+ All Categories
Home > Documents > Fluvial Sequences Low Gradient System

Fluvial Sequences Low Gradient System

Date post: 30-May-2018
Category:
Upload: naufal2010
View: 221 times
Download: 0 times
Share this document with a friend

of 14

Transcript
  • 8/14/2019 Fluvial Sequences Low Gradient System

    1/14

    Journal of Sedimentary Research, 2007, v. 77, 225238Research Article

    DOI: 10.2110/jsr.2007.024

    THREE-DIMENSIONAL SEISMIC ARCHITECTURE OF FLUVIAL SEQUENCES ON THE LOW-GRADIENTSUNDA SHELF, OFFSHORE INDONESIA

    Y. DARMADI, * B.J. WILLIS, { AND S.L. DOROBEK 1

    Department of Geology and Geophysics, Texas A&M University, College Station, Texas 77845-3115, U.S.A.

    ABSTRACT : The sequence stratigraphy of fluvialdeltaic strata deposited near the center of the wide, gradually subsidingSunda Shelf differs from standard highstand examples because low accommodation and low depositional gradients restrictedthe thickness of fluvialdeltaic sequences and the depth of lowstand fluvial incision. In this setting sea-level fluctuations mayhave had less influence on fluvial depositional patterns, and fluvial stratigraphy is more likely to be defined by changes in riverdischarge during gradual aggradation. The upper Muda Formation in West Natuna Basin, offshore Indonesia, is a 225-m-thick,dominantly fluvial succession that prograded onto the middle part of the Sunda shelf from Pliocene to Holocene. The succession

    changes upward from offshore shelf deposits, through a thin interval of deltaic deposits, a succession of fluvial deposits, andfinally a thin transgressive succession directly beneath the modern sea floor.A high-resolution seismic survey (17 km by 40 km in area) of Belida Field near the middle of the Sunda shelf provides

    exceptional images of changing fluvial architecture within five erosionally based, fluvial sequences. Sequences, each several tensof meters thick, are characterized by low-relief basal incisions that extend laterally onto interfluves cut by a network of gullies.Although a regressive stacking pattern of successive sequences is defined to some extent by an increase in the relief of basalerosion and a decrease in average sequence thickness, it is best defined by progressive increase in the extent of gully formationonto interfluve areas (indicating longer periods of interfluve exposure).

    Channel size tends to decrease upward within individual sequences, recording both a decrease in the size of the largestchannels and greater preservation of smaller tributary and floodplain drainage channels. The largest channels preserved directlyabove the better developed sequence boundaries tend to be more sinuous and to have more extensive inner-bend accretiondeposits than those that cut down from horizons higher within sequences. This suggests that incised channels had more streampower to erode their banks relative to the sediment loads they carried. Although mechanisms controlling the episodic shallowincision and then aggradation of these fluvial systems are difficult to constrain from variations observed in this small study area,

    the consistent thickness of the Muda Formation observed in regional 2D seismic sections is evidence against significantsyndepositional tectonic deformation or differential subsidence. Low regional gradients and great distance of the study areafrom the paleo-shelf edge (, 1000 km) are evidence against direct sea-level influence. Larger channel sizes directly abovesequence boundaries supports temporal changes in channel discharge, perhaps related to climatic changes within the drainagebasin. Recognition of episodic channel incision and bypass alternating with periods of floodplain aggradation influencespredictions of channel-deposit connectivity based on channel proportion.

    INTRODUCTION

    Vertical changes in channel patterns within fluvial successions havebeen widely inferred in sequence stratigraphic models of fluvial systems.Stratigraphic predictions are generally based on alluvial architecture

    models (Leeder 1978; Allen 1978; Bridge and Leeder 1979), combinedwith ideas about changes in sediment aggradation rates of nearshoresediments during changes in sea level (Shanley and McCabe 1991, 1993;Wright and Marriott 1993; Plint et al 2001; Olsen et al 1995; Legarreta

    and Uliana 1998; Posamentier 2001; see also recent review in Blum andTornqvist 2000). These models generally predict that fluvial gradientssteepen and floodplain widths decrease as aggradation slows and riverchannels incise during sea-level fall, leading to preferential preservation of channel deposits relative to overbank deposits and braided channelpatterns. Sea-level rise is predicted to decrease fluvial gradients and widenfloodplains as sediments aggrade and river incisions fill, leading to greaterpreservation of floodplain deposits and meandering channel patterns.Based on these general ideas, sandier fluvial successions have commonlybeen interpreted as deposits of braided river systems, and muddier fluvialsuccessions as deposits of meandering river systems.

    Studies of fluvial architecture in outcrop typically provide only isolatedtwo-dimensional views of fluvial architecture. Although vertical varia-tions in the proportion of channel to overbank deposits over tens tohundreds of meters are commonly observed, few examples have adequateage control to constrain changes in aggradation rate associated with

    * Present address: ConocoPhillips Indonesia, Gatot Subroto, Jakarta 12930,Indonesia

    { Present address: Chevron Energy Technology Company, 1500 LouisianaStreet, Houston, Texas 77002, U.S.A.

    1 Present address: Maersk Oil and Gas, Esplanaden 50, 1263 Copenhagen K,Denmark

    Copyright E 2007, SEPM (Society for Sedimentary Geology) 1527-1404/07/077-225/$03.00

  • 8/14/2019 Fluvial Sequences Low Gradient System

    2/14

    observed upsection changes in the proportion of channel deposits. It isalso generally difficult to define changes in channel pattern from vertical

    cross sections (Bridge 1985). The result is that most stratigraphic studiesof ancient fluvial deposits are strongly inferential with regards to linksbetween channel-deposit abundance, aggradation rate, and channelpattern and few studies provide robust tests of these ideas.

    Recent studies of Quaternary fluvial systems that have been extensivelycored anddated (e.g.,Blum 1993;Tornqvist 1993; Blum et al 1996; Tebbenset al. 1999; Tornqvist et al. 2000; Blum and Tornqvist 2000) are beginningto provide a more complex picture of controls on fluvial architecture.Variations in sea level and climate are better known for these younger timeintervals, and thus extrabasinal controlling factors are better constrained.In some cases, these studies have also documented temporal changes inchannel patterns based on examination of abandoned river courses orpreserved terraces (e.g., Starkel 1991; Blum et al. 1995; Aslan and Autin1999). One important insight from Quaternary studies is that climaticchanges that affect fluvial water and sediment discharge are probably asimportant as sea-level variations in defining alluvial architecture. Forexample, Blum and Lancaster (2005) showed that deepest incision of theMississippi valley occurred during the beginning of eustatic sea-level rise asmelting continental glaciers within the drainage basin significantlyincreased high river discharges. Khadkikar and Rajshekhar (2005)indicated that rivers incised along the western coast of India during themost recent period of eustatic sea-level risedue to increased river dischargesrelated to intensification of monsoon rainfall. A limitation of Quaternarystratigraphic studies is that for the most part they give insight only intofluvialdeltaic response to the very rapid sea-levelchanges that characterizethe last eustatic cycle, and that changes in fluvial architecture are defined

    only by isolated cores and river cutbank exposures (rare exceptions to thelatter include the unprecedented coring studies of the RhineMeuse delta,

    summarized in Berendsen and Stouthamer 2001).Improvements in 3D seismic acquisition and processing show promise inproviding new views of changingalluvial architecture.Although 3D seismicrecords do not show the details of facies variations that can be documentedin large outcrops, nor are they constrained by the resolution of datingachieved in some studies of Quaternary deposits, they have vertical andlateral extent and continuity that is far beyond that which can be obtainedusing other methods. Most importantly for analysisof alluvial architecturalmodels, 3D seismic volumes can be used to link records of vertical changesin channel-deposit abundance and incision depth with planview changes inchannel-deposit width, distribution, and pattern. The few published studiesof fluvial deposits documented by modern 3D seismic data showmagnificent views of complex channel-body geometries and floodplainfeatures imaged in horizontal time slices (e.g., Burnett 1996; Carter 2003;Posamentier 2001; Miall 2002; Zeng and Hentz 2004). Additional studiesare needed that better link vertical variations with planview changes of fluvial deposits before the full potential of these new data in assessingalluvial architecture models will be realized.

    This study is based on a 3D seismic survey of Belida field in the WestNatuna Basin, South China Sea (Fig. 1). The focus is on the Pliocene toHolocene river-dominated, upper Muda Formation, which containsfluvial features that are well imaged in this seismic volume. The goal of this study is to relate stratigraphic variations defined by vertical changesin channel size and abundance with planview changes in the distributionand pattern of observed channel bodies and evaluate possible factors thatmay have caused these changes.

    F IG. 1.Ancient drainage patterns across thesouthwestern South China Sea (Sunda shelf) thatformed during the Last Glacial Maximum andpossibly also during older Quaternary eustaticlowstands (after Pelejero et al. 1999). Map showsshallow shelf areas exposed during lowstands

    and major paleo-drainage systems that can beidentified by present-day seafloor bathymetry(Molengraaff 1921; Tjia 1980). Belida Field,located in the broad shallow southwestern partof the South China Sea, is nearly 1000 kilometersaway from the continental shelf edge and a fewhundred kilometers basinward of the modernshoreline. Inset map of the study area showsintersection of each 200th northsouth cross-lineand eastwest inline seismic traces (dots), loca-tion of the seismic cross section in Figure 3(dashed line AA9 ), and the outline of the areashown in Figures 4 and 5.

