Matthews, M. D., 1996, Migration—a view from the top, in D. Schumacher and M.A. Abrams, eds., Hydrocarbon migration and its near-surface expression:AAPG Memoir 66, p. 139–155.

Chapter 11

Migration—A View from the Top

Martin D. MatthewsTexaco International Exploration DivisionBellaire, Texas, U.S.A.

Abstract

Many mechanisms have been proposed for hydrocarbon migration, and many processes have beendescribed that modify the composition of migrating hydrocarbons. Examination of subsurface and surface dataindicates that all the proposed mechanisms and processes are active. However, many play minor roles only rec-ognizable in special situations. The dominant migration mechanism is as a free phase, rising under the forces ofbuoyancy within carrier and reservoir rocks, and capillary imbibition in the transition from sources and seal intocarrier rocks. The migration pathway is determined by three-dimensional heterogeneity at all scales, from theindividual pore systems to the interrelationships of facies. The dominant process modifying the composition ofmigrating hydrocarbons is phase partitioning, as evidenced by subsurface and surface data on hydrocarbonexpulsion, migration, and accumulation.

In the near surface, many processes act to modify this seepage, particularly biogenic activity and diffusion(both chemical and mechanical). Free-gas surface hydrocarbon survey measurements, however, are dominatedby this seepage mechanism. It explains (1) the spatial variability of the data; (2) the relationship of high-magni-tude sites to high-permeability geologic features such as faults, fractures, unconformities, and outcropping reser-voir units; (3) the compositional relationship of subsurface reservoired hydrocarbons to source rocks and thelack of relationship to ineffective source rocks; the variation of magnitude with time, both long and short; andcurrent estimates of the rate of transport in the near surface.

INTRODUCTION

This paper is concerned with postexpulsion migration.With the exception of boundary conditions, there is littleor no difference in the processes involved in secondaryand tertiary migration. The knowledge gained by study-ing one directly constrains the behavior of the other.Postexpulsion migration is concerned with all aspects ofmigration, through mixed geologic sequences (reservoirs,conduits, seals, faults, and fractures) and ultimately to thesurface. This mixed sequence of rocks is simplified intotwo classes: relatively large pore networks and relativelysmall pore networks. Because these terms are used in ref-erence to adjacent systems, they may refer to a contactbetween sandstone and shale, fine and coarse sandstone,shale and evaporite, or even adjacent pore networks. Porenetworks are chosen as the reference because they are aprime control on flow rate and volume.

Conventional conclusions on how hydrocarbonsmigrate are dominated by studies of flow in reservoirs byengineers, laboratory simulations, and theory of flow inporous media. These studies rarely encompass tens of

years, while the time frame of reservoir charging is prob-ably much longer. The transport of hydrocarbonsthrough an entire stratigraphic section has been general-ly ignored. Notable exceptions are the work of hydrolo-gists such as Tóth (1988 and this volume) and basin mod-elers such as Welte and Yükler (1980), Durand et al.(1983), and Nakayama and Lerche (1987).

Migration is like Einstein’s watch. Observations con-cerning its operation can be made, but since opening thesystem is not permitted, only hypotheses about its oper-ation, consistent with those observations, can be made.The movement of hydrocarbons in the deep and shallowsubsurface is a complex balance of processes. We candraw conclusions based only on our understanding ofthe basic principles of science and our observations. Wemay never know if these conclusions are correct for anygiven situation.

First, I give an overview of the forces and processesresponsible for hydrocarbon migration. This is followedby a description of constraints placed on migration bysubsurface and near-surface studies. This information isthen used to generate a model of migration, consistentwith all these data.

139

MIGRATION OVERVIEW

Three principal questions about migration arise. (1)How is cross-formational flow of hydrocarbons accom-plished? (2) Does the form change during migration and,if so, which form is dominant under what conditions? (3)How does the dominant form (or forms) control the vol-umes and compositions transported? To consider thesequestions, it is appropriate to review the constraintsplaced on our understanding of the subsurface migrationprocess. The interrelationships of some of these con-straints are shown in Figure 1.

Energy

The movement of hydrocarbons from one location toanother requires energy. Many sources of energy are pos-sible, and to one extent or another all may be active. Somearise directly as a result of the presence of hydrocarbonsin given location, including buoyancy, chemical potential

(related to concentration differences), and compaction ofthe sediment (squeezing the hydrocarbons from the col-lapsing pore space). Some act indirectly on the hydrocar-bons as conditions around them change due to burial,such as thermal expansion and water motion due to com-paction. Others come about due to decreases in pressureand temperature as a result of the upward motion ofhydrocarbons through the rock, such as phase changeand volume expansion.

Physical Form

Numerous models for the physical form of migratinghydrocarbons have been proposed. These can be groupedinto three categories: dissolved in water, separate phase,and transport along a continuous kerogen network.

Hydrocarbons dissolved in water occur as true solu-tion and micellar solution. Both of these forms enable thehydrocarbons to move one molecule at a time and thuspresent minimal restriction to movement. True solution isa function of pressure, temperature, salinity, molecularweight, and mixtures of components present. Micellarsolution increases the capacity of water to carry molecularhydrocarbon species by the use of naturally occurringhydrocarbon solubilizers.

Separate phase migration of hydrocarbons occurs asdiscrete droplets (smaller than a pore throat) and as alarger continuous mass extending across several pores.Dispersions of separate phase hydrocarbons occur as dis-crete droplets, colloids, or emulsions, either gaseous orliquid. The taller the hydrocarbon mass, the greater thebuoyancy force. Separate phase migration of hydrocar-bons is subject to capillary forces at its contact with water.Continuous kerogen networks are only expected in richsource rocks and can be discounted as a dominant factorin post-expulsion migration.

Processes Causing Chemical ChangesDuring Migration

During postexpulsion migration, many processes arecapable of altering the chemical characteristics of thehydrocarbons expelled from the source rock. Theseinclude water washing, adsorption, phase partitioning,mixing, and biodegradation. Water washing is the selec-tive removal of the more water-soluble components.Adsorption of migrating hydrocarbons onto the mineralsubstrate it passes through can lead to both selectiveremoval of hydrocarbons and selective retardation of themigration rate of hydrocarbons, as in a chromatographiccolumn. Phase partitioning is the concentration of differenthydrocarbon species into gaseous and liquid phases withchanges in pressure and temperature. Mixing alters com-position through inclusion of hydrocarbons from otherkerogen particles along the migration path, combiningmigration streams from two or more source rocks, andprecipitation of asphaltenes and other high molecularweight compounds by the addition of methane. Bio-degradation is the biologic alteration of hydrocarbons.

140 Matthews

by

org

anis

ms

mo

dif

icat

ion

mix

ing

ph

ase

sep

arat

ion

So

luti

on

diffusion

diffusion

adso

rpti

on

leakagete

rtia

rym

igra

tio

n

Sourceexpulsion

pri

mar

y m

igra

tio

n

re-m

igra

tio

n

wat

er w

et p

ore

sys

tem

Reservoir

Reservoirm

igra

tio

np

rim

ary

entrapment

entrapment

leakageSourceexpulsion

Figure 1—Schematic diagram of subsurface migration.Rock framework is shown in boxes, events in black letter-ing, and processes in gray.

Concentration MechanismsProcesses capable of concentrating hydrocarbons from

the migrating stream into a reservoired accumulationinclude pressure minimums, critical orifices, exsolution,and capillary forces. Perfect seals, which do not leak at all,are the type of seals most petroleum geologists visualize,but they rarely occur. Pressure minimums are a perfect seal.When all the forces acting on a hydrocarbon mass areresolved, the hydrocarbons remain in the minimum aslong as it exists. There is no migration out of that mini-mum. Critical orifices apply specifically to dissolvedspecies. They act as molecular sieves, allowing particlessmaller than the orifice to pass, while retaining particleslarger than the orifice. If seals were uniformly composedof subcritical pore throats, this would also form a perfectseal. Nonporous unfractured rock, such as a salt, willform such a seal under certain circumstances. Exsolutionof hydrocarbons is facilitated by increasing salinity,decreasing temperature, and decreasing pressure. Once aseparate phase is formed, capillary forces become effec-tive. Capillary forces arise at the interface between twoimmiscible mobile phases across a restricted opening. Asthe pressure difference across a capillary restrictionincreases, the interface deforms and the nonwettingphase eventually penetrates the restriction.

Near-Surface Boundary ConditionsSurface geochemical observations are made close to

the end of the subsurface migration path. They offer theadvantage of virtually complete spatial availability ofsampling locations, but are subject to certain boundaryconditions (Figure 2) that must be considered in interpre-tation of the data. Four interfaces are of interest: water-wet pores with free water, free water with the atmos-phere, water-wet pores with soil air, and soil air with theatmosphere.

