FORMATION OF DEEP INCISIONS INTO TIDE-DOMINATED RIVER DELTAS: IMPLICATIONS FOR THE STRATIGRAPHY OF THE SEGO SANDSTONE, BOOK CLIFFS, UTAH, U.S.A.

Transcription

1 FORMATION OF DEEP INCISIONS INTO TIDE-DOMINATED RIVER DELTAS: IMPLICATIONS FOR THE STRATIGRAPHY OF THE SEGO SANDSTONE, BOOK CLIFFS, UTAH, U.S.A. BRIAN J. WILLIS 1 AND SHARON L. GABEL 2 1 Department of Geology and Geophysics, Texas A&M University, College Station, Texas , U.S.A. willis@geo.tamu.edu 2 Department of Earth Sciences, State University of New York, Oswego, New York 13126, U.S.A. ABSTRACT: The Upper Cretaceous Sego Member of the Mancos Shale in east-central Utah is composed of tidally deposited sandstones interbedded with intervals containing marine shales and thin wave-deposited sandstones. The tidal sandstones have been interpreted to comprise multiple amalgamated estuarine valley fills above a major sequence boundary that is incised into distal marine deposits of the underlying Buck Tongue Shale. According to this interpretation, the deposits constitute the base of a thick transgressive succession within the Sevier foreland clastic wedge. Alternatively, lower Sego sandstones have recently been interpreted by the authors to be tide-dominated river delta deposits in an overall regressive interval of the foreland succession. A cross section showing facies variations and stratigraphic surfaces within the lower Sego Member along 100 kilometers of the eastern Book Cliffs is used in this paper to refine depositional and sequence stratigraphic interpretations of these tidal deposits, particularly the origin of deep incisions within the Sego sandstones. A strike-oriented section of the lower Sego Member reveals tidal sandstone lenses, kilometers wide and tens of meters thick, separated laterally by areas tens of kilometers wide that have no tidal sandstones. Facies variations within the tidal sandstones reflect depositional processes on tide-dominated deltas that prograded obliquely into the shallow intracratonic Western Interior Seaway of North America. Two major episodes of delta progradation and subsequent transgressions are interpreted to have formed tidal sandstone layers within the lower Sego Sandstone. Thinner tidal sandstone layers, of more local extent, are interpreted to reflect changes in tidal current strength but not necessarily shifts in shoreline position or major changes in water depth. Multiple origins are suggested for incised erosion surfaces within the lower Sego sandstones. Firstly, the tide-dominated deltaic sandstones are floored by laterally discontinuous erosion surfaces. These basal surfaces, marking abrupt vertical changes in depositional process and coarsening of grain size, are interpreted to record tidal scouring of the sea floor during shoreline progradation. Secondly, sharp-based upward-coarsening deltaic sandstones are cut by numerous deep incisions with tens of meters of relief, and locally they are entirely removed by erosion. Some of this deep erosion within the lower Sego Sandstone may reflect the upstream avulsion of incised distributaries during forced regression of deltas into the basin, whereas others may have formed by tidal erosion of abandoned distributaries during more gradual progradation and subsequent retrogradation of deltas. Highly variable facies trends within incision fills are related to pronounced changes in sediment supply to different distributaries on the delta, to differences between incisions cut during delta regression versus those cut during the start of transgression, and to varying distances from delta axes. When the entire coast was transgressed, the influence of tides waned and wave-generated currents ravined delta tops. Depositional interpretations of the Sego Member indicate that the Buck Tongue to Neslen interval lies in a broadly regressive part of the foreland basin fill. INTRODUCTION The Sego Member of the Mancos Shale contains tide-deposited sandstones within a thick clastic wedge that prograded away from the Cretaceous Sevier orogenic belt into the Western Interior Seaway of North America (Fig. 1). Stratigraphic studies of deposits exposed along the Book Cliffs of central Utah show that the Sego sandstones lie within a finergrained interval of the foreland basin succession that separates coarsergrained fluvial wedges of the underlying Castlegate Member from overlying Blue Castle Tongue Member (Fig. 