    226 B.J. WILLIS ET AL. J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    3/14

    GEOLOGIC SETTING

    The West Natuna Basin consists of a series of small, roughly eastwestoriented depocenters with intervening basement ridges that formedduring Paleogene rifting across the southern South China Sea (Sundashelf) region (Fig. 1; White and Wing 1978; Wongsosantiko andWirojudo 1984; Daines 1985; Ginger et al. 1993; Murray 2003). Manyhalf-graben depocenters of the West Natuna Basin experienced significantcontraction during late Oligocene to Miocene time, which formed theSunda Folds, classical examples of inverted basins (Ginger et al. 1993;Olson and Dorobek 2000). The Upper Miocene to Recent MudaFormation, about 600750 m thick, is a broadly recognized lithostrati-graphic unit in the West Natuna Basin that formed during a period of slow regional subsidence after inversion had ended around 10 Ma(Ginger et al. 1993; Phillips et al. 1997; Madon and Watts 1998; McClayand Bonora 1998; Murray 2003). Thus, strata of the Muda Formation arelargely undeformed.

    The Muda Formation is defined at its base by a thick interval of outer-shelf mudstone facies that generally define the topseal for underlyingsandstone and carbonate reservoirs in many basins across the Sunda shelf (Mattes 1979; Cossey et al. 1982; Sutoto 1991; Mayall et al. 1995; Michaeland Adrian 1996; Matthews et al. 1997; Maynard and Murray, 2003). TheMuda Formation was generally thought to consist entirely of marine

    mudstones before the 1990s, because well logs and drill cuttings wererarely collected until deeper hydrocarbon-bearing intervals were pene-trated. Recent studies have combined high-resolution seismic data andshallow sediment cores to document variations within the upper 50 100 m of the Muda Formation and its lateral stratigraphic equivalents.These studies demonstrated that shallow intervals of the MudaFormation consist of deposits that formed during the last sea-level cycle,including: (1) fluvial channels deposited during the Last GlacialMaximum lowstand, (2) transgressive deposits formed during the post-Pleistocene sea-level rise, and (3) shallow-shelf deposits formed during thepresent-day highstand (e.g., Aleva 1973; Batchelor 1979; Emmel andCurray 1982; Evans et al. 1995; Hanebuth and Statteger 2003; Wong et al.2003). Deeper intervals of the Muda Formation remain poorlydocumented.

    Belida field, in the western part of West Natuna Basin (Maynard et al.2003), is in the center of the Sunda shelf, nearly 1000 km up dip (west) of the modern shelf slope break. Currently in 56 m deep water, this area islikely to have been subaerially exposed during Pliocene to Holocenemaximum glacio-eustatic lowstands (Fig. 1). Undeformed post-inversionstrata in Belida field (Fig. 2), including the uppermost Arang Formationand the overlying Muda Formation, can be divided into three intervals.Seismically, each of these intervals is defined by strata with channel-formreflections that are separated by strata with more-continuous horizontal

    F IG. 2.Seismic cross section and chronostratigraphy of Belida Field, West Natuna Basin (after Palynova 2003). Hydrocarbon reservoirs are found within invertedhalf-graben structures (below 750 ms twtt). The Muda Formation, formed during approximately the last 10 Myr, was deposited after tectonic inversion had ended andthe region gradually subsided. The base of the Muda Formation is defined by an abrupt fining to shale, above an interval of channels in the uppermost Arang Formation.This study addresses an interval with increasing channels in the upper 500 ms twtt of this seismic record.

    MUDA FORMATION THREE-DIMENSIONAL FLUVIAL ARCHITECTURE 227J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    4/14

    reflections. Each interval is estimated to record several million years of deposition, on the basis of proprietary biostratigraphic data from a wellin an adjacent area of the West Natuna Basin (Darmadi 2005) and bycomparisons with Morley et al. (2003). The oldest of these intervalsinclude channel-form reflections and overlying parallel-horizontal reflec-tions in the upper Arang Formation that may have been mildly influencedby the last stages of basin inversion. The second interval is defined bychannel-form reflections within the uppermost Arang Formation andparallel-horizontal reflections in the basal part of the Muda Formation.The third interval includes channel-form reflections in the upper part of the Muda Formation and overlying parallel-horizontal reflections thatdirectly underlie the modern sea floor. Each of these intervals areinterpreted to be third-order stratigraphic sequences (i.e., have durationsof a few million years). Their boundaries, however, do not necessarilyrecord an abrupt coarsening across a single discrete allostratigraphicsurface (in contrast to the definition of Van Wagoner et al. 1987), butrather appear to be characterized by a gradual upward change in faciesacross a succession of smaller-scale internal unconformity-boundedintervals. This study focuses on smaller-scale vertical variations withinthe youngest of these third-order stratigraphic intervals.

    METHODS

    A 3D seismic-reflection survey acquired in 2001 by Conoco Indonesiaacross the 17 km by 40 km area of Belida field has in-line spacing of 25 mand cross-line spacing of 12.5 m. Because most wells in Belida fieldinclude information only from deeper stratigraphic levels wherepetroleum reservoirs are located (i.e., Eocene to upper Miocene strata),the thickness of Muda deposits was estimated by a comparison of verticalseismic variations to velocity logs that extend through the shallow part of four wells in Belida field. Given the relatively shallow depth of the MudaFormation, average seismic velocity is estimated to be 1650 meters persecond. The highest frequencies recorded in the seismic data are nearly100 Hz (dominant frequencies ranged from 50 to 60 Hz), potentiallyallowing stratal features less than 10 m thick to be resolved. Seismicmapping was accomplished using combined in-line and cross-line verticalseismic sections and horizontal time slices of successive 4 ms twtt(milliseconds two-way travel time) thick horizons.

    Seismic root-mean-squared amplitude variations were used to interpretdepositional features and geomorphologic patterns in time slices.Interpretation of stratigraphic trends defined by vertical changes of thefluvial architecture was a challenge because channel deposits of differentsize and incision depth that are cut down from a variety of horizons aresuperimposed within individual seismic slices. By scrolling throughsuccessive horizontal time slices and vertical seismic cross sectionssimultaneously, individual channel-form reflections were assigned to thestratigraphic horizon from which they incised. Although many channelbodies where so highly fragmented by incision of subsequent channelsthat it was difficult to define their dimensions, where bodies were morecomplete, their width, depth, and sinuosity were recorded relative to theirincision horizon. Planview maps of these more complete channels were

    compiled for successive stratigraphic intervals to show vertical changes inchannel size, direction, and fluvial style.

    Seismic Variations within the Upper Muda Formation

    A seismic section of the upper 500 ms twtt interval of Belida field(Fig. 3) shows the location and cross-sectional geometry of channels inthe upper part of the Muda Formation. This seismic cross section extendsfrom a high-amplitude, laterally continuous horizontal reflection in theMuda Formation at 430 ms twtt upward to the modern sea floor at 56 mstwtt, this represents a thickness of about 225 m. Channels imaged withinthis interval become more abundant and on average larger upsection.

    Larger channels appear to cut downward preferentially from particularhorizons, an observation that was used to subdivide this interval intoa succession of thinner sequences. Boundaries of these thinner sequenceswere defined by tracing particular reflections across the seismic volumethat marks the base of clusters of more deeply incised channels. Althougheach sequence boundary is defined by a reflection that is continuousacross the seismic volume, the basal surface of the oldest sequence(Sequence Boundary 1 or SB 1) is less clearly an allostratigraphic surface(versus a clustering of channel deposits cut down from slightly differenthorizons) than those that define the four younger sequences. Theyoungest two sequences (above SB 4 and SB 5) are complexly nestedwithin the same interval, and the boundary separating these sequences(SB 5) could be defined only locally because the associated interfluvesmerge within seismic resolution with those of SB4. Individual sequencesare 50 to 100 ms twtt (about 3060 m) thick and are interpreted torepresent deposition over about 0.5 Myr on the basis of simpleinterpolation of the estimated age range of the entire Muda Formation.Amplitude variations across representative horizontal seismic time slicesthrough this interval (Fig. 4) are discussed in detail below. Becausechannels incised from different sequences are commonly superimposedwithin horizontal time slices, maps of channel bodies incised froma specific stratigraphic interval are presented for a similar set of intervalsas shown in each of the presented seismic slices (Fig. 5).

    Below Sequence 1

    Reflections in the basal part of the seismic cross section (Fig. 3;between 500 and 400 ms twtt) are laterally continuous and subhorizontal.Lateral variations in amplitude observed in horizontal seismic time slicesthrough this interval are broad and gradual over 5 to 10 km. Above400 ms twtt, sparse small channels are defined by high-amplitudereflections, each less than 100 meters wide, 1020 ms twtt (, 612 m)deep, and about 2 km in channel-bend wavelength (Fig. 4A). Thesechannels are generally oriented southwestnortheast and are locallyhighly sinuous (i.e., ratio of channel-centerline to straight-line distance of 2.53.5). Bar accretion deposits, visible on the insides of some channel-bend segments, indicate migration dominated by expansion. The fewchannel bends with asymmetric accretion patterns (Figs. 4A, 5A) suggestpaleoflow to the northeast. Although interchannel amplitude variationsare more variable upsection, they remain gradual over a few kilometersand locally appear to be lobate in planview (Fig. 4A). Two largerchannels defined by low-amplitude reflections, on the order of a kilometerwide and nearly 45 ms (, 25 m) deep, cut down from horizons just belowSequence Boundary 1 (400430 ms twtt). These larger channels are nearlysymmetrical in cross section, have low sinuosity, and lack evidence of bank accretion deposits. Smaller, higher-sinuosity channels, like thoselower within this interval, are also observed.