Water-Wet Pores with Free Water

The behavior of migrating hydrocarbons at the inter-face between water-wet pores and free water is particu-larly instructive because, in addition to sampling by cor-ing and sniffers, it is subject to direct observation andmeasurement by submersibles and free diving. At shal-low depths below the sediment–water interface, the con-centration of hydrocarbons generally increases down-ward, as shown by Abrams (1992, and this volume).Detailed depth profiles (Brooks et al., 1978) show that thisgeneral increase is subject to considerable variation andlocal reversals of hydrocarbon concentration with depth.This variation is similar to that observed in high-sensitiv-ity mud logs at deeper depths. Separate phase hydrocar-bons escape through the sediment water interface as bothbubbles (gas dominated) and globules (liquid dominat-ed). The rate of escape varies greatly, from intermittent ona scale exceeding days to nearly continuous. The occur-rence of separate phase hydrocarbons at the sea floor isdirect proof that postexpulsion migration in a separatephase exists in the subsurface as well.

Sporadic escape suggests that the process often pro-ceeds as small separate masses. Nearly continuous bub-ble trains support the occasional existence of longer con-tinuous hydrocarbon stringers in the subsurface. As thesestringers reach the free water surface, they break up dueto surface tension. Gas bubbles rise rapidly due to theirlow density and expand as they move upward to lowerpressure. If a bubble grows too large, it breaks up intosmaller bubbles. Bubbles moving through the water col-umn dissolve into the undersaturated water. If the bubbleis small, it may dissolve before it reaches the water–atmosphere interface. If the bubble contains significantquantities of liquids, these liquids plate out at the bubblesurface, preventing the light gases from dissolving. Therate of bubble transport in the water column is high, andunder most conditions, the lateral transport is slight(<45˚). The train of bubbles widens from a narrow pointor line of entry at the sea floor to a cone that is most oftennarrower than 1 km at the surface. Surrounding this coneof bubbles is a zone of water with elevated dissolvedhydrocarbons formed by solution of portions of the bub-ble train. This behavior is demonstrated by the occur-rence of well-defined peaks on sniffer surveys surround-ed by broad shoulders elevated above background.

Chapter 11—Migration—A View from the Top 141

ph

ase

sep

arat

ion

So

luti

on

adso

rpti

on

bu

bb

le e

xpan

sio

n

mo

dif

icat

ion

by

org

anis

ms

oxi

dat

ion

WA

TE

R W

ET

PO

RE

S

WA

TE

R

SO

IL A

IR

AIR

diffusion

evaporationdiffusion

evaporation

Figure 2—Schematic diagram of near-surface migration.Media through which hydrocarbons migrate are shown inboxes and processes are in gray lettering.

Oil globules move more slowly and may be signifi-cantly displaced by water motion. Detection of dissolvedhydrocarbons escaping through the sediment water inter-face is more difficult because this transport is not visible.Any dissolved hydrocarbons escaping through this inter-face are subject to rapid mixing and dilution by the over-lying water mass. Broad regions of elevated hydrocarbonconcentrations are detected by sniffer surveys. Wherethese occur near the sea floor they may indicate (1) escapeof dissolved hydrocarbons over a broad area , (2) inter-mittent escape of bubbles over a large area, or (3) disper-sive mixing and destruction of a discrete bubble train.

Free Water with the Atmosphere

Once a bubble reaches the water—atmospheric inter-face, it breaks and is rapidly mixed with the atmosphere.Globules form a slick, evaporating into the atmosphereand forming biodegraded masses or remaining as sepa-rate globules. Globules and biodegraded masses continueto lose volatiles to both the water and atmosphere andbiodegrade until they sink or run aground. Hydro-carbons dissolved in the water column either escape tothe atmosphere or are consumed by organisms. Williamset al. (1981) pointed out that the level of backgroundhydrocarbons in sea water is higher in productive areasthan in the open ocean. The degree to which solution ofbubbles in the water column and transport of dissolvedhydrocarbons through the sea floor each contribute tothis phenomenon is unknown.

Water-Wet Pores with Soil Air

The behavior of hydrocarbons at the interface betweenwater-wet pores and soil air is similar to that describedabove. Hydrocarbons in solution either come out of solu-tion and join those already in the gas or liquid phase orare consumed by organisms. Hydrocarbons in a liquidphase spread out to form a free-floating mass subject tobiological action, just like a slick at the sea–air interface.These hydrocarbons can be sampled from shallow waterwells. Portions of this mass are slowly transportedupward into the soil air by gaseous migration, throughdiffusion or effusion, either alone or with a carrier gas,such as methane. Hydrocarbons in a gas phase moveupward due to buoyancy and diffusion, aided by baro-metric pumping.

The concentration of hydrocarbons decreases towardthe soil air–free air interface due to dilution of the migrat-ing gases with atmospheric gases. The concentration atany one point is therefore a function of the rate of hydro-carbon transport (controlling the volume of hydrocar-bons delivered to that location), the rate of transport ofatmospheric gases (controlling the volume of nonhydro-carbon gases delivered to that location), and the rate ofdestruction of hydrocarbons (and methane production)by organic activity. Statistically, the deeper the sample, thebetter, just like that observed offshore below the sedi-ment–water interface. Studies at Gulf Research andDevelopment Company suggest that for drilled holes, adepth of 3.6 m (12 ft) provides stable readings. Probe

studies suggest that a depth of 1.2 m (4 ft) also providesstable data, but with reduced magnitudes. Depth profilesindicate that both increases and decreases in concentra-tion can occur at any depth, with higher concentrationsgenerally occurring below less permeable layers, similarto the behavior of mud logs but with reduced variation.This suggests that the concentration of hydrocarbons inthe shallow surface (and in the deeper subsurface) isdependent on the three-dimensional permeability struc-ture of the soil and rocks being sampled rather than a sim-ple vertical mixing curve in a homogeneous medium.Despite this vertical variability, at any given depth, siteswith high concentrations tend to be consistently high andsites with low values tend to be consistently low.

Soil Air with the Atmosphere

At the interface between soil air and the atmosphere,the concentration of hydrocarbons is fixed at or close tothe average atmospheric concentration. Study of the con-centration of hydrocarbons in tents sealed to the groundby Gulf Research and Development Company indicatethat at low-level sites, the exchange occurs in both direc-tions. The concentration of hydrocarbons in the air mayincrease above that in the atmosphere, presumably due topreferential expulsion of hydrocarbons from the soil air,or it may decrease, presumably due to preferentialremoval of hydrocarbons from the air by the soil. Theseexchanges are rapid, occurring within hours. At high-level sites, the concentrations also vary but remain aboveatmospheric concentrations. Antropov et al. (1981) foundsimilar behavior in the atmosphere over petroleum fieldsusing adsorption gas lasers.

LESSONS FROM SOURCE AND RESERVOIR STUDIES

I have made a wide variety of observations about thecharacteristics of rocks during burial, their ability totransmit fluids, and the static distribution of hydrocar-bons in them. These are summarized in this sectionunder the following topics: source–reservoir correlation,physical sizes, pore throat heterogeneity, fluid inclusions,phase behavior, hydrocarbon solution in water, diffu-sion, rates of hydrocarbon motion, mud log data, andseismic evidence.

Source–Reservoir Correlation

The fact that oils can be chemically correlated with aparticular source rock places a fundamental constraint onany proposed migration process. It strongly indicates thatthe expulsion–migration process does not significantlyeffect the overall geochemistry of the migrated product.However, there are also some differences. Thompson(1988) points out that reservoired hydrocarbons in closeproximity to one another and geochemically matched tothe same family or source are frequently dissimilar in

142 Matthews

their light ends and other properties. This suggests thatthe migration and entrapment process has some controlover the chemical characteristics of reservoired hydrocar-bons, but in most cases this only affects the detailedchemistry, not the overall structure.

Physical Sizes

The majority of hydrocarbon molecules are larger thanthe size of small shale pores and approximately equal tothe median pore size of shales (Figure 3). Methane mole-cules are slightly larger than the size of small shale pores,and water molecules are slightly smaller. The lower rangeof oil colloids is within the range of median shale pores,while larger colloids and other hydrocarbon particulates,such as micelles and bitumen, are in the range of thelargest shale pores. This suggests that water, methane,and the smaller paraffins and aromatics are capable ofmoving through shales in solution, but that the largermolecules, aggregates, and separate phases undergo var-ious degrees of molecular filtration and capillary block-age in attempting to pass through all but the largest shalepores or open fractures. The effect of burial on shale per-meability (proportional to pore throat size) is shown inFigure 4. Note that shale compaction results in a logarith-mic decrease in permeability down to a depth of about 3.5km. Beyond this depth, two factors vie for dominance:continued collapse of the pore network and the formationof micro- and macrofractures as the shale becomes morebrittle (Neglia, 1979).

This suggests that diffusion and aqueous solution trans-port may be practical for the smaller hydrocarbon mole-cules, but becomes increasingly less likely with increasingmolecular size. Direct transport of larger separate phaseparticles, while possible through the large pores, becomes

increasingly less likely as the traversed path lengthens dueto the increased probability of a continuous large pore net-work terminating into a small pore throat. Indeed, even theflow of the comparatively small water molecule oftenrequires significant pressure gradients to overcome therestrictions to flow common in shales.