2; Young 1955; Fouch et al. 1983; Van Wagoner et al. 1990; Van Wagoner 1991, 1995; Olsen et al. 1995; Kirschbaum and Hettinger 1998; McLaurin and Steel 2000; Willis 2000). Exceptional exposures of the Sego Sandstone in the eastern Book Cliffs (Green River, Utah, to Grand Junction, Colorado) occur along a 100 km southwest northeast oriented outcrop belt that exposes a dominantly depositional strike-oriented cross section. There, the westward-thinning Buck Tongue Member of the Mancos marine shale onlaps the eastward-thinning Castlegate fluvial wedge. The Sego Member erosionally overlies the Buck Tongue Member, and fine-grained fluvial and coastal-plain deposits of the Nelson Formation in turn overlie these tidal sandstones. The Sego Member has been divided into lower and upper sandstones, separated by marine shales of the Anchor Mine Tongue (Young 1955). The Neslen Formation contains fluvial channels and overbank successions, with numerous coals and tide-influenced horizons, that aggraded behind vertically stacked, wavedominated shorelines in western Colorado (Kirschbaum and Hettinger 1998). The Blue Castle Tongue is a layer of coarser-grained fluvial deposits above the basal Neslen Formation that is associated with regression of shorelines farther east into Colorado. Although it is generally agreed that thicker sandstones within the Sego Member reflect deposition dominated by tidal currents (Van Wagoner 1991, 1995; McLaurin and Steel 2000; Willis 2000; Willis and Gabel 2001), there is little agreement about the depositional environments in which these tidedominated deposits formed or about the stratigraphic implications of these tidal deposits for the larger-scale foreland basin stratigraphy. Most thick shoreline successions within the Sevier foreland basin fill are fluvial- or wave-dominated, and it has commonly been suggested that tide-influenced deposits formed only during transgression of incised valleys or embayments on low-lying delta-top coastal plains. Major erosion surfaces that define the bases of the underlying Castlegate and overlying Blue Castle fluvial wedges are generally agreed to be sequence boundaries that record pronounced basinward (eastward) shifts in deposition away from the rising mountain belt (to the west). In contrast, the architecture and sequence stratigraphic interpretations of the finer-grained deposits separating these fluvial wedges, including the Sego Sandstone, remain controversial. Sequence stratigraphic interpretations of deposits between the Castlegate and Blue Castle fluvial wedges depend critically on how one interprets abrupt facies transitions and high-relief erosion surfaces within the Sego Sandstone Member. Van Wagoner (1991) interpreted the tidal Sego sandstones to be a complex stack of incised valleys with estuarine fills formed during multiple high-frequency changes in sea level. Yoshida et al. (1996) suggested that the Sego Sandstone is underlain by a major sequence boundary that marks an abrupt fall in relative sea level, and that the Sego sandstones record the start of long-term episodic transgression. The interpretation of Yoshida et al. (1996) was supported by evidence that the Sego Sandstone erosionally cuts out the Buck Tongue as this shale thins landward onto the underlying Castlegate Sandstone. McLaurin and Steel (2000) questioned evidence for regional erosion at the base of the Sego Sandstone, JOURNAL OF SEDIMENTARY RESEARCH, VOL. 73, NO. 2, MARCH, 2003, P Copyright 2003, SEPM (Society for Sedimentary Geology) /03/ /$03.00

2 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 247 and suggested instead that the Buck Tongue and thinner shales within the Sego sandstone could be correlated many tens of kilometers landward into tide-influenced layers within the fluvial-dominated proximal parts of the foreland fill (see discussion and reply in Yoshida et al. 2001, and McLaurin and Steel 2001, respectively). Resolving this controversy hinges on interpreting depositional environments, controls on stratal architecture (including erosion surfaces), and regional changes in basin accommodation during deposition of the Sego tidal sandstones. Facies variations within tidal environments can be complicated by the shifting influence of fluvial sediment supply and the extent of sediment reworking by tidal currents. The strength of tidal currents along coastlines is controlled by a complex set