    Sequence 1

    Sequence 1 (generally between 350 to 250 ms twtt; on average about60 m thick) is underlain by Sequence Boundary 1. Basal deposits are low-to moderate-sinuosity channel deposits up to 2 kilometers wide and50 ms twtt (, 30 m) deep (Figs. 3, 4B). The sequence boundary isdefined by laterally mapping a reflection cut by the most deeply incised of these channels through the entire seismic volume (Fig. 6A). Although thisboundary delineates an interval along which larger channels areconcentrated (Fig. 3), it cannot be claimed conclusively that it representsa higher-order allostratigraphic surface relative to those that extendlaterally away from similar-scale channels found slightly lower and higherwithin the stratigraphy. The channel used to define this sequenceboundary shallows and thins toward the northeast (Fig. 6A), a directionthat can be inferred to be downstream on the bases of accretion depositsin adjacent channels.

    228 B.J. WILLIS ET AL. J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    5/14

  • 8/14/2019 Fluvial Sequences Low Gradient System

    6/14

    F IG. 4.Selected horizontal time slices showing the distribution and geometry of seismic-amplitude variations in different horizons of the upper Muda Formation atBelida Field. A) Horizon below Sequence Boundary 1 shows small meandering channels (arrows point to a few specific examples) that cross an area characterized by

    broad, gradual amplitude variations. B) Larger channels with moderate to high sinuosity dominate horizons in the middle interval of Sequence 1 (arrows point to twoexamples). C) Channels with widely varying dimensions observed in the upper interval of Sequence 1. The largest channels have fairly straight, southnorth orientation(arrows point to two examples). D) Horizon near the top of the incision fill above Sequence Boundary 2 (area of deepest underlying incision is marked by the dashed line(i ). Areas adjacent to this incision show local dendritic gully formation (arrow ii ). E) Large channel with well imaged bar accretionary deposits accumulated in the area of deepest incision along Sequence Boundary 3 (arrow i ). The very base of Sequence Boundary 4 incision also intersects this time slice (arrowii ). Small channels that cutdownward from within the upper part of Sequence 2 (arrow iii ) vary in sinuosity and pattern. Larger channels that cut down from the upper part of Sequence 3 are allfairly straight (arrow iv). F) Time slice intersecting the upper part of the basal channel along Sequence Boundary 3 (arrow i ) shows gully formation into adjacentfloodplain areas (arrow ii ). Straighter large channels in the upper part of Sequence 3 also penetrate this horizon (arrow iii ). Inner-bank bar accretion in the basal channelof Sequence 4 is also observed (arrow iv). G) Incised channels cut down from Sequences 4 and 5 into nearly the same interval. The largest channels incised from bothsequences contain well imaged inner-bank bar accretion deposits (arrow i points to bar in a Sequence 4 channel, and arrow ii points to bar in a Sequence 5 channel).Narrower channels with adjacent dendritic gullies cut down from Sequence 4. Straighter channels without adjacent dendritic gullies cut down from higher intervals withinSequence 4. H) Channels with adjacent dendritic gullies (i ) decrease in abundance and average size and increase in sinuosity from northwest to southeast in the upper partof Sequences 4 and 5. A smaller sinuous channel in the upper part of Sequence 5 (ii ) has adjacent dendritic gullies that overlie deposits of Sequence 4.

    230 B.J. WILLIS ET AL. J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    7/14

    upper part of Sequence 1 are uniformly smaller (less than a kilometerwide and 10 ms twtt deep) and have low sinuosity (Figs. 3, 4C, 5C). Onaverage they are more uniformly south-to-north oriented than thoselower in the sequence. Interchannel amplitude variations in time slicesalso are more subdued than those below Sequence Boundary 1.

    Sequence 2

    Sequence 2 (generally between 250 to 170 ms twtt; on average about50 m thick) is defined at its base by a reflection with nearly 90 ms twtt(, 55 m) of relief (Fig. 6B). The dominant incision along this sequenceboundary is a nearly two kilometers wide, 60 ms twtt (, 40 m) deep,northsouth oriented, straight channel. Reflections in this channel areinitially concordant with its base, and then they successively become morehorizontal. Several smaller channels that also cut into this basal surface(Figs. 4D, 5D) are generally parallel to the larger channel, and togetherdefine a nearly 10-km-wide incision trend (Fig. 6B). In one location, smalldendritic gullies expand over a few kilometers onto the adjacent interfluveadjacent to this incised trend (Fig. 4D).

    Above the interval of incision, isolated larger channels (kilometerswide) have a fairly straight northsouth orientation. In the upper part of this sequence narrower channels have progressively more west-to-easttrends and have more variable planform geometry, ranging from highersinuosity to straighter segments with downstream branching or anasto-mosing geometries (Fig. 4E, F). Smaller channels are defined bycontrasting amplitude variations (i.e., some have higher amplitudes andothers lower amplitudes than surrounding deposits). The complexdistribution of smaller channel segments, many of which are near thelimit of seismic resolution, makes it difficult to define patterns of amplitude variation within strata between channel bodies.

    Sequence 3

    Sequence 3 (generally between 170 to 120 ms twtt, on average about30 m thick) is defined at its base by a sequence boundary with slightlyover 50 ms twtt (, 30 m) of relief (Fig. 5C). The stratigraphically lowestparts of this sequence boundary define a sinuous, westeast incised trend,in contrast to the southnorth trend of incisions that characterizeunderlying sequence boundaries. A large sinuous channel with well-imaged bar accretion deposits is found in the deepest part of the mainincision along Sequence Boundary 3 (Fig. 4E, F). The extent of theseaccretion surfaces and adjacent channel fill in cross section (Fig. 3), andplanview extent of bar accretion sets, indicate that this channel was nearly2 kilometers wide and 40 ms twtt (, 25 m) deep, and had a wavelengthof about 58 kilometers. Accretion sets extend more that four times thewidth of the adjacent channel fill, which has a sinuosity of about 3.Several similarly large channels, with more modest sinuosity, incise fromslightly higher stratigraphic levels into Sequence Boundary 3.

    Time slices that cut through less incised areas along SequenceBoundary 3 show well developed dendritic gullies expanding severalkilometers away from the most deeply incised areas (Figs. 4F, 5F).Channels in Sequence 3, incised downward from above the horizon cut by

    gullies, are less than a kilometer wide and 20 ms twtt (,

    12 m) deep, andhave low sinuosity (Fig. 4F, G). Amplitude variations along thesechannel bodies suggest bar wavelengths of about 5 km, similar to thosein the high-sinuosity channel at the base of Sequence 3 (Fig. 5F).

    Sequences 4 and 5

    Sequences 4 and 5 (generally between 120 to 80 ms twtt, on averageabout 25 m thick) are nested within nearly the same interval, and thustheir deposits are difficult to differentiate. The combined erosion surfaceat the base of these sequences has about 75 ms twtt (, 50 m) of relief.Both sequence boundaries rise locally to a distinctive horizon with

    extensive dendritic gullies, some extending away from adjacent channelsonto interfluves over distances greater than 10 km. The distinctionbetween these two sequences is that gullies associated with incision of SB5 locally erode into the tops of channels in Sequence 4. In contrast,channels of Sequence 5 have gullies along their margins, but these gulliesdo not extend across the main axes of these channel deposits.

    The basal deposit of Sequence 4 is a moderately sinuous channel withwell imaged inner-bend bar accretion deposits (Figs. 4G, 5H). Thischannel is 3 km wide and nearly 70 ms twtt (50 m) deep, and has a bendwavelength of 14 km. Asymmetry of accretion deposits in planviewsuggests flow from northwest to southeast (Figs. 4 and 5, F and G). Otherlarge channel deposits that are cut into this interval are narrower,straighter, a bit thinner (about 1 km wide and 30 ms twtt deep), and lackobvious adjacent bar accretion deposits. Amplitude variations along oneof these straighter channels suggest that channel bars are about 3 km inlength (Fig. 4G). Crosscutting relationships suggest that these straighterchannels are younger in age than the large channel that floors thesequence, even though they cut downward from nearly the same horizon.The basal channel of Sequence 5, incised from nearly the same interval asthose in Sequence 4, has low sinuosity and bars with well imagedaccretion surfaces that suggest dominantly downstream migration fromwest to east (Figs. 4G, 5G). There are distinct gullies along the margin of

    this channel and along a narrower adjacent channel that crosses above thepath of the largest channel in Sequence 4.Above the gullied surfaces of Sequences 4 and 5, channel segments with

    varying size and sinuosity are generally discontinuous across horizontaltime slices. There is a decrease in channel size and abundance from westto east across the study area. Strata are poorly imaged above this intervalbecause of the strength of the reflection from the modern ocean floor.