Pore Throat Heterogeneity and InterstitialContinuity

The range of shale pore size is more than five orders ofmagnitudes (Figure 3). The effect of this heterogeneity isdiagramatically represented in two dimensions in Figure5, which shows the interconnectivity of 200 pore throatsof five different sizes (1 to 5 units) randomly scattered ina regular pattern. Entry of hydrocarbons from a largepore network is assumed to occur at the bottom of theunit and exit into a similar unit at the top. There is no con-nection through this network that is composed exclusive-ly of the largest (5-unit) pore throats. This is representa-tive of a no-flow condition for hydrocarbons and is rarelyif ever achieved in nature. It would represent a very finegrained, highly compacted, unfractured rock, such as anevaporite.

A more typical fine-grained condition is shown inFigure 5a. The continuous networks of 4- and 5-unit porethroats are highlighted. Note that only one is continuousthrough the unit (5% of the potential exit pore throats).Continuous pore throat networks of increasingly largerdiameter are shown in Figure 5b (3, 4, and 5 units) andFigure 5c (2, 3, 4, and 5 units). Note that the percentage ofexit pore throats increases to 50% and 80%, respectively.The connection of all pore throats (1 to 5 units) representsa total flow condition, such as is expected for a reservoirrock rather than a seal.

Chapter 11—Migration—A View from the Top 143

Figure 3—Comparativesizes of shale pores andmolecules of hydrocar-bon, water, and kerogen.

The relationship between net path permeability andfrequency of occurrence of a given path is summarized inFigure 6. The net path permeability of a given pore net-work is determined by the smallest pore throat in thatpath. Net paths with small permeability are thereforemore common than paths with large net permeability,and the longer any particular path is, the greater thechance it will include a pore throat smaller than it cur-rently contains. A three-dimensional model would havesimilar behavior, but the potential connections to largepore throats would rise (connections to other small porethroats have no effect), increasing the probability of alarge pore network. This pore network is a modificationof Momper’s (1978), Lindgreen (1985), and England’s(1987) expulsion models.

The connectivity of these pore throats governs theextent to which hydrocarbons, either as separate phasesor dissolved molecules, are retarded by the seal. Theusual situation is the sealing of a separate phase becausemost shales have pores in the ranges shown in Figure 3.To examine the flow of separate phases, the pore throatsin Figure 5 are considered as fixed and the diameter of thecurvature at the hydrocarbon–water interface is allowedto vary among the three diagrams (Figure 5a, b, c). At aconstant interfacial tension, capillary pressure dependson the curvature of the interface between the hydrocar-bon phase and water. The radius of curvature is deter-mined by the diameter of the pore throat through whichit must pass and the difference in pressure across thatpore throat. If that pressure is small and the pore throat issmall, there cannot be any separate phase hydrocarbon

flow through the unit. The unit thus acts as a perfect seal,ignoring losses due to noncapillary processes.

As the pressure rises, however, the radius of curvatureat the pore throat interface decreases until a point isreached when hydrocarbons move through the largestcontinuous pore throat network present (Figure 5a). Ifthis flow balances the rate of pressure buildup, a steadystate arises between pressure buildup below the seal andleakage through the seal. If, however, the rate of pressurebuildup continues, the curvature of the interface at thepore throats will continue to decrease and flow will takeplace through increasingly small pore throats (Figures 5b,c). The heterogeneous nature of seals thus allows them toadjust their rates of hydrocarbon leakage dynamically.

One of the principal causes of pressure build up inreservoirs is the buoyancy of the hydrocarbon phase. As

144 Matthews

Figure 4—Graph of permeability in Tertiary shales withdepth. (After Neglia, 1979.)

Permeability (darcys)

De

pth

(km

)

(c)

(b)

(a)

Figure 5—Diagrams of a pore network showing five possi-ble pore throat sizes from 1 to 5 (smallest to largest, frombottom to top). (a) Pore throat sizes 4 and 5, with net-works shown in black. (b) Pore throat sizes 3 and larger,with networks of sizes 4 and 5 shown in black and size 3shown in vertical line pattern. (c) Pore throat sizes 2 andlarger, with networks of sizes 3, 4 and 5 shown as beforeand size 2 shown in horizontal line pattern.

the reservoired column increases in height, the pressureat the interface increases, decreasing the radius of curva-ture at the interface and increasing the leakage throughthe seal. Similarly, as the rate of reservoir charging slowsdown, the rate of leakage also slows down. This model isessentially in agreement with laboratory simulations ofseparate-phase flow in porous medium by Catalan et al.(1992) and Dembicki and Anderson (1989) and withSchowalter’s (1979) observation that continuous fila-ments are formed only if the hydrocarbon concentrationexceeds 4.5–17% of the pore volume. Examination ofgaseous hydrocarbons in the cap rock of Snorre field inthe North Sea (Lieth et al., 1993) indicates that, althoughdiffusion plays a role in loss from the reservoir, bulk flowdue to buoyancy is probably the dominant process.

On rare occasions, seals may be dominated by finepore throats, in which case, the rock acts like a molecularsieve. For dissolved phases, the size of the molecules isfixed. Therefore, the pore throat units are considered toincrease from Figure 5a to 5c. A rock dominated by finepore throats is diagramatically represented by Figure 5a.If we consider the pore throats of 4-unit width to repre-sent 0.5 nm, the seal will slowly pass methane, and per-haps some ethane, due to the small percentage of avail-able paths. The case in which the large pore (4-unit)throats are 0.2 nm is represented by Figure 5b. The largernormal paraffins and smaller aromatics can pass throughthe single large pore (4-unit) network, and methane andthe smaller normal paraffins can pass through the morenumerous 3-unit pore networks as well as the larger net-work. In Figure 5c, the large pore (4-unit) network isabout 0.4 nm, the normal size of median pores. Underthese conditions, the large pore network permits the pas-sage of methane, the entire range of normal paraffins andaromatics, and all but the largest asphaltenes. The small-er pore networks pass the smaller hydrocarbons at ahigher rate because of their greater frequency. Lindgreen(1987) documented molecular sieving in fine pore sys-tems in source rocks and estimated that only one-fifth ofthe total pore volume was capable of transmitting normalparaffins.

Fluid Inclusions

Fluid inclusions in diagenetic cements are samples ofpaleopore fluids that were trapped during precipitationof that cement. By incorporating the composition andPVT properties of the inclusions with geologic historyand the diagenetic sequence in the area, a series of snap-shots of the pore fluid characteristics as a function of timecan be obtained (Burruss et al., 1983). These studies showthat reservoir charging occurs in a discrete time interval,either as a series of events or as a continuously evolvingprocess (Karlsen et al., 1993). Jensenius and Burruss(1990) found that some oils in inclusions seem to be a mix-ture of a biodegraded hump of C12 to C30 compoundsand water-soluble low carbon number compounds. Theysuggest that the low carbon number component mayarise due to exsolution from a later warm aqueous phaserising from depth. McLimans (1987) suggests that hydro-carbon inclusions in reservoirs originate from a separatephase consisting of drops from 1–40 µm in diameter (therange of oil in water emulsions) (Figure 3), sufficientlydispersed to move freely within the reservoir pore space.

Phase Behavior

Figure 7 shows the phase behavior of hydrocarbonswith depth. The exact configuration of this diagram is afunction of the particular mixture of hydrocarbons pre-sent; for simplicity, a single composition is considered. Asproducts are expelled and migrate vertically, phase parti-tioning occurs. Consider an expulsed oil at a depth of 5km with 30% gas. This oil migrates vertically without anysignificant changes until it reaches its bubble point atabout 4 km. At this point, both a gas and oil phase coex-ist. The single gas bubble would contain about 25% dis-solved oil. If these two phases continued to maintain con-tact and migrate vertically, the proportion of gas in the oilphases would decrease and the proportion of oil in thegas phase would similarly decrease. At a depth of 2 km,the gas phase would contain about 5% oil and the oilphase would contain about 15% gas, but the total propor-tion of gas would still be 0.3. Silverman (1965) combinedphase behavior with selective trapping to produce a “sep-aration-migration” mechanism.

If the two phases separated at 3 km and continued tomigrate vertically, at a depth of 2 km, the migrated gasphase would have 85% gas and the migrated oil phasewould have 20% gas. The migrated gas and oil phaseswould both have a gas cap with 10% dissolved oil and oillegs with 5% gas, identical to the original unseparatedproduct. The amounts of gas caps and their chemicalcompositions would, however, be different from oneanother and from the gas cap of the unseparated product.Separation-migration can significantly alter the grosscomposition of the migrated and trapped intervals. Thegeneral trends with migration from depth are as follows:hydrocarbon liquids lose low molecular weight com-pounds to a gaseous phase, becoming more dense, andhydrocarbon gases lose high molecular weight com-

Chapter 11—Migration—A View from the Top 145

Figure 6—Graph of expected frequency of net path permeability.

pounds to a liquid phase, becoming less dense (Englandet al., 1987).

