbert, T. D., 1987, Hunting for paleoclimatic periodicities in a geologic time series with an uncertain time scale: J. Geophys. Res. 92, B13, p. 14,027 - 14,040.
Perlmutter, M. A., and Matthews, M. D., 1989, Global Cyclostratigraphy: Sedimentologic response to orbitally forced climatic change: (abs.) in 101st Geol. Soc. Am. Ann. Program with Abstracts, p. A138.
Picard, M. D., 1955, Subsurface stratigraphy and lithology of Green River Formation in the Uinta Basin, Utah: Am. Assoc. Petroleum Geologists Bulletin, v. 39, no. 1, p. 75 - 101.
Picard, M. D., 1957, Green River and lower Uinta Formations -subsurface stratigraphic changes in central and eastern Uinta basin, Utah, in: Otto G. Seal, ed., Guidebook to the Geology of the Uinta Basin, IAPG. p. 116 - 130.
Picard, M. D., 1963, Duration of Eocene Lake, Uinta Basin, Utah: G. S. A. Bulletin, v. 74, p. 89 - 90.
Picard, M. D., 1985, Hypotheses of oil-shale genesis, Green River Formation, northeast Utah, northwest Colorado, and southwest Wyoming: in Picard, M. D., ed., Geology and energy resources, Uinta Basin of Utah: Utah Geological Association, Salt Lake City, Utah, p. 193 - 210.
Picard, M. D., and High, L. R., 1972, Paleoenvironmental reconstructions in an area of rapid facies change, Parachute Creek Member of Green River Formation (Eocene), Uinta Basin, Utah: Geological Society of America Bulletin, v. 83, p. 2689 - 2708.
Pitman, J. K., Fouch, T. D., and Goldhaber, M. B., 1982, Depositional Setting and Diagenetic Evolution of Some Tertiary Unconventional Reservoir Rocks, Uinta Basin,, Utah: Am. Assoc. Petroleum Geologists Bulletin, v. 66, no. 10, p. 1581 - 1596.
Prentice, M. L., and Matthews, R. K., 1988, Cenozoic ice-volume history: development of a composite oxygen isotope record: Geology, v. 16, p. 963 - 966.
Press, W. H., Flannery, B. P., Teukolsky, S. A., and Vetterling, W. T., 1986, Numerical Recipes: The Art of Scientific Computing. Cambridge University Press, New York. 818 p.
Press, W. H., and Teukolsky, S. A., 1988, Search algorithm for weak periodic signals in unevenly spaced data: Computers in Physics, v. 1, no. 12, p. 77 - 81.
Rider, M. H., 1986, The geological interpretation of well logs. Blackie Halsted Press, New York. 175 p.
Ryder, R. T., Fouch, T. D., and Elison, J. H., 1976, Early Tertiary sedimentation in the western Uinta Basin, Utah: Geol. Soc. Am. Bull., v. 87, p. 496 - 512.
Selley, R. C., 1985, Elements of Petroleum Geology. Freeman, New York. 449 p.
Sergin, V. Y., 1980, Origin and mechanism of large-scale climatic oscillations: Science, v. 209, p. 1477 - 1482.
Smith, J. W., Trudell, L. G., and Robb, W. A., 1972, Oil yields and characteristics of Green River Formation Oil Shales at WOSCO EX-1, Uinta County, Utah: US Bureau of Mines Report of Investigations 7693. 145 p.
Smith, M. C., 1986, Bibliography of the Geology of the Green River Formation, Colorado, Utah, and Colorado: USGS Open-file report 86-466, 86 p.
Stanfield, K. E., Rose, C. K., McAuley, W. S., and Tesch, W. J. Jr., 1957, Oil Yields of sections of Green River Oil Shale in Colorado, 1952 - 1954: US Bureau of Mines Report of Investigations 5321, 132 p.
Stanfield, K. E., Smith, J. W., Smith, H. N., and Robb, W. A., 1960, Oil Yields of sections of Green River Oil Shale in Colorado, 1954 - 1957: US Bureau of Mines Report of Investigations 5614, 186 p.
Stanfield, K. E., Smith, J. W., and Trudell, L. G., 1963, Oil yields of sections of Green River Oil Shale in Utah, 1952 - 1962: US Bureau of Mines Report of Investigations 6420, 217 p.
Stanfield, K. E., Smith, J. W., and Trudell, L. G., 1967, Oil Yields of sections of Green River Oil Shale in Colorado, 1957 - 1963: US Bureau of Mines Report of Investigations 7051, 284 p.
Surdam, R. C., and Stanley, K. O., 1979, Lacustrine sedimentation during the culminating phase of Lake Gosiute, Wyoming: Geological Society of America Bulletin, v. 90, pt. 1, p. 93 - 110.
Van Houten, F. B., 1964, Cyclic lacustrine sedimentation, upper Triassic Lockatong Formation, central New Jersey and adjacent Pennsylvania: Kansas State Geol. Surv. Bull., v. 169, p. 497 - 531.