    Interpretation of Depositional Patterns

    The upper interval of the Muda Formation comprises an overallprogradational fluvialdeltaic succession formed on a broad, low-gradient platform under conditions of relatively slow subsidence.Horizontal reflections near the base of the study interval (Figs. 2, 3;400 to 700 ms twtt) image marine strata deposited during transgression

    across the interval of fluvial channels in the lower Muda Formation(Fig. 2, from 700 to 900 ms twtt). The particularly high-amplitudereflection at about 430 ms twtt may be a maximum flooding surface.Gradual amplitude variations across horizontal time slices reflect broadlithologic variations within these marine deposits. Minor crenulationsalong these generally horizontal reflections are aligned through manysuccessive horizons and probably record deformation around gas pipesthrough unconsolidated sediments. Small, high-sinuosity channels thatare imaged at about 400 ms twtt, and that are joined by larger, straighterchannels in horizons stratigraphically a bit higher, may reflect prograda-tion of a distributive (delta top) fluvial system. Diffuse amplitudevariations observed in time slices in areas between these channel bodiesare broadly lobate, and may record lithologic variations across adjacentmouth-bar or interdistributary-bay deposits.

    The wide variety of channel sizes in the overlying sequences suggeststhe presence of rivers that carried very different discharge. The largestchannels (kilometers wide and 2550 m thick) are inferred to be depositsof major trunk rivers, whereas smaller channels of variable size andplanview geometry are inferred to be tributaries or floodplain drainagechannels. Reflection geometries that define bar accretion deposits andchannel fills within the largest channels (e.g., Fig. 7) suggest that they hada channel-forming cross-sectional area of about 60,000 m2 . Althoughaccurate estimates of channel-forming discharge require information onvertical trends in grain size within these channel deposits (which is notavailable), broadly constrained estimates based on assuming a mean grainsize of fine sand suggest mean flow rates of about 0.6 m/s and thus

    MUDA FORMATION THREE-DIMENSIONAL FLUVIAL ARCHITECTURE 231J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    8/14

  • 8/14/2019 Fluvial Sequences Low Gradient System

    9/14

    a channel-forming discharge of about 35,000 to 40,000 m3 /s (see methodsin Willis 1993a). This suggests that the largest channel deposits imaged inthe study area were formed by rivers with discharge nearly as great asaverage floods of the modern Mekong River.

    Vertical clusters of large channels along particular stratigraphicintervals, used to define the bases of sequences, are interpreted to recordperiods of lower accumulation rate relative to avulsion rate of the trunkriver(following standardsequence stratigraphicmodels for fluvial deposits;Bridge and Leeder 1979; Shanley and McCabe 1993; Wright and Marriott1993). Intervals higher within sequences, with fewer and generally smallerchannel deposits, are interpreted to record periods of more rapidaggradation that led to increased preservation of tributary channels andfloodplain deposits. The standard interpretation of these types of verticalvariations is that channel incisions are related to sea-level falls, andsubsequent periods of floodplain aggradation are related to increasingaccommodation during sea-level rise. An alternative is that for fluvialsystems that decrease in slope onto a consistently low-gradient shelf, a fallin sea level leads to fluvial floodplain aggradation as the river systemadjusts to greater distances to the regressing coastline (Summerfield 1985;Posamentier et al. 1992; Holbrook et al. 2006). Clusters of channels alongthe base of sequences separating episodes of floodplain aggradation couldthen be interpreted to record more gradual erosion associated with bypass

    of sediment basinward during continued sea-level fall and lowstand, ratherthan knickpoint erosion landward. In either case, interpretations of a regional control on sequence development in this succession need to beviewed with some caution, because there likely were multiple river systemstraversing the study area and the region spanned by the seismic volume isprobably significantly smaller than the floodplain width across which themain trunk river avulsed. Thus, it is possible that observed verticalvariations in fluvial architecture may also reflect longer-term lateral shiftsin thepositionof thetrunkriveror temporal changes in itsfloodplain widthrelated to adjustments of adjacent rivers and varying aggradation rate of their associated floodplains.

    Tributaries cut into the floodplain adjacent to the larger trunk rivers, asdefined by distinctive dendritic gully systems observed in time slices, arethe best evidence for incision of the trunk river into the floodplain. Thesefeatures indicate the formation of valleys with adjacent interfluves(Posamentier 2001). Such gullies are not expected to form adjacent tonon-incised river channels, because their growth to scales visible inseismic slices is inhibited by overbank deposition during periodic riverfloods. In this study area gullies are observed to downcut only fromsequence boundaries.

    In the study interval, there are increases in the abundance and extent of gullyformation along successive sequence boundaries. No evidenceof gullyformation was observed cut into Sequence Boundary 1. Evidence of areallyrestricted gullies is observed very locally along the most deeply incisedchannel along Sequence Boundary 2. Evidence of nearly continuous gullyformation, extending a few kilometers onto adjacent interfluves, isobserved along the most deeply incised channels of Sequence 3. The bestimaged and most extensive gullies extend broadly over many kilometersacross the amalgamated interfluves of Sequence Boundaries 4 and 5. More

    extensive gully systems are interpreted to record longer periods of trunk-

    river incision during the formation of successive sequence boundaries, andperhaps also greater river incision depth.

    Progressive decrease in mean sequence thickness and increase in gullyformation along successive sequence boundaries suggests that the upperinterval of the Muda Formation comprises a forward-stepping sequenceset. That is, the formation of successive sequence boundaries is inferred tobe associated with longer-term bypass of sediment, longer-term loss of accommodation, and possible regression of shorelines farther basinward.The poorly imaged channels in the uppermost part of this sequence set(upper interval of Sequence 5; Fig. 4H) may indicate that channel-filllithologies became more similar to adjacent floodplain deposits duringinitial transgression. Deposits that cap Sequence 5 are thin (comprising atmost the upper 4060 ms twtt of sediment), and probably formed duringcontinued marine transgression of this prograding sequence set. Thespecific horizon that records marine transgression is poorly imaged due tothe strength of the sea-floor reflection.

    There are systematic changes in channel size and pattern upward withineach sequence (Fig. 7). The size of channels generally decreases upwardwithin each sequence. Although the deepest channels cut into sequenceboundaries clearly record incision greater than associated channel flowdepths (values greater than about 40 m), the thickness of bar accretiondeposits along inner banks of large channel bends suggest that 2040 m

    deep channels were common. Maximum relief along sequence boundariesis thus only slightly greater than the thickness of accretion sets associatedwith the largest individual channels, indicating that total depths of valleyincision were not much greater than the depths of the associated riverchannels. Decrease in mean channel size upward within sequences reflectsdifferent dimensions of incised and overlying unincised major channelsand also preservation of smaller tributary and floodplain channels.

    Sequence boundaries with better defined incision (i.e., SequenceBoundaries 35) are overlain by higher sinuosity channels with moreobvious bar accretion deposits, and channels that incise from horizonsstratigraphically higher in these sequences are generally straighter. Thisobservation is contrary to the general assumption that steeper streamgradients along the axis of an incised valley lead to straighter braidedchannel patterns, and that channels become more sinuous as slopesdecline during transgressive valley filling (Wright and Marriott 1993).Near the center of the broad exposed Sunda shelf stream gradients mayhave been uniformly very low during Muda deposition. Thus, moresinuous major channels may record slopes steepened by incision and thestraighter, lower-sinuosity channels that show no evidence of lateralmigration may have flowed across lower gradients (Fig. 8). Upwardchange in the sinuosity of larger channels from the base of Sequence 1 tothose that incise downward from the middle of this sequence may indicatethat the deepest channel in this non-incised sequence did not flow downthe steepest gradients. Given that the sinuous channels at the bases of themost deeply incised sequences are also the largest (as indicated by thescale of inner-bank accretion deposits), it may alternatively be thatchannels with greater discharge had greater stream power relative tosediment load. In this case, their sinuous planform (relative to the largechannels higher in the same sequence) may reflect an increased ability of

    these channels to erode their banks during high discharge, rather than

    r

    F IG. 5.Map-view traces of channels compiled by combining observations from several successive time slices through the upper Muda Formation at Belida Field.Although mapped intervals of channels correspond in a general way to seismic slices shown in correspondingly labeled parts of Figure 4, the seismic slices are complicatedby channels that incise downward from intervals higher in the stratigraphy. A) Channels that incise downward from the interval below Sequence Boundary 1, between 412to 368 ms twtt. B) Channels that incise downward from the basal and middle intervals of Sequence 1, between 354 and 310 ms twtt. C) Channels that incise downwardfrom the upper interval of Sequence 1, between 306 and 278 ms twtt. D) Channels that incise downward from the basal interval of Sequence 2, between 278 and 250 mstwtt. E) Channels that incise downward from the upper interval of Sequence 2 and the most deeply incised deposits of Sequence 3, between 214 and 190 ms twtt. F)Channels that incise downward from the upper interval of Sequence 3 and most deeply incised deposits of Sequence 4, between 180 and 160 ms twtt.G) Channels thatincise downward from the uppermost interval of sequence 3, in sequence 4, and the lowermost interval of Sequence 5.H) Channels that incise downward from theuppermost intervals of Sequence 4 , between 134114 ms twtt.