Price et al. (1983) showed experimentally that suffi-cient oil can be transported in gas solution to charge reser-voirs. Thompson (1988) documented that migration frac-tionation is a viable mechanism to explain the geochem-istry of the Gulf Coast and more than half of all other U.S.oils. Migration fractionation involves reservoired oilbeing partially dissolved in an excess of methane, subse-quent loss of the gaseous phase, leaving behind a residualoil with elevated light aromatics. The leaked gaseousphase condenses as pressure and temperature arereduced during vertical migration, forming retrogradecondensates.

Hydrocarbon Solution in Water

Figure 8 shows the aqueous solubility of normal alka-nes at 25˚C. The solubility decreases linearly with increas-ing carbon number, up to C12. Normal alkanes with car-bon numbers greater than 12 show a much higher solu-bility than would be expected by extrapolation of thelower carbon number data. This may represent an

approach to a solubility threshold or the microdroplets inthe laboratory studies (Tissot and Welte, 1984). This com-positional relationship differs significantly from thatfound in most reservoired oils. There are, however, a fewlight oils that show this relationship (Tissot and Welte1984; Quanxing and Qiming, 1991). The occurrence ofthese light oils as a separate phase demonstrates thatsolution transport does occur and that it can reach condi-tions causing it to separate into a separate phase in con-siderable quantity. The infrequent occurrence of oils withthis signature suggests that this process of migration is anexception rather than the rule.

DiffusionStudies of source and reservoir contacts show that dif-

fusion of dissolved hydrocarbons occurs in the subsur-face. Connan and Cassou (1980) and Leythaeuser et al(1984) have studied the relationship of sandstone andshale extracts in immature terrestrial organic sequences.They have shown that small quantities of hydrocarbonshave been selectively depleted in the shales and accumu-lated in the sandstones. The greatest depletion in theshales occurred near the sandstone contacts while thegreatest accumulation in the sandstone was associatedwith fractures. Paraffins, particularly the lighter com-pounds, were preferentially migrated. The sandstoneextracts resemble a condensate or light oil while thedepleted shale extract looked less mature and slightlybiodegraded compared to the center of thick shales.

The gradients shown in Figure 9 are explained by dif-fusion (Leythaeuser et al., 1983). However, benzene andtoluene (water-soluble aromatics) had anomalously hightransfer rates from the siltstone into the sandstone. Thissuggests that some transport of hydrocarbons may havebeen assisted by transport of water from the siltstone intothe sandstone.

146 Matthews

Figure 7—Graph of depth and temperature versus gas togas + oil ratio showing hydrocarbon phase behavior.(After Pepper, 1991.)

Figure 8—Graph of aqueous solubility versus carbonnumber for hydrocarbons at 25˚C. (After Tissot and Welte,1984.)

Rates of Hydrocarbon Motion

Estimation of rates of hydrocarbon migration fromsource to reservoir are difficult at best. Estimates of ratesof diffusion can be calculated theoretically, measuredexperimentally, and compared to concentration gradientsaway from concentrated sources. Active transport sys-tems are much more difficult to constrain due to theuncertainties in the time of expulsion and entrapment.

Diffusion of light hydrocarbons in a water-filledporous system is extremely slow. Figure 10 shows thetime it takes a light hydrocarbon to diffuse a given dis-tance and reach a concentration level equal to one-half theconcentration in an irreducible source. The larger diffu-sion coefficient is more typical of sandstone and thesmaller one is more typical of shale. This is consistentwith the findings of Zarrella et al. (1967) who showed thatdissolved benzene and toluene followed a diffusion gra-dient horizontally for miles in sandstone but was verti-cally absent in the over- and underlying sections. Thissuggests that diffusion is not a practical transport mecha-nism in shale, except for short distances or at its bound-ary with sandstone. England et al. (1987) calculated thatfor distances greater than 10–100 m, diffusion is insignifi-

cant relative to bulk flow. An exception to this is methaneand ethane. Diffusion would be expected to transportlight hydrocarbons only tens of meters into a shale cap-ping a reservoir. Despite this, significant quantities oflight hydrocarbons can diffuse into a cap rock (Krooss etal., 1992). If the shale was fractured, the diffusion coeffi-cient would only increase slightly , because the fractureswould occupy a small percentage of the shale. Diffusionof larger hydrocarbons is less likely as an appreciabletransport mechanism because the diffusion of hydrocar-bons is related to the size of the molecule (Figure 9).

Cathles and Smith (1983) estimated that periodic fluidmigration events greater than 1000 times the normal rateswere necessary to form Mississippi Valley type lead-zincdeposits by expulsion of basin brines. They suggest thatoverpressure conditions build to a point where the sec-tion fractures, releasing large volumes of fluid over shortperiods of time. As the fluids escape, the pressure isreleased and the fractures close. The pressures then buildup and the cycle repeats at intervals of about 1 m.y. atdepths of more than ~3000–4500 m (~10,000–15,000 ft).

The flux rate of hydrocarbon transport in the subsur-face is viewed as both parallel and series processes.Parallel processes are diffusive transport, aqueous trans-port in solution and as micelles, and separate phase trans-port. Series transport is the set of processes that are domi-nant sequentially along the most effective migration path.Any compositional constraint placed by any process(such as solution in water) can impact the characteristicsof the migrated product from then on. From a linear ratestandpoint, the least efficient process along this path is therate-limiting step, controlling the overall rate of theprocess. In a sequence of sandstone and shale, the ratedelimiter would be the least permeable shale. However,flux rates of a nonwetting separate phase, particularlythrough shales, are self-adjusting. This is accomplishedby changing the area through which the linear processoperates. Small dips in the section focus water and hydro-carbon flow laterally to an area where vertical flowbecomes dominant (Roberts, 1980). Free-phase hydrocar-

Chapter 11—Migration—A View from the Top 147

Figure 9—Graph of depth versus n-alkane fraction show-ing the concentration of hydrocarbons at the transitionfrom a source to a sandstone.

Figure 10—Graph showing the time needed for hydrocar-bons to diffuse a given distance and reach a concentra-tion of one-half the concentration of a nondepletedsource.

Fraction of n-Alkanes

bons accumulate in large pore systems underlying therestrictive shale, spreading out into a commercial or non-commercial accumulation. This increases the hydrocar-bon flux rate through the overlying shale by increasing itssurface area, making up for the low rate per unit area.

Mud Log Data

Mud log data are the most abundant subsurface geo-chemical data available. In general, they are an adequaterepresentation of vertical variations in light gases and flu-orescing compounds. Vertical trends in light gases arehighly variable, but generally show an increase as thesource-generative section is approached. They also showorders of magnitude changes in a few feet to tens of feetand rapid compositional shift of light gases and heaviercompounds. High values are generally related to theoccurrence of mature source rocks, top of overpressuredzones, faults, and reservoirs.

Detailed geochemical studies, including isotopicanalysis, often show a smoothly increasing gradient, con-sistent with increasing maturity. Superimposed on thisgradient are isolated spikes of hydrocarbons with matu-rities characteristic of much deeper conditions. This sug-gests migration from a deeper source, through an inter-vening section, with little mixing or other effects from theintervening section.

Seismic Evidence

Vertical chains of seismic bright spots have been seen inthe Gulf of Mexico, Nigeria, and other offshore areas. Inmany cases, the bright spots are associated with sand-stones adjacent to faults and are occasionally recognizedwithin fault zones. The most dramatic chain of bright spotsis at Ekofisk in the North Sea where they are associatedwith a gas chimney of poor data and a low-velocity sag atthe reservoir level (Van den Bark and Thomas, 1980).

LESSONS FROM NEAR-SURFACESTUDIES

Surface geochemical surveys represent the end of themigration pathway. As such, they offer the best opportu-nity for understanding the spatial patterns of migrationand cross-formational flow. The compositional and mag-nitude information is, however, subject to the boundaryconditions previously discussed. Therefore, caution mustbe exercised in comparing this information with deeperstudies. Of particular importance is the volume change offree gas associated with decreasing pressure as the hydro-carbons migrate toward the surface. This volume increasemakes the gas more visible to seismic profiling, makes itmore measurable through collection, and increases itsrate of motion due to buoyancy. Topics discussed in thissection include seepage patterns, compositional relation-ships to the subsurface, variation with time, and the rateof transport.

Seepage Patterns

Surface macroseepage occurs in areally restricted loca-tions and is direct evidence of hydrocarbon migration ina separate phase. Link (1952) studied worldwide occur-rence of macroseeps and concluded that they occur dom-inantly along high-permeability pathways such as faults,fractures, unconformities, and pore networks in outcrop-ping reservoirs. The linearity and spatial variability ofmacroseeps is particularly well illustrated by Preston(1980).

Microseep studies by Jones and Drozd (1983), Zorkinet al. (1977), and others often demonstrate the preferen-tial pathway concept, particularly the role of faults andfractures. Preferential seepage up outcropping reservoirrock and through the fractured crest of an anticline isdemonstrated by Matthews et al. (1984). The role offaults in focusing seepage is demonstrated by Jones andThune (1982), Matthews et al. (1984), Burtell et al. (1986),and Abrams (1992). Wakita et al. (1978) has shown thatgas microseepage up faults occurs in spatially distinctareas (“spots”), similar to that shown for gas macroseep-age by Preston (1980). In macroseepage, there is a clear,visible distinction between where seepage is occurringand where it is not occurring. The replacement of visualobservation with sensitive instrumentation removes theability to conveniently distinguish areas of no seepagefrom seepage. Instead, there is a seepage continuumfrom the smallest level the machine can detect to visibleseepage.

The dominant mechanism of seepage at large-magni-tude sampling locations is the same—bulk flow of a sep-arate phase. Microseepage magnitudes of light hydrocar-bons in soil air vary from atmospheric levels, to elevatedmoderate levels, to percent levels. Atmospheric levels insoil air undoubtedly represent as close to no seepage aspossible. Percent levels represent massive active seepageas a gas phase. Moderate levels also exist and are ambigu-ous in their mode of origin and their significance. Theymay represent the end of a dissolved phase migrationpath or diffusive mixing of a moderately active gas phasewith atmospheric air in the soil profile.

Experience shows, however, that when closely spacedsamples are taken, the boundaries between high- and low-magnitude sites are usually less than 400 m and often justa few meters. In one instance, soil gas magnitudes variedseveral orders of magnitude within 100 m. The boundarywas recognized to occur over a distance of centimeters bythe presence of bubble formation at the soil surface as rainbegan to saturate the soil and by the drying pattern whenthe rain had ceased (V. T. Jones, personal communication,1978). The spatial proximity of surface hydrocarbon con-centrations that differ by orders of magnitude, in bothland and marine surveys, requires effusive transport of aseparate phase, not solution transport.

High-resolution marine seismic surveys reveal thepresence of shallow gas in the upper 600 m of sedimentand up into the water column. The existence of consider-able quantities of shallow gas has been proved manytimes by the collapse of offshore platforms. Analysis of

148 Matthews

hazard surveys by Salisbury (1990) suggests two path-ways, fault planes, and localized pervasive vertical leak-age (gas chimneys). Where the vertical seepage is inter-rupted by a competent seal, horizontal migration alongsiltstone and sandstone layers is observed.

This pattern of migration is consistent with theonshore pattern of gas migration over a leaking gas stor-age reservoir at Mont Belview, Texas, studied by GulfResearch. Bubble trails in the water column, revealed byecho sounders and side-scan sonar, indicate that in manycases the transport of gases is an active process (Judd,1990). Side-scan sonar shows that these gas plumes in thewater column often originate from pockmarks ormounds on the sea floor, arranged along what appear tobe the surface trace of faults. McQuillin and Fannin (1979)report occasional plumes of sediment in the water col-umn up to heights of 10 m from the sea bed. Numeroussniffer surveys by Gulf Research and others have docu-mented the association of these plumes with both ther-mogenic and biogenic hydrocarbons. In a few cases,unidentified nonhydrocarbon gases were also observed.Gravity core studies (Kennicutt et al., 1988) and analysisof collected bubbles reveals that oils reach the sea floor asa separate phase in considerable quantities (Clarke andCleverly, 1991).

Compositional Relationships withSubsurface

Jones and Drozd (1983) and Williams et al. (1981)demonstrated that light gas compositions in the surfaceand subsurface are similar in broad compositional classesfor both soils and marine waters. The majority of oil seepsresemble reservoired hydrocarbons (Kennicutt et al.,1988). These correlations indicate bulk flow with little orno modification by other processes along the migrationpathway to the surface. Numerous studies by Gulf andTexaco Research on land and in the oceans demonstratethe spatial proximity of zones of high-magnitude compo-sitionally dissimilar samples that are correlatable to thecomposition of reservoired hydrocarbons in the subsur-face. These relationships indicate that multiple pathwaysfrom depth exist and that often little or no mixing or alter-ation occurs along these paths. However, Illich et al.(1984) reported an oil seep that is compositionally similarto that expected from solution transport. The occurrenceof an oil phase of this composition at the surface maymean that it leaked from a subsurface accumulationformed by solution transport or that it evolved from solu-tion transport near the surface.

Sweeney (1988) found that in the organically lean shal-low sediments of the North Sea, both diffusion and free-phase transport of thermogenic hydrocarbons occurs.The diffusion transport of these hydrocarbons to the nearsurface was effectively eliminated by bacterial oxidationin the sediment. The free-phase transport, however, haslimited contact area with the bacterial zone, resulting inleakage to the sediment–water interface. Where the

process is efficient, a continuous stream of bubbles movesfrom the sediment into the water column. Where theprocess is inefficient, a patchy distribution of shallow gaspockets is formed. These pockets increase in size untilthey overcome the capillary restrictions of the sedimentor, more likely, catastrophically break through to thewater column by rupturing the sediment.

Variation with Time

Surface seepage at any particular location is time vari-able. Attempts to reoccupy a previous sampling site usu-ally result in merely being close to the original locationand confounding spatial changes with temporal changes.Broad patterns hold over time, provided the subsurfaceconditions do not change significantly within that timeinterval. The survey of Dickinson and Matthews (1993)was performed over two field seasons, but the pattern ofanomalies held between the two surveys. In subsequentyears, other surveys were undertaken enlarging the sur-vey area. These surveys extended edge anomalies of theprevious surveys seamlessly. There is, however, a changein surface magnitudes that is recognized in a longer timeframe. Significant reservoir pressure decreases related toproduction lower the magnitudes of surface gases(Horvitz, 1969).

Permanent stations remove the uncertainty of spatialpositioning in temporal sampling. These stations showremarkable repeatability, usually less than a factor of ten,if human activity is minimal. The near surface, however,is not a static system. Variations in surface hydrocarbonconcentrations are linked to atmospheric pressurechanges and earthquakes. Barometric pressure changespropagate downward in excess of 150 m (Figure 11)(Meents, 1958). This periodic pressure change is capableof causing minute expansions and contractions ofgaseous free phases within water-wet systems. Seismicevents also cause short-term variations (in hours) inhydrocarbon concentration (Zorkin et al., 1977; Antropov,1981; Burtell, 1989).

Rate of Transport

Linear rate measurements at the surface are seldommade because of the uncertainty associated with depth oforigin and cross-sectional area. Linear rate estimates inthe upper 200 m are on the order of tens of meters per day,based on known times of injection of gas into storagereservoirs and subsurface coal burns (Jones and Thune,1982). Arp (1992) summarized estimates of vertical seep-age velocities as 75–300 m per year. Volume rate estimateshave been made in the marine environment by collectingbubbles. The average oil seep flow rate compiled byClarke and Cleverly (1991) is 50 m3 (300 bbl) per year.These rates clearly indicate separate phase migrationalong narrow migration pathways. Diffusion rates andsolution transport are much slower but operate overmuch larger areas.

Chapter 11—Migration—A View from the Top 149

CONCLUSIONS

Much variation exists in the characteristics of subsur-face and surface hydrocarbons. Since it is clear that a sin-gle mechanism is incapable of explaining these varia-tions, we must conclude that hydrocarbon migrationoccurs by many mechanisms and each case may or maynot have significant local effects. Thus, our task is to dis-cover the dominant pattern of migration and the natureof the exceptions. It is my opinion that the dominantmigration mechanism is as a free phase interacting with aheterogeneous rock framework.

Except in rare instances, separate phase migrationoccurs along the entire preferential migration path, fromsource to surface. The only significant process responsiblefor modifying the composition of the migrating hydro-carbons is phase separation, aided by the characteristicsof the seals along the migration path. The success of geo-chemical source–reservoir correlation techniques is thestrongest of many arguments indicating that the domi-nant migration mechanism is as a free phase and not assolution. Solution transport, both by diffusion and activewater transport, does occur and probably represents thelargest reservoir of light gas in the subsurface. It is domi-nantly a dispersive mechanism, selectively removing themore soluble compounds from hydrocarbon masses,both stationary and in transport, leaving what have beencalled water-washed accumulations. The kilometer-longdiffusion gradients of benzene and toluene within reser-voirs demonstrate the effectiveness of this process.

The lack of significant transport of benzene andtoluene across adjacent shales argues against both diffu-sion and active aqueous solution transport as a dominantmechanism of accumulation. The rapid variation withdepth of mud log composition and magnitude indicatesthat diffusion generally does not extend smoothly to thesurface or great distances vertically. The ability to concen-trate hydrocarbons by solution, either through diffusionor advection, is also limited by the constraints of low sol-ubility and the limited supply of compaction water atmaturation depths. The saturation of water by hydrocar-

bons and subsequent migration is, however, supportedby a few instances of surface measurements characteristicof solution transport. The existence of a few oil fields withcompositions similar to that expected from solutiondemonstrates that accumulations may form by thisprocess, but they are the exception, not the rule. Mostreservoired hydrocarbons have compositions that arecontrolled by the characteristics of their source rock, notthe processes of their transportation. Diffusion of lighthydrocarbons into cap rocks explains the abundance ofreservoirs undersaturated with light gases.