    MUDA FORMATION THREE-DIMENSIONAL FLUVIAL ARCHITECTURE 233J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    10/14

    reflecting temporal changes in downstream gradients. The sinuosity of smaller channels probably reflects more local stream gradients andsediment loads, which could have been influenced by local floodplaindepositional features (e.g., levees and crevasse splays adjacent to thelarger channels). Thus, smaller channels are less likely to reflect regionalstratigraphic trends. For example, some small sinuous rivers alongSequence Boundary 5 appear to change downstream into straighterincised gullies (Fig. 4G).

    Origin of Sequences

    The origin of sequences in the upper interval of the Muda Formation isdifficult to constrain from the limited area of this study. Posamentier(2001) suggested fluvial incision can occur due to: (1) sea-level fall acrossa region with steepening downstream gradients, for example across anexposed shoreface or the continental shelf edge; (2) erosion of channelsacross areas of tectonic uplift; and (3) climatic variations in riverdischarge that leave underfit river deposits along the courses of larger-discharge channels. Although none of these mechanisms can be ruled outdefinitively based on the data available from Belida field, they can beassessed based on observed variations within sequences and the broaderdepositional setting.

    The Muda Formation at Belida field is located in middle parts of regional drainage systems that extended across much of the Sunda shelf during Pliocene to Holocene sea-level lowstands (Fig. 1; Murray 2003;Murray and Dorobek 2004). Regional seismic profiles extending eastwardacross all of the West Natuna Basin and parts of the adjacent EastNatuna and Nam Con Son basins show that the Muda Formation (andits lateral stratigraphic equivalents) vary little in thickness across thenearly 1000 km distance to the present-day Sunda shelf edge (Dorobek,unpublished). Sediment aggradation across the Sunda shelf thus appearsto have been broadly laterally consistent during Pliocene to Recent time.This observation, and the repetitive nature of fluvial variations thatdefine sequences in the upper Muda Formation, argues against localtectonic controls on fluvial aggradation and erosion rates.

    Posamentier (2001) suggested that rivers crossing a low-gradient,

    lowstand-exposed shelf are likely to remain non-incised over large areaswhen sea-level fall is not extensive enough to expose the shelf edge. To theextent that this is true, a sea-level control on upper Muda Formationsequences would require that large regions of the Sunda shelf, perhapsdistances of 1000 km or more, responded essentially synchronously tosea-level changes. Variations in the incision of interfluves observed onsuccessive sequence boundaries would then also suggest that gullyformation occurs very slowly relative to rates that trunk rivers inciseshelf-edge knickpoints landward across the shelf. Similarly, fluvialaggradation rates and channel behavior also would be required torespond very rapidly across broad regions to downstream-gradientdeclines as the shelf was episodically flooded. There is little evidence forsuch extensive and rapid response of fluvial systems to high-frequencysea-level variations, in as much as stream gradient changes required toadjust river profiles at distances of many hundreds of kilometers from theshelf edge would be negligible. Burns et al. (1997) suggest that effects of sea-level changes on river gradients should be only a few backwaterlengths (river depth divided by downstream slope), a distance thatshould not exceed a few hundred kilometers, even considering theprobable low gradients of the exposed Sunda shelf.

    r

    F IG. 6.Topographic maps of sequence boundaries defined by tracing specificreflections through the seismic volume of the upper Muda Formation atBelida Field.

    234 B.J. WILLIS ET AL. J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    11/14

    An alternative mechanism for sea-level control on sequence formationin the upper Muda Formation is that fluvial incision across a successionof falling-stage shorelines on the shelf formed during episodic short-term

    regressions and transgressions. This is supported by the modest depths of incision along sequence boundaries relative to sizes of the largest internalchannel deposits (less than two to one). Given expected low shelf gradients, this model would still require significant regressions andtransgressions to allow the accumulation of shoreline deposits that arethick enough to produce adequate fluvial incision, followed by very rapidlandward expansion of incised valleys cut into falling-stage shorelineslocated basinward. Support for this idea would require identification of coeval Muda Formation shoreline deposits basinward of Belida field.Given evidence that this sequence set is dominantly regressive, it might beexpected from this model that younger sequences would be less deeplyincised as successive delta-front knickpoints shifted farther basinwardonto the shallow shelf. This supposition is not supported by observationsfrom this example.

    Although climate controls on river discharge are difficult to constrainfrom our available data, they would be expected given the icehouseconditions of the Pliocene to Recent and low-latitude setting of the studyarea. Climatically modulated changes in river discharge could affect riverchannel size, sediment transport capacity, and patterns of erosion anddeposition over large areas of these drainage basins. Large changes inriver discharge during the Holocene caused by changes in monsoonal

    precipitation patterns have been reported to occur in rivers that drain intothe Himalayan foredeep, as recorded by incision of floodplains extendingacross alluvial fans of the Ganges plains to low-lying areas on the top of

    the GangesBrahmaputra delta (Goodbred 2003; Gibling et al. 2005).Whereas incision and changes in river pattern are expected to attenuatelandward away from shelf-edge knickpoints exposed during sea-levelfalls, effects of variations in river channel discharge caused by climaticchanges are more likely to influence the entire drainage basin. Bogart andvan Balen (2000) speculated that river incision may reflect lags betweenincreases in river water discharge and associated increases in sedimentsupply. The observation of more sinuous channels with better developedbank accretion deposits along the most deeply incised areas of sequenceboundaries may support this supposition, in as much as it suggests thatincised rivers would have more stream power to erode their banks.Similarly, decrease in the size of the largest channels upward withinsequences may indicate declining river discharge relative to sedimentloads, rather than a backfilling of incised drainages due to declininggradients caused by sea-level rise.

    DISCUSSION

    Although sequence stratigraphic concepts have been widely applied tofluvial successions, most well constrained examples are from depositsformed near the coast, where fluvial deposition can be related fairly

    F IG. 7.Close-up view of largest channel incised into Sequence Boundary 4, in horizontal time slice and cross section.

    F IG. 8.Dimensions and sinuosity of chan-nel-form reflections shown relative to the strati-graphic horizon from which they originated. Thesea-level curve of Haq et al. (1987) is shown onthe right.

    MUDA FORMATION THREE-DIMENSIONAL FLUVIAL ARCHITECTURE 235J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    12/14

    directly to stacking patterns in shallow marine deposits and shorelineregressiontransgression trends. For example, where fluvial channels areincised into regressive shoreface deposits, fluvial incisions are partiallyfilled with transgressive estuarine facies and the succession is capped bya marine flooding surface (summaries in Shanley and McCabe 1993;Zaitlin et al. 1994; Blum and Tornqvist 2000). For these examples

    knickpoint erosion into a delta front edge exposed during sea level fall isassumed to have a major influence on valley development.

    For successions deposited farther inland, sequence divisions aregenerally defined based on vertical variations in the proportion of channel to overbank deposits, and less commonly by recognizing laterallycontinuous erosion surfaces, abrupt changes in sandstone petrography, orintervals with mature paleosols (e.g., Aitken and Flint 1995; Legarretaand Uliana 1988; Miall and Arush 2001; McCarthy and Plint 1998).These interpretations tend to be more conjectural, however, because theinfluence of sea level fluctuations are expected to attenuate landward andthus, other allostratigraphic mechanisms that can control variations inalluvial architecture become increasingly probable explanations forobserved variations. Increasing channel sandstone proportion is notalways associated with declining aggradation rate as predicted by sea-

    level-driven alluvial sequence stratigraphic models (e.g., Read and Dean1985; Willis 1993b; and discussion of hypothetical cases in Heller andPaola 1996), and several authors have suggested that in many cases fluvialchannels aggrade or simply extend basinward following sea level fallsrather than incising (Miall 1991; Posamentier 2001).

    Given the prevalence of erosion surfaces of varying depth within fluvialdeposits, evidence for valley incision from outcrop exposures is generallybased on recognizing extensive erosion surfaces with relief significantlygreater than the thickness of individual channel bar accretion sets andadjacent channel fills. In this paper we follow Posamentier (2001) insuggesting that distinctive dendritic gullies adjacent to channel marginsobserved in horizontal time slices of 3D seismic volumes are adequateevidence to demonstrate that a river channel was incised into itsfloodplain, despite basal erosion surfaces with nearly the same relief asthe thickness of adjacent point-bar deposits. The occurrence of similarlowstand bypass channels of the Burdekin River across the low-gradientshelf of northeastern Australia led Fielding et al. (2003, 2005) to proposea distinction between entrenched channels and incised valleydeposits. Although entrenched channels have not been recognized inoutcrop, this may reflect a lack of observation rather than the absence of these features. For example, in the Siwaliks of Pakistan the second authorof this paper observed multi-storied sandstone bodies that terminated atcutbank margins against overbank successions that lacked levee deposits,but he failed to walk out the surface correlative with the terminating basalerosion surface of the channel sandstone body into the adjacent overbankdeposits to look carefully for evidence of gully formation.