Micellar transport of hydrocarbons is not considered adominant mechanism. The amount of solubilizer in themicelle is many times the amount of solubilized hydro-carbon. These solubilizers are seldom detected in reser-voirs and source rocks in abundant quantities. Transportof hydrocarbons along a kerogen network is limited torocks with TOC contents over about 1%. These constitutea small percentage of most migration paths. It may be animportant expulsion mechanism within source rocks, butit would seem to render these source rocks relatively inef-fective as a top seal. Since many source rocks do act as topseals, the mechanism is thought to be an exception.

Selective adsorption of hydrocarbons has been pro-posed to occur along the migration path. This is notthought to be a significant factor. The movement of a freephase through the pore network minimizes contact areaand time, reducing the opportunity for adsorption. Thepore network does not act as a chromatographic column,except as one that is saturated. Only in the first and laststages of a migration event would any chromatographicseparation be noticeable.

General Model

Separate phase migration is significantly effected bythe heterogeneous nature of rocks, particularly those clas-sified as seals. The heterogeneity of shales is such that,over a large area, capillary restrictions are believed torarely be 100% effective. In most cases, there is at least onepathway that will leak, albeit slowly. As the accumulation

150 Matthews

Figure 11—Graph show-ing atmospheric-drivenpressure changes in agas sandstone at ~150 m(500 ft) depth. (AfterMeents, 1958.)

grows, the number of leakage pathways also grows, bothbecause of increased buoyancy pressure opening up pre-viously restricted pathways and because the increasedareal extent of the growing accumulation intersects otherleakage pathways. Figure 5, previously discussed from apore system viewpoint, also serves as a model for macro-scopic leakage behavior. Depositional facies are not inter-nally random; there is structure to their deposition andspatial characteristics. They are fractile in nature. Thus,there are local concentrations of large pore throat net-works and more extensive regions dominated by smallpore networks. This translates into areally restrictedregions of high flux rate and broad regions of almost noflow due to capillary heterogeneity. The rate-limiting stepin postexpulsion migration of hydrocarbons is the transi-tion from coarse to fine pore throat systems. This systemis self-adjusting. A consequence of this process is thedynamic formation, maintenance, and destruction of oiland gas fields.

Postexpulsion ModelThe motion of hydrocarbons during postexpulsion

migration involves four repetitive conditions. First, ascapillary imbibition causes hydrocarbons to move into acarrier from discrete pore networks in the source or a seal,they thin and separate into bubbles or droplets. Second,within the carrier bed, buoyancy (assisted by gas expan-sion) is a major control on migration. Water velocitiesgenerated by compaction are generally insignificant com-pared to buoyancy-driven velocities and as a first approx-imation can be ignored. In these cases, lateral transport ofhydrocarbons is caused by the dip of the contact betweenthe seal and the carrier bed. Water velocities due to topo-graphic driven flow are, however, significantly higherand can locally dominate buoyancy. This can result in sig-nificant lateral transport due to water flow, both updipunder the seal and downdip (Hubbert, 1953). Once atrapping configuration is reached, the hydrocarbonsaccumulate under a seal. At this point, the hydrocarbonsbecome a coherent slug and cease to act as discretedroplets. Buoyancy pressure increases as the mass grows.Third, the flow of hydrocarbons from a reservoir into aseal is dominated by the penetration of a separate phase.This is accomplished by buoyant pressure overcomingcapillary restriction.

Finally, within the seal, the hydrocarbons movethrough discrete pathways of one or more continuousthreads until they encounter a carrier bed. At this point,capillary imbibition once again becomes dominant andthe first condition (above) is repeated. This breaks thethread at each carrier unit, constraining the buoyant forceto reservoir–seal couplets. As the hydrocarbons movevertically through the section, they continuously reequili-brate into gas and oil phases. These phases may stay incontact, but are often partitioned into different accumula-tion sites due to differences in migration pathways(spillage of oil and capillary leakage of gas or oil) and thevector differences in gas and oil when water flow is sig-nificant.

The rates of leakage through the sequence of sealsbetween the source rock and the earth’s surface controlthe pattern of hydrocarbon accumulations along the pos-texpulsion migration route and both the magnitudes andspatial properties of surface seepage. This complex sys-tem is idealized into three successive seals for purposes ofdiscussion (Figure 12).

Shortly after hydrocarbons begin to be expelled fromthe source rock (time 0–2, Figure 12) they begin to appearat the surface in low-magnitude, spatially dispersed sites.This results in an elevated background population withoccasional spikes (abnormally high magnitudes) causedby preferential migration pathways extending from thesource rock to the surface. Some of the spikes are locatednear the crest of structures, while others are scatteredalong the surface projection of migration pathways tothese structures. Once the hydrocarbons reach the upperboundary of the first carrier unit above their source (sealA), they migrate to a trapping configuration (time 0–1,Figure 12). The hydrocarbons accumulate there, buildingup both buoyancy pressure (because of the increasedheight of the hydrocarbon mass) and water pressure(because the hydrocarbons restrict the area the water hasto flow through). If seal A has many large pore networks,only a small pressure gradient is required to carry hydro-carbons through it, leaving a small amount behind. If thevolume rate of charge into the carrier is greater than canbe accommodated by the large pore networks at the topof the trapping configuration, both hydrocarbon heightand areal extent increase until an accommodation ofreservoir charge and leakage is accomplished, and thereservoir enters a steady-state condition (time 1–6, Figure12). Leakage through seal A increases steadily as pressureopens new pathways and the lateral expansion of thehydrocarbon mass encounters new pore networks capa-ble of leaking. The accumulation below seal A disappears(time 6–8, Figure 12) as the rate of charging falls below therate of leakage.

The hydrocarbons leaking through seal A migrate inthe new carrier unit until they reach a trapping configu-ration under seal B (for convenience, the finest seal in thesystem). A large mass of hydrocarbons can be accommo-dated under seal B. As this reservoir begins to grow (time1–4, Figure 12), both in size and thickness, the rate of leak-age also grows due to increased pressure and increasedopportunity. The reservoir may even fill to the spill point(time 4, Figure 12). As leakage through seal B increases,regional surface seepage increases in magnitude andnumber (times 2–4, Figure 12). Surface locations at theend of preferential migration pathways connected to thetrapping configuration under seal B markedly increase inspike spatial density due to the subsurface concentrationof hydrocarbons that are beginning to form coherent clus-ters of higher magnitudes rather than isolated spikes.When seal B reaches maximum capacity (time 4, Figure12), the rate of leakage is controlled by the rate of expul-sion from the source rock.

The hydrocarbons leaking through seal B (time 2,Figure 12) progress through a carrier system until they

Chapter 11—Migration—A View from the Top 151

encounter seal C. For illustrative purposes, seal C onlyaccumulates hydrocarbons under it when the leakagerate from B is near a maximum (time 4, Figure 12). Atlower leakage rates from B, seal C is essentially transpar-ent to migrating hydrocarbons. Surface leakage grows inmagnitude, extent, and density as long as the reservoirunder seal B is growing or being charged beyond capaci-ty (time 2–5, Figure 12) and will reach a maximum whenreservoir C is fully charged (time 5). From time 5 to 6 inFigure 12, the reservoirs below all three seals are chargedto maximum capacity and surface seepage is also at amaximum, being controlled by the rate of expulsion fromthe source rock. During these times, surface seepage mag-nitudes are at a maximum and the clustering of thesespikes is best developed, in response to migration up themaximum number of preferential pathways connectingthe surface to the accumulations. When the rate of expul-sion from the source rock falls below the rate needed tomaintain the accumulation below seal A (after time 6,Figure 12), the accumulation below seal B is at first main-tained by a high rate of leakage of hydrocarbons storedbelow seal A (time 6–7, Figure 12).

Surface seepage remains constant during this timewith high magnitudes and a well-developed clustering ofspikes, supported by equilibrium leakage from the accu-mulations below seals B and C. The accumulation belowseal B decreases in thickness and areal extent as chargingfrom below decreases and becomes insignificant (aftertime 7, Figure 12). The rate of leakage through seal Bsimultaneously decreases as the reservoir diminishes.Surface seepage decreases slightly at his time (time 7–8,Figure 12), with the magnitudes diminishing and theclustering of spikes beginning to degrade as more andmore preferential pathways are abandoned due to capil-lary forces overcoming the upward pressure of the hydro-carbons. Seepage is maintained by the excess rate ofcharging below seal C. Once the accumulation below sealC begins to decrease, the rate of surface seepage is sup-ported by the steady-state seepage through seal C (time8–9, Figure 12). The surface pattern of spike clusteringbegins to degrade and the magnitudes decrease slightly.Once the accumulation below seal C is depleted, the sur-face seepage falls sharply (time 9, Figure 12) and is sup-ported entirely by the low rate of seepage through thebest seal of the sequence (seal B). Seal C is totally trans-parent at these seepage rates. As the accumulation belowseal B winds down, the surface seepage does also,decreasing in magnitude and both spatial coherence andfrequency of spikes, once again returning to low levelswith scattered low-magnitude spikes, indicating a gener-al lack of focused migration of hydrocarbons.