    Posamentier (2001) distinguished non-incised and incised lowstandchannels using seismic cross sections similar to those presented here, andsuggested that, although both types of channel bodies were depositedduring an overall fall in sea level, incised channels likely form only whenthe sea level falls below the antecedent shelf edge. The channels hediscussed, in the southern Java Sea, were interpreted to record fluvialchannels and valleys cut directly into shelf deposits (i.e., regressions wereso rapid that prograding shallow marine deposits were not preserved).The deposits imaged in this study are somewhat different, in as much asseveral hundred meters of net aggradation occurred during deposition of these sequences, basal deposits of this succession appear to be deltaic, andthe overall succession is interpreted to be progradational. Channels in thissystem episodically aggraded and then incised. The occurrence of depositsformed by very large rivers suggests that regional drainages were free toavulse over wide areas of the Sunda shelf between episodic incisions,because it is unlikely that deposits of this large river are restricted bychance to the small area covered by this study. In this setting, distinctionbetween non-incised, entrenched, and incised valley channel bodies ismore tenuous, and depends critically on the resolution in the seismicimage of features like bar accretion surfaces and interfluve gullies. Theremay be a complete gradation between these types.

    Where lowstand incision is not pronounced, as in this case, it becomes

    difficult to distinguish channel incisions caused by climatically controlleddischarge variations, very subtle floodplain deformation, and landwardknickpoint erosion following sea-level falls. Channel incisions initiated bythese different mechanisms are unlikely to be regionally synchronous.Given that gully incisions appear to extend only short lateral distancesaway from the entrenched channels (order of kilometers to a few tens of kilometers), it is not clear that the sequences defined here are of regionalextent. Episodic entrenchment and river avulsion during interveningperiods of floodplain aggradation may leave a series of laterallydiscontinuous falling-stage erosion surfaces rather than a single regionalallostratigraphic sequence boundary. Similarly, incisions into the shelf edge caused by knickpoint erosion following a sea-level fall may beregionally discontinuous, where river courses in areas upstream remainnon-incised and continue to avulse over time (Leeder and Stewart 1996;Tornqvist et al. 2003).

    Alluvial architecture models used to calculate subsurface channelsandstone connectivity generally assume a stationary distribution of channel bodies, on the basis of the underlying idea that floodplainaggradation is dominated by levee growth adjacent to major channels,which expand at uniform rates to trigger a statistically constantdistribution of river channel avulsions through time (Leeder 1978; Allen1978; Bridge and Leeder 1979). In this subsurface example, the largestchannels are incised and aggradational intervals appear to be dominatedby fluvial systems with smaller channels. Although the scale of the studyarea and seismic resolution of floodplain depositional elements may notbe adequate to define alluvial ridges adjacent to the major channels, anabundance of smaller channel systems within intervals that separatemajor channel bodies agrees with models that predict floodplainaggradation dominantly occurs during avulsion events (Smith et al.1989) rather than the growth of extensive alluvial ridges. Miall (2006) alsonoted that the wide variety of channel sizes and fluvial styles imaged in3D seismic data of the Pleistocene Pilong Formation (offshore Malaysia)suggests pronounced changes in channel-body dimensions and perhapsdeposits of multiple river systems. This degree of complexity is notcurrently addressed in alluvial architecture models. High-frequencyepisodes of channel entrenchment and aggradation, as observed in thisstudy, are also likely to increase horizontal relative to vertical channeldeposit connectively, compared to predictions for systems with equalchannel sandstone proportion formed under more uniformly aggradingconditions and a stationary distribution of channel avulsions. Alluvialarchitecture models that can incorporate sequence stratigraphic trends

    F IG. 9. Relationship between slope and channel sinuosity defined by flumeexperiments of Schumm and Khan (1972).

    236 B.J. WILLIS ET AL. J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    13/14

    also require better understanding of extrabasinal influences on patterns of fluvial deposition. In summary, the next generation of alluvial sequencestratigraphic models will likely require detailed 3D seismic images of a broader range of fluvial depositional systems as well as well-constraineddata on contemporaneous eustatic fluctuations and regional differentialsubsidence patterns.

    CONCLUSIONS

    The upper Muda Formation at Belida Field in the West Natuna Basinis a 225-meter-thick, progradational sequence set that coarsens upwardfrom marine shelf and deltaic deposits to fluvial deposits. These depositsare capped by modern marine deposits of the Sunda shelf. Five fluvialsequences, each of which is tens of meters thick, are defined by theclustering of channels along specific horizons that are separated byintervals with fewer channel deposits. Sequence boundaries are generallymarked by erosion greater than the thickness of bar accretion depositsadjacent to individual channels; gully formation on interfluves alsocharacterizes sequence boundaries. Deeper erosion and greater extent of gully formation on successive sequence boundaries suggests greater long-term loss in accommodation and greater bypass of sediment throughtime. Major channels in basal parts of the better developed sequences(Sequences 3 through 5) are larger and more sinuous, and show greaterpreservation of bar accretion deposits than channels higher in thesesequences. These vertical changes may indicate that sequence-boundaryincision was associated with climatic changes in river discharge, ratherthan changes in river gradient associated with sea-level fluctuations. Thearchitecture of fluvial deltaic sequences deposited on a low-gradientcontinental shelf differs from better documented highstand examplesbecause sea-level fall can lead to fluvial aggradation and fluvial channelsmay not become deeply incised. Current sequence stratigraphic modelsfor fluvial deposits fail to address the variety of possible controls onalluvial architecture and the range of possible responses of fluvial systemsto changing allocyclic controls.

    ACKNOWLEDGMENTS

    This research was supported by a scholarship to the first author from theEducational Consortium of the Coordinating Board of Oil and Gas of Indonesia. Chuck Caughey of ConocoPhillips was especially supportive. Wealso thank the Belida Group of ConocoPhillips Indonesia in Jakarta,including Rahadian Adhyaksawan, Sugiharto Danudjaja, Paulus Hamid,Fadia Hamid, and Redo Waworuntu for help in acquisition of these data.Roald Brotherton and Vaughn Ball are thanked for permission to publish.

    REFERENCES

    AITKEN , J.F., AND FLINT , S.S., 1996, Variable expressions of interfluvial sequenceboundaries in the Breathitt Group (Pennsylvanian), eastern Kentucky, USA, inHowell, J.A., and Aitken, J.F., eds., High-resolution Sequence Stratigraphy:Innovations and Applications: Geological Society of London, Special Publication104, p. 193206.

    ALEVA , G.J.J., 1973, Aspects of historical and physical geology of the Sunda Shelf essential to the exploration of submarine tin placers: Geologie en Mijnbouw, v. 52, p.7991.

    ALLEN , J.R.L., 1978, Studies in fluviatile sedimentation: an exploratory quantitativemodel for the architecture of avulsion-controlled suites: Sedimentary Geology, v. 21,p. 129147.

    ASLAN , A., AND AUTIN , W.J., 1999, Evolution of the Holocene Mississippi Riverfloodplain, Ferriday, Louisiana; insights on the origin of fine-grained floodplains:Journal of Sedimentary Research, v. 69, p. 800815.

    BATCHELOR , B.C., 1979, Discontinuously rising Late Cainozoic eustatic sea-levels, withspecial reference to Sundaland, Southeast Asia: Geologie en Mijnbouw, v. 58, 1020 p.

    BERENDSEN , H.J.A., AND STOUTHAMER , E., 2001, Palaeogeographic development of theRhineMeuse delta: Assen, The Netherlands, Van Gorcum, 268 p.

    BLUM , M.D., 1993, Genesis and architecture of incised valley fill sequences: A LateQuaternary example from the Colorado River, Gulf Coastal Plain of Texas, inWeimer, P., and Posamentier, H.W., eds., Siliciclastic Sequence Stratigraphy, RecentDevelopments and Applications: American Association of Petroleum Geologists,Memoir 58, p. 259283.

    BLUM , M.D., AND LANCASTER , R., 2005, Incision of the lower Mississippi valley (southcentral USA) during deglacial sea-level rise (abstract): Geological Society of America,Abstracts with Programs, v. 37, 233 p.

    BLUM , M.D., AND TORNQVIST , T.E., 2000, Fluvial responses to climate and sea levelchange: a review and look forward: Sedimentology, v. 47, p. 248.

    BLUM , M.D., M ORTON , R.A., AND DURBIN , J.M., 1995, Deweyville terraces anddeposits of the Texas Gulf Coastal Plain: Gulf Coast Association of GeologicalSocieties, Transactions, v. 45, p. 5360.

    BLUM , M.D., M ISNER , T.J., C OLLINS , E.S., SCOTT , D.B., MORTON , R.A., AND ASLAN, A.,1996, Middle Holocene sea-level rise and highstand at + 2 m, central Texas Coast:Journal of Sedimentary Research, v. 71, p. 581588.

    BOGART , P.W., AND VAN BALEN , R.T., 2000, Numerical modeling of the response of alluvial rivers to Quaternary climate change: Global and Planetary Change, v. 27, p.147163.

    BRIDGE , J.S., 1985, Description and interpretation of fluvial deposits: a criticalperspective: Sedimentology, v. 42, p. 382389.

    BRIDGE , J.S., AND LEEDER , M.R., 1979, A simulation model of alluvial stratigraphy:Sedimentology, v. 26, p. 617644.

    BURNETT , M., 1996, 3-D seismic expression of a shallow fluvial system in west centralTexas, in Weimer, P., and Davis, T.L., eds., Applications of 3-D Seismic Data toExploration and Production: American Association of Petroleum Geologists, Studiesin Geology 42, p. 4556.

    BURNS , B.A., HELLER , P.L., M ARZO , M., AND PAOLA, C., 1997, Fluvial response ina sequence stratigraphic framework: example from the Montserrat Fan Delta, Spain:Journal of Sedimentary Research, v. 67, p. 311321.

    CARTER , D.C., 2003, 3-D seismic geomorphology: insights into fluvial reservoirdeposition and performance, Widuri Field, Java Sea: American Association of Petroleum Geologists, Bulletin, v. 87, p. 909934.

    COSSEY, S.P.J., VALENTA , W.T., P AYOT , F., AND HO, T.T., 1982, Seismic hydrocarbonindicators in the South China Sea: geological and geophysical aspects: IndonesianPetroleum Association, 11th Annual Convention Proceedings, v. 1, p. 335355.

    DAINES , S.R., 1985, Structural history of the West Natuna Basin and the tectonicevolution of the Sunda region: Indonesian Petroleum Association, 14th AnnualConvention Proceedings, p. 3961.

    DARMADI , Y., 2005. Three-dimensional fluvialdeltaic sequence stratigraphy, Pliocene Recent Muda Formation, Belida Field, West Natuna Basin, Indonesia [Mastersthesis]: Texas A&M University, 74 p.

    EMMEL , F.J., AND CURRAY , J.R., 1982, A submerged late Pleistocene delta and otherfeatures related to sea level changes in the Malacca Strait: Marine Geology, v. 47, p.192216.

    EVANS , C.D.R., B RETT , C.P., J AMES , J.W.C., AND HOLMES , R., 1995, Shallow seismicreflection profiles from the waters of East and Southeast Asia: an interpretationmanual and atlas: British Geological Survey, Technical Report WC/94/60, p. 6667.

    F IELDING , C.R., T RUEMAN , J.D., D ICKENS , G.R., AND PAGE , M., 2003, Anatomy of theburied Burdekin River channel across the Great Barrier Reef shelf: How does a majorriver operate on a tropical mixed siliciclastic/carbonate margin during sea-levellowstand?: Sedimentary Geology, v. 157, p. 291301.

    F IELDING , C.R., TRUEMAN , J.D., D ICKENS , G.R., AND PAGE , M., 2005, Geomorphologyand internal architecture of the ancestral Burdekin River across the Great BarrierReef shelf, north-east Australia, in Blum, M.D., Marriott, S.B., and Leclair, S.F., eds.,Fluvial Sedimentology VII: International Association of Sedimentologists, SpecialPublication 35, p. 321347.

    GIBLING , M.R., T ANDON , S.K., SINHA , R., AND JAIN , M., 2005, Discontinuity-boundedalluvial sequences of the southern Gangetic Plains, India: aggradation anddegradation in response to monsoonal strength: Journal of Sedimentary Research,v. 75, p. 369385.

    GINGER , D.C., ARDJAKUSUMAH , W.O., H EDLEY , R.J., AND POTHECARY , J., 1993, Inversionhistory of the West Natuna Basin: examples from the cumi-cumi PSC: IndonesianPetroleum Association, 22nd Annual Convention Proceedings, p. 635658.

    GOODBRED , S.L., 2003, Response of the Ganges dispersal system to climate change:a source-to-sink view since the last interstade: Sedimentary Geology, v. 162, p. 83104.

    HANEBUTH , T.J., AND STATTEGER , K., 2003, The stratigraphic evolution of the SundaShelf during the past fifty thousand years, in Sidi, F.H., Nummedal, D., Imbert, P.,and Darman, H., eds., Tropical Deltas of Southeast Asia; Sedimentology,Stratigraphy, and Petroleum Geology: SEPM, Special Publication 76, p. 189200.

    HAQ, B.U., HARDENBOL , J., AND VAIL, P.R., 1987, Chronology of fluctuating sea levelssince the Triassic: Science, v. 235, p. 11561167.

    HELLER , P.L., AND PAOLA , C., 1996, Downstream changes in alluvial architecture: an

    exploration of controls on channel stacking patterns: Journal of SedimentaryResearch, v. 66, p. 297306.HOLBROOK , J., SCOTT , R.W., AND OBOH-IKUENOBE , F.E., 2006, Base-level buffers and

    buttresses: a model for upstream versus downstream control on fluvial geometry andarchitecture within sequences: Journal of Sedimentary Research, v. 76, p. 162174.

    KHADKIKAR , A.S., AND RAJSHEKHAR , C., 2005, Holocene valley incision during sea leveltransgression under a monsoonal climate: Sedimentary Geology, v. 179, p. 295303.

    LEEDER , M.R., 1978, A quantitative stratigraphic model for alluvium, with specialreference to channel deposit density and interconnectedness, in Miall, A.D., ed.,Fluvial Sedimentology, Canadian Society of Petroleum Geologists Memoir 5, p.587596.

    LEEDER , M.R., AND STEWART , M.D., 1996, Fluvial incision and sequence stratigraphy:alluvial responses to relative sea-level and their detection in the geologic record, inHesselbo, S.P., and Parkinson, D.N., eds., Sequence Stratigraphy in British Geology:Geological Society of London, Special Publication 103, p. 2539.

    MUDA FORMATION THREE-DIMENSIONAL FLUVIAL ARCHITECTURE 237J S R

  • 8/14/2019 Fluvial Sequences Low Gradient System

    14/14

    LEGARRETA , L., AND ULIANA , M.A., 1998, Anatomy of hinterland depositionalsequences: Upper Cretaceous fluvial strata, Neuquen basin, West-Central Argentina,in Shanley, K.W., and McCabe, P.J., eds., Relative Role of Eustasy, Climate, andTectonism in Continental Rocks: SEPM, Special Publication 59, p. 8392.

    MADON , M.B., AND WATTS , A.B., 1998, Gravity anomalies, subsidence history and thetectonic evolution of the Malay and Penyu basins (offshore Peninsula Malaysia):Basin Research, v. 10, p. 375392.

    MATTES , E.M., 1979, Udang Field: a new Indonesian development: IndonesianPetroleum Association, 8th Annual Convention Proceedings, p. 177184.

    MATTHEWS , S.J., FRASER , A.J., LOWE, S., TODD , S.P., AND PEEL, F.J., 1997, Structure,stratigraphy and petroleum geology of the SE Nam Con Son Basin, offshore Vietnam,in Fraser, A.J., Matthews, S.J., and Murphy, R.W., eds., Petroleum Geology of Southeast Asia: Geological Society of London, Special Publication 126, p. 89106.

    MAYALL , M.J., BENT , A., AND ROBERTS , D.M., 1995, Miocene carbonate buildups in theNam Con Son Basin, offshore socialist Republic of Vietnam: American Association of Petroleum Geologists, Bulletin, v. 88, p. 12341235.

    MAYNARD , K., PRABOWO , W., G UNAWAN , J., WAYS , C., AND BROTHERTON , R., 2003,Maximizing the value of a mature asset, The Belida Field, West Natunacana detailed subsurface re-evaluation really add value late in field life?: IndonesianPetroleum Association, 29th Annual Convention and Exhibition Proceedings, v. 2, p.291305.

    MAYNARD , K., AND MURRAY , I., 2003, One million years from the upper ArangFormation, West Natuna Basin, implications for reservoir distribution and faciesvariation in fluvial deltaic deposits: Indonesian Petroleum Association, 29th AnnualConvention and Exhibition Proceedings, v. 1, p. 270276.

    MCCARTHY , P.J., AND PLINT , A.G., 1998, Recognition of interfluve sequence boundaries:Integrating paleopedology and sequence stratigraphy: Geology, v. 26, p. 387390.

    MCCLAY , K., AND BONORA , M., 1998. Tectonic evolution of Sunda Shelf and the WestNatun Basin, Indonesia, Part A: regional overview: Final Report presented to ConocoIndonesia Inc., 57 p.

    MIALL , A.D., 1991, Stratigraphic sequences and their chronostratigraphic correlation:Journal of Sedimentary Petrology, v. 61, p. 497505.

    MIALL , A.D., 2002, Architecture and sequence stratigraphy of Pleistocene fluvialsystems in the Malay Basin, based on seismic time slices analysis: AmericanAssociation of Petroleum Geologists, Bulletin, v. 86, p. 12011216.

    MIALL , A.D., 2006, Reconstructing the architecture and sequence stratigraphy of thepreserved fluvial record as a tool for reservoir development: A reality check: AmericanAssociation of Petroleum Geologists, Bulletin, v. 90, p. 9891002.

    MIALL , A.D., AND ARUSH , M., 2001, The Castlegate Sandstone of the Book Cliffs, Utah:sequence stratigraphy, paleogeography, and tectonic controls: Journal of SedimentaryResearch, v. 71, p. 537548.

    MICHAEL , E., AND ADRIAN , H., 1996, The petroleum system of West Block B PSC, SouthNatuna Sea, Indonesia: Indonesian Petroleum Association, 25th Annual Conventionand Exhibition Proceedings, v. 1, p. 465479.