As expulsion stops, the system winds down. As eachsuccessive reservoir ceases to leak, the residual hydrocar-bons in the reservoirs and along the migration pathwaysare purged by diffusion and active transport in solution bymigrating waters. Where temperature and other biologi-cal requirements permit, biodegradation also assists theseprocesses. Indeed, these processes are active from themoment maturation starts, but are not dominant during

152 Matthews

1 2 3 4 5 6 7 9

Surface Leakage

Rate Deliveredto C

Rate Deliveredto B

Rate ofExpulsion

CumulativeExpulsion

0 8Time

AccumulationBelow C

AccumulationBelow B

AccumulationBelow A

Figure 12—Schematic diagram of the postexpulsionmodel showing the evolution of subsurface accumula-tions and seepage from a source through three seals (A, B, and C) to the surface. See text for details.

active expulsion, when separate phase migration over-whelms their abilities. As the hydrocarbons move intolower temperature and pressure conditions, phase separa-tion proceeds, partitioning compounds into gaseous andliquid phases. During postexpulsion migration, these twophases may be physically separated by a combination oftrapping configurations and seal behavior.

These processes, combined with the potential for mul-tiple source rocks, at different levels of maturity, and atdifferent migration routes to different reservoirs and mix-ing within the same reservoir, can lead to a very complexsystem with spatially distinct and mixed surface hydro-carbon compositions. The preferential occurrence ofhydrocarbon-filled inclusions along fractures and at spe-cific diagenetic stages supports the hypothesis thathydrocarbon migration occurs in a relatively narrow timeinterval. The combination of a relatively narrow migra-tion route with a short time interval minimizes contact ofthe migrating hydrocarbons with water during migra-tion. Thus, hydrocarbon migration can be thought of as“slug” flow, an idea supported by the observations ofZarrella et al. (1967). If migrating oil were exposed towater over a long time with a large area-to-volume ratio,one would expect that much of the benzene and tolueneproduced in the source rock would have been removed.This does not seem to be the case, supporting some sortof solution-protected migration.

As a slug of hydrocarbons migrates through shallow-er hydrocarbons, the small contact area and time mini-mizes compositional changes in the migrating phase. Ifthe upward-driving force is greater than the driving forceof the shallow accumulations, the amount of mixing isslight. This is similar to an open hole completion: the for-mations with higher excess pressure generally dominateflow into the borehole. Thus, the general backgroundtrend of increased concentration and increased gas wet-ness observed in many mud logs is dominantly a productof the static local generation of unmigrated hydrocarbonswith no drive. The higher concentrations observed insandstones, with compositions similar to the overalltrend of compositional change in the well, represent local-ly generated hydrocarbons that have accumulated in asandstone near the source. The higher concentrations,with compositions dissimilar to the nearby backgroundcomposition and isotopic values suggestive of highermaturation conditions, represent migrated hydrocarbonsthat were generated from a deeper source. These activelymigrating hydrocarbons are the ones that make it to thesurface, retaining their compositional similarity to thereservoired hydrocarbons, despite the presence of inter-vening minor accumulations.

This theory predicts that at every coarse–fine interfacethe rate of transport slows to adjust itself with time to bal-ance input and output. Siddiqui and Lake (1992) pointout that one result of capillary trapping theory is that adifferent hydrocarbon column height should exist atevery coarse–fine grain transition and that this is general-ly not noted. Wells are usually drilled overbalanced.Despite this, mud logs show a wide range of concentra-

tion variation that may be evidence of this hydrocarbonconcentration. A well is essentially a one-dimensionalsampling of a three-dimensional field. It should not beexpected that all sandstone–shale transitions should havehydrocarbon concentration at them. Well-executed mudlogs show, however, many small variations in concentra-tion occur and that the higher concentrations are general-ly correlated with sandstones below shales.

Surface seepage is the end of the migration process.The narrowness of the migration pathways, from bothreservoirs and source rocks, to the surface and the speedof this migration maintain the chemical signature of themigrating hydrocarbons. Chemical and mechanical diffu-sion broadens and decreases the magnitude of seepage asthe separate hydrocarbon phase reaches the surface.Biogenic activity uses part of the diffused hydrocarbonseepage as a feedstock, altering its composition andadding biogenic methane. The presence of mature hydro-carbons at the surface of the earth is indisputable proofthat an active source rock exists in the subsurface. Thecomposition of these hydrocarbons is similar to thosegenerated by the source or modified by phase separationduring migration and entrapment. The spatial pattern ofseepage is controlled by the complexities of the migrationpathway to the surface. Thus, the challenge in joining sur-face geochemistry with subsurface information is defin-ing this subsurface migration pathway and improvingour ability to predict subsurface accumulations along thispathway.

Acknowledgments—I wish to thank Hollis Hedberg, TedWeismann, Ed Driver, Bill Glezen, Vic Jones, Dick Mousseau,Bob Pirkle, and numerous other workers at Gulf Research forthe many discussions that helped generate my interest in thistopic and formulate many of the opinions presented here. I par-ticularly wish to thank Grover Schrayer, Bill Roberts, andVaughn Robison for discussion and reviewing the manuscript.

REFERENCES CITED

Abrams, M. A., 1992, Geophysical and geochemical evidencefor subsurface hydrocarbon leakage in the Bering sea,Alaska: Marine and Petroleum Geology, v. 9, p. 208–221.

Antropov, P. Y., 1981, Laser gas analysis in solving geologicand production problems: International GeologicalReview, v. 23, no. 3, p. 314–318.

Arp, G. K., 1992, Effusive microseepage: a first approximationmodel for light hydrocarbons movement in the subsur-face: Association of Petroleum GeochemicalExplorationists Bulletin, v. 8, p. 1–17.

Barker, C., 1980, Distribution of organic matter in a shaleclast: Geochimica et Cosmochimica Acta, v. 44,p. 1483–1492.

Brooks, J. M., B. B. Bernard, and W. M. Sackett, 1978,Characterization of gases in marine waters and sediments,in J. R. Waterson and P. K. Theobald, eds., 7thInternational Geochemical Symposium: Association ofExploration Geochemistry, Golden, Colorado, p. 337–345.

Chapter 11—Migration—A View from the Top 153

Burruss, R. C., K. R. Cercone, and P. M. Harris, 1983, Fluidinclusion petrography and tectonic history of the Al AliNo. 2 well: evidence for the timing of diagenesis and oilmigration, northern Oman foredeep: Geology, v. 11,p. 567–570.

Burtell, S. G., 1989, Geochemical investigations at Arrowheadsprings, San Bernardino, and along the San Andreas faultin southern California: Master’s thesis, University ofPittsburgh, Pittsburgh, Pennsylvania.

Burtell, S. G., V. T. Jones, R. A. Hodgson, K. Okasa, M.Kuniyasu, and T. Ando, 1986, Remote sensing and surfacegeochemical study of Railroad Valley, Nye County,Nevada, detailed grid study: Fifth Thematic MapperConference, Remote Sensing for Exploration Geology,Reno, Nevada, Sept. 29–Oct. 2.

Catalan, L., F. Xiaowen, I. Chatzis, and F. A. L. Dullien, 1992,An experimental study of secondary oil migration: AAPGBulletin, v. 76, p. 638–650.

Cathles, L. M., and A. T. Smith, 1983, Thermal constraints onthe formation of Mississippi Valley-type lead-zincdeposits and their implications on episodic basin dewater-ing and deposit genesis: Economic Geology, v. 78,p. 983–1002.

Clarke, R. H., and R. W. Cleverly, 1991, Petroleum seepageand post-accumulation migration, in W. A. England and E.J. Fleet, eds., Petroleum migration: Geological Society ofLondon Special Publication No. 59, p. 265–271.

Connan, J., and A. M. Cassou, 1980, Properties of gases andpetroleum liquids derived from terrestrial kerogen at vari-ous maturation levels: Geochimica et Cosmochimica Acta,v. 44, p. 1–23.

Dembicki, Jr., H., and M. J. Anderson, 1989, Secondary migra-tion of oil: experiments supporting efficient movement ofseparate, buoyant oil phase along limited conduits: AAPGBulletin, v. 73, p. 1018–1021.

Dickinson, R. G., and M. D. Matthews, 1993, Regionalmicroseep survey of part of the productive Wyoming-Utah thrust belt: AAPG Bulletin, v. 77, p. 1710–1722.

Durand,B., P., Ungerer, A. Chiarelli, and J. L. Oudin, 1983,Modelisation de la migration de L’Huile: application àdeux exemples de bassins sédimentaires: Eleventh WorldPetroleum Congress, London, Chichester, v. 1, no. 3,p. 3–11.