    MOLENGRAAFF , G.A.F., 1921, Modern deep-sea research in the East Indian archipelago;Geographical Journal, v. 57, p. 95121.

    MORLEY , R.J., M ORLEY , H.P., AND RESTREPO -PACE, P., 2003, Unravelling the tectonicallycontrolled stratigraphy of the West Natuna Basin by means of palaeo-derived midTertiary climate changes: Indonesian Petroleum Association, 29th Annual Conven-tion Proceedings, v. 1, p. 561584.

    MURRAY , M.R., 2003. Regional tectonics, differential subsidence, and sedimentdispersal patterns: implications for sediment flux to the southern South China Seaand regional filling of sedimentary basins during Pliocene to Recent time [Mastersthesis]: Texas A&M University, 120 p.

    MURRAY , M.R., AND DOROBEK , S.L., 2004, Sediment supply, tectonic subsidence, andbasin-filling patterns across the southwestern South China Sea during Pliocene toRecent time, in Clift, P., Wang, P., Kuhnt, W., and Hayes, D., eds., ContinentOceanInteractions with East Asian Marginal Seas: American Geophysical Union, Geo-physical Monograph 149, p. 235254.

    OLSON , C.C., AND DOROBEK , S.L., 2000, Styles and significance of structural inversionacross the Nam Con Son Basin, offshore SE Vietnam (abstract): AmericanAssociation of Petroleum Geologists, Annual Meeting, 109 p.

    OLSEN , T., STEEL , R., H OGSETH , K., SKAR , T., AND ROE, S., 1995, Sequential architecturein a fluvial succession: sequence stratigraphy in upper Cretaceous Mesaverde group,Price Canyon, Utah: Journal of Sedimentary Research, v. 65, p. 265280.

    PALYNOVA , LTD ., 2003, Sequence characterization and fingerprinting: Jakarta, Indonesia,ConocoPhillips, Internal Company Report, 120 p.

    PELEJERO , C., GRIMALT , J.O., H EILIG , S., KIENAST , M., AND WANG , L., 1999, High

    resolution UK

    37 temperature reconstructions in the South China Sea over the past220 kyr: Paleoceanography, v. 14, p. 224231.PHILLIPS , S., LITTLE , L., MICHAEL , E., AND VAN ODELL , R., 1997, Sequence stratigraphy

    of Tertiary petroleum systems in the West Natuna Basin, Indonesia: IndonesianPetroleum Association, Petroleum Systems of SE Asia and Australasia, ConferenceProceedings, p. 381390.

    PLINT , A.G., M CCARTHY , P.J., AND FACCINI , U.F., 2001, Nonmarine sequencestratigraphy: updip expression of sequence boundaries and systems tract in a highresolution framework, Cenomanian Dunvegan Formation, Alberta Foreland Basin,Canada: American Association of Petroleum Geologists, Bulletin, v. 86, p. 19672001.

    POSAMENTIER , H.W., 2001, Lowstand alluvial bypass systems: incised vs. unincised:American Association of Petroleum Geologists, Bulletin, v. 85, p. 17711793.

    POSAMENTIER , H.W., A LLEN , G.P., J AMES , D.P., AND TESSON, M., 1992, Forcedregressions in a sequence stratigraphic framework: concepts, examples, andexploration significance: American Association of Petroleum Geologists, Bulletin,v. 76, p. 16871709.

    READ , W.A., AND DEAN , J.M., 1982, Quantitative relationships between numbers of fluvial cycles, bulk lithological composition and net subsidence in a ScottishNamurian basin: Sedimentology, v. 29, p. 181200.

    SCHUMM , S.A., AND KHAN , H.R., 1972, Experimental study of channel patterns:Geological Society of American, Bulletin, v. 83, p. 17551770.

    SHANLEY , K.W., AND MCCABE, P.J., 1991, Predicting facies architecture through

    sequence stratigraphyan example from the Kaiparowits Plateau, Utah: Geology,v. 19, p. 742745.SHANLEY , K.W., AND MCCABE, P.J., 1993, Alluvial architecture in a sequence

    stratigraphic framework; a case history from the Upper Cretaceous of southernUtah, USA, in Flint, S.S., and Bryant, A.D., eds., The Geological Modelling of Hydrocarbon Reservoirs and Outcrop Analogues: International Association of Sedimentologists, Special Publication 15, p. 2155.

    SMITH , N.D., C ROSS, T.A., D UFFICY , J.P., AND CLOUGH , S.R., 1989, Anatomy of anavulsion: Sedimentology, v. 36, p. 123.

    STARKEL , L., 1991, Long-distance correlation of fluvial events in the temperate zone, inStarkel, L., Gregory, K.J., and Thornes, J.B., eds., Temperate Palaeohydrology:Chichester, John Wiley, p. 473495.

    SUMMERFIELD , M.A., 1985, Plate tectonics and landscape development on the Africancontinent, in Morisawa, M., and Hack, J., eds., Tectonic Geomorphology: Boston,Allen and Unwin, p. 2751.

    SUTOTO , A., 1991, Reservoir geology of the Belida Field South Natuna Sea, Block B.The petroleum system of West Block B PSC, South Natuna Sea, Indonesia:Indonesian Petroleum Association, 25th Annual Convention and ExhibitionProceedings, v. 1, p. 453478.

    TEBBENS, L.A., VELDKAMP , A., WESTERHOFF , W., AND KROONENBERG , S.B., 1999, Fluvialincision and channel downcutting as a response to late glacial and Early Holoceneclimate change: the River Meuse lower reach: Quaternary Science, v. 14, p. 5975.

    TJIA, H.D., 1980, The Sunda shelf, Southeast Asia: Annals of Geomorphology, v. 4, p.405427.

    TORNQVIST , T.E., 1993, Holocene alternating of meandering and anastomosing fluvialsystem in RhineMeuse delta (central Netherlands) controlled by sea level rise andsubsoil erodibility: Journal of Sedimentary Petrology, v. 63, p. 683693.

    TORNQVIST , T.E., WALLINGA , J., MURRAY , A.S., DE WOLF , H., CLEVERINGA , P., AND DEGANS, W., 2000, Response of the RhineMeuse system (west-central Netherlands) tothe last Quaternary glacio-eustatic cycles: a first assessment: Global and PlanetaryChange, v. 27, p. 89111.

    TORNQVIST , T.E., WALLINGA , J., AND BUSSCHERS , F.S., 2003, Timing of the last sequenceboundary in a fluvial setting near the highstand shorelineInsights from opticaldating: Geology, v. 31, p. 279282.

    VAN WAGONER , J.C., M ITCHUM , R.M., P OSAMENTIER , H.W., AND VAIL, P.R., 1987,Seismic stratigraphy interpretation using sequence stratigraphy, Part 2: key definitionsof sequence stratigraphy, in Bally, A.W., ed., Atlas of Seismic Stratigraphy: AmericanAssociation of Petroleum Geologists, Studies in Geology 27, p. 1114.

    WILLIS , B.J., 1993a, Interpretation of bedding geometry within ancient river-channeldeposits, in Marzo, M., and Puigdefabregas, C., eds., Alluvial Sedimentation:International Association of Sedimentologists, Special Publication 17, p. 101114.

    WILLIS , B.J., 1993b, Evolution of Miocene fluvial systems in the Himalayan foredeepthrough a two kilometer thick succession in northern Pakistan: Sedimentary Geology,v. 88, p. 77122.

    WHITE , J.R., AND WING , R.S., 1978, Structural development of the South China Sea withparticular reference to Indonesia: Indonesian Petroleum Association, 7th AnnualConvention Proceedings, p. 159177.

    WONG , H.K., L UDMANN , T., HAFT , C., AND PAULSEN , A.M., 2003, Quaternarysedimentation in the Molengraaff paleo-delta, northern Sunda Shelf (Southern SouthChina Sea), in Sidi, F.H., Nummedal, D., Imbert, P., and Darman, H., eds., TropicalDeltas of Southeast Asia; Sedimentology, Stratigraphy, and Petroleum Geology:SEPM, Special Publication 76, p. 201218.

    WONGSOSANTIKO , A., AND WIROJUDO , G.K., 1984, Tertiary tectonic evolution and relatedhydrocarbon potential in the Natuna Area: Indonesian Petroleum Association, 13thAnnual Convention, Proceedings, v. 1, p. 161183.

    WRIGHT , V.P., AND MARRIOTT , S.B., 1993, The sequence stratigraphy of fluvial

    depositional systems: the role of floodplain sediment storage: Sedimentary Geology,v. 86, p. 203210.ZAITLIN , B.A., DALRYMPLE , R.W., AND BOYD , R., 1994, The stratigraphic organization of

    incised-valley systems associated with relative sea-level change, in Dalrymple, R.W.,Boyd, R., and Zaitlin, B.A., eds., Incised-Valley Systems: Origin and SedimentarySequences: SEPM, Special Publication 51, p. 4560.

    ZENG , H., AND HENTZ , T.F., 2004, High-frequency sequence sedimentology: Applied toMiocene, Vermilion Block 50, Tiger Shoal area, offshore Louisiana: AmericanAssociation of Petroleum Geologists, Bulletin, v. 88, p. 153174.

    Received 11 December 2005; accepted 30 September 2006.

    238 B.J. WILLIS ET AL. J S R


Recommended