England, W. A., A. S. Mackenzie, D. M. Mann, and T. M.Quigley, 1987, The movement and entrapment of petrole-um fluids in the subsurface: Journal of the GeologicalSociety of London, v. 144, p. 327–347.

Horvitz, L., 1969, Hydrocarbon geochemical prospecting afterthirty years, in W. B. Heroy, ed., Unconventional methodsin exploration for petroleum and natural gas: Dallas,Southern Methodist University Press, p. 925–940.

Hubbert, M. K., 1953, Entrapment of petroleum under hydro-dynamic conditions: AAPG Bulletin, v. 37, p. 1954–2026.

Hunt, J. M., 1984, Primary and secondary migration of oil, inR. F. Meyer, ed., Exploration for heavy crude oil and nat-ural bitumen: AAPG Studies in Geology 25, p. 345–349.

Jensenius, J., and R. C. Burruss, 1990, Hydrocarbon–waterinteractions during brine migration: evidence from hydro-carbon inclusions in calcite cements from Danish NorthSea oil Ffields: Geochemica et Cosmochimica Acta, v. 54,p. 705–713.

Jones, V. T., and R. J. Drozd, 1983, Prediction of oil or gaspotential by near-surface geochemistry: AAPG Bulletin,v, 67, p. 932–952.

Jones, V. T., and H. W. Thune, 1982, Surface detection of retortgases from an underground coal gassification reactor insteeply dipping beds near Rawlings, Wyoming: Society ofPetroleum Explorationists, SPE 11050, 24 p.

Judd, A. G., 1990, Shallow gas and gas seepages: a dynamicprocess?: Safety in Offshore Drilling, v. 25, p. 27–50.

Karlsen, D. A., Nedkvitne, T., Larter, S. R., Bjørlykke, K., 1993,Hydrocarbon composition of authigenic inclusions: appli-cation to elucidation of petroleum reservoir filling history,Geochimica et Cosmochimica Acta, v. 57, p. 3641–3658.

Kennicutt II, M. C., J. M. Brooks, and G. J. Denoux, 1993,Leakage of deep, reservoired petroleum to the near sur-face of the Gulf of Mexico continental slope: MarineGeochemistry, v. 24, p. 39–59.

Krooss, B. M., D. Leythaeuser, and R. G. Schaefer, 1988, Lighthydrocarbon diffusion in a caprock: Chemical Geology,v. 71, p. 65–76.

Leith, T. L., I. Kaarstad, J. Connan, J. Pierron, and G. Caillet,1993, Recognition of caprock leakage in the Snorre field,Norwegian North Sea: Marine and Petroleum Geology,v. 10, p. 29–41.

Leythaeuser, D., A. Mackenzie, R. G. Schaefer, and M. Bjørøy,1984, A novel approach for recognition and quantificationof hydrocarbon migration effects in shale-sandstonesequences: AAPG Bulletin, v. 68, p. 196–217.

Leythaeuser, D., R. G. Schaefer, and H. Pooch, 1983, Diffusionof light hydrocarbons in subsurface sedimentary rocks:AAPG Bulletin, v. 67, p. 889–895.

Lindgreen, H., 1985, Diagenesis and primary migration inUpper Jurassic claystone source rocks in North Sea: AAPGBulletin, v. 69, p. 525–536.

Lindgreen, H., 1987, Experiments on adsorption and molecu-lar sieving and inferences on primary migration in UpperJurassic claystone source rocks, North Sea: AAPG Bulletin,v. 71, p. 308–321.

Link, W. K., 1952, Significance of oil and gas seeps in worldoil exploration: AAPG Bulletin, v. 36, p. 1505–1541.

McLimans, R. K., 1987, The application of fluid inclusions tomigration of oil and diagenesis in petroleum reservoirs:Applied Geochemistry, v. 2, p. 589–603.

McQuillin, R., and N. G. T. Fannin, 1979, Explaining theNorth Sea’s lunar floor: New Scientist, v. 83, p. 90–92.

Matthews, M. D., V. T. Jones, and D. M. Richers, 1984,Remote sensing and surface hydrocarbon leakage:Proceedings of the International Symposium on RemoteSensing and the Environment, Third ThematicConference, Colorado Springs, April 16–19, p. 663–670.

Meents, W. F., 1958, Tiskilwa drift–gas area, Bureau andPutnam counties, Illinois: Illinois State Geological SurveyCircular 253, 15 p.

Momper, J. A., 1978, Oil migration limitations suggested bygeological and geochemical considerations, in C. Barker,W. H. Roberts, and R. J. Cordell, eds., Physical and chemi-cal constraints on petroleum migration: AAPG ContinuingEducation Course Notes 8, p. B1–B60.

Nakayama, K., and I. Lerche, 1987, Two-dimensional basinanalysis, in B. Doligez, ed., Migration of hydrocarbons insedimentary basins: Second IFP Exploration ResearchConference, Technip, Paris, p. 597–611.

Neglia, S., 1979, Migration of fluids in sedimentary basins:AAPG Bulletin, v. 63, p. 573–597.

Pepper, A. S., 1991, Estimating the petroleum expulsionbehavior of source rocks: a novel quantitative approach, inW. A. England and E. J. Fleet, eds., Petroleum migration:

154 Matthews

Geological Society of London Special Publication 59,p. 9–31.

Preston, D., 1980, Gas eruptions taper off in northwestOklahoma: Geotimes, October, p. 18–20.

Price. L. C., L. M. Wagner, T. Ging, and C. W. Blount, 1983,Solubility of crude oil in methane as a function of pressureand temperature: Organic Geochemistry, v. 4, p. 210–221.

Quanxing, Z., and Z. Qiming, 1991, Evidence of primarymigration of condensate by molecular solution in aqueousphase in Yacheng field, offshore South China: Journal ofSoutheast Asian Earth Sciences, v. 5, p. 101–106.

Roberts, W. H., 1980, Design and functionof oil and gas traps,in W. H. Roberts and R. K. Cordell, eds., Problems inpetroleum migration: AAPG Studies in Geology 10,p. 217–240.

Sajgo, C., J. R. Maxwell, and A. S. Mackenzie, 1983,Evaluation of fractionation effects during the early stagesof primary migration: Organic Geochemistry, v. 5,p. 65–73.

Salisbury, R. S. K., 1990, shallow gas reservoirs and migrationpaths over a central North Sea diapir: Safety in OffshoreDrilling, v. 25, p. 167–180.

Schowalter, T. T., 1979, Mechanics of secondary hydrocarbonmigration and entrapment: AAPG Bulletin, v. 63,p. 723–760.

Siddiqui, F. I., and L. W. Lake, 1992, A dynamic theory ofhydrocarbon migration: Mathematical Geology, v. 24,p. 305–327.

Silverman, S. R., 1965, Migration and segregation of oil andgas, in A. Young and J. E. Balley, eds., Fluids in subsurfaceenvironments: AAPG Memoir 4, p. 53–65.

Sweeney, R. E., 1988, Petroleum-related hydrocarbon seepagein a recent North Sea sediment: Chemical Geology, v. 71,p. 53–64.

Thompson, K. F. M., 1988, Gas-condensate migration and oilfractionation in deltaic systems: Marine and PetroleumGeology, v. 5, p. 237–246.

Tissot, B. P., and D. H. Welte, 1984, Petroleum formation andoccurence: Berlin, Springer-Verlag, 699 p.

Tóth, J., 1988, Groundwater and hydrocarbon migration, inW. Back, J. S. Rosenshein, and P. R. Seaber, eds.,Hydrogeology: Geology of North America, v. O-2,p. 485–502.

Van den Bark, E., and O. D. Thomas, 1980, Ekofisk: first ofthe giant oil fields in western Europe, in M. T. Halbouty,ed., Giant oil and gas fields of the decade 1968–1978:AAPG Memoir 30, p. 195–224.

Wakita, H., 1978, Helium spots: caused by diapiric magmafrom the upper mantle: Science, v. 200, April 28,p. 430–432.

Welte, D. H., and A. Yükler, 1980, Evolution of sedimentarybasins from the standpoint of petroleum origin and accu-mulation, an approach for a quantitative basin study:Organic Geochemistry, v. 2, p. 1–8.

Williams, J. C., R. J. Mousseau, and T. J. Weismann, 1981,Correlation of well gas analysis with hydrocarbon seeps:Proceedings of the American Chemical Society NationalMeeting, Atlanta, Georgia, March 29–April 3.

Zarrella, W. M., R. J. Mousseau, N. D. Coggeshall, M. S.Norris, and G. J. Schrayer, 1967, Analysis and significanceof hydrocarbons in subsurface brines: Geochimica etCosmochimica Acta, v. 31, p. 1155–1166.

Zorkin, L. M., S. L. Zabairaevi, E. V. Karus, and K. K.,Kilmetov,1977, Experience of geochemical prospecting inpetroleum and gas deposits in the seismically active zoneof the Sakhalin Island: Izvestiya Vysshikh UchebnykhZavedeniy. Geologiya i Razvedka, v. 20, p. 52–62.

Chapter 11—Migration—A View from the Top 155