of variables related to the geometry of bathymetric shoaling, the position of amphidromic nodes within the basin, and tidal resonance where basin width matches a harmonic of tidal wavelength (Ippen 1966; Johnson and Belderson 1969; Pugh 1987; Reynaud et al. 1999b). Studies of Recent and Holocene tidal systems identified marked changes in tidal range along shorelines and through time, and localized areas deeply eroded by tidal currents at a variety of water depths (Belderson et al. 1986; Dalrymple et al. 1990; Reynaud et al. 1999a; Reynaud et al. 1999b). Spatial and temporal changes in tidal intensity within basins make it difficult to interpret stratigraphic changes in depositional processes, stacking patterns of tidal sandstones, and the extent of shoreline erosion (Johnson and Belderson 1969). Because controls are complex, local changes in tidal current strength and the balance among river-, wave-, and tidal-influence on deposition may not reflect changes in shoreline proximity or depositional water depths. Modern river deltas prograding into coastal areas with strong tidal currents show a highly variable distribution of grain size along the shoreline, with elongate sand bodies extending into deeper waters away from areas of active sediment supply and deeply incised tidal channels dissecting delta fronts in areas temporarily abandoned when sediment feeding distributaries avulsed to other locations (Hori et al. 2002; Allison 1998; Harris et al. 1993; Harris et al. 1996; Kuehl et al. 1989; Kuehl et al. 1997). Thus, in contrast to that commonly inferred from deposits of riverdominated deltas and wave-dominated shorelines, local changes in sediment caliber, sedimentary structures, and the basinward extent of different sandstone layers in deposits of tidal systems do not necessarily reflect regional regression and transgression of shorelines due to changes in basin accommodation. This has important implications, inasmuch as interpreting changes in water depth and shoreline position is fundamental to defining sequence stratigraphic divisions. In contrast to previous depositional interpretations, Willis and Gabel (2001) suggested that the lower Sego Sandstone was deposited during episodic progradation and transgression of tide-dominated deltas. These interpretations were based on an examination of two 10-km-long outcrops, spaced about 50 km apart along strike. This study reports the results of a larger-scale tracing of stratigraphic surfaces and facies in the lower Sego Sandstone between and adjacent to the two areas documented in Willis and Gabel (2001). The goal of this paper is to define the broader architecture of these tide-dominated deposits, particularly the larger-scale distribution of tidal sandstones and the location of high-relief erosion surfaces, in order to refine previous sequence stratigraphic interpretations of the Castlegate Sego Neslen interval. After describing and interpreting the larger-scale stratal patterns within the lower Sego Sandstone exposed along the dominantly strike-oriented eastern Book Cliffs outcrop belt, we combine this information with recently published dip-oriented cross sections of the Sego Sandstone interval (Willis 2000; McLaurin and Steel 2000) to examine contrasting depositional interpretations of this foreland-basin fill. LOWER SEGO LITHOFACIES Deposits of the lower Sego interval can be divided into three lithofacies associations: (1) gray shales with thin wave-deposited sandstones; (2) tidal, erosionally based, inclined-bed sets; and (3) channel-form bedsets with FIG. 1. Paleogeographic map showing the location of the study area along the shoreline of the early Campanian Western Interior Seaway of North America (after Roberts and Kirschbaum 1995). Note that the seaway embayment at our study area was hypothesized by Roberts and Kirschbaum (1995) to explain the tide-influenced deposits observed in the Sego Member and basal Neslen Formation. varying degrees of tidal influence. Sedimentologic variations within these lithofacies have been previously described and interpreted in detail by Van Wagoner (1991), Willis (2000), and Willis and Gabel (2001), and therefore these variations are only summarized here. Cross sections in Figure 3 show the distribution of these lithofacies within the lower Sego interval mapped at two different scales. Wave-Deposited Bedsets Wave-deposited bedsets comprise 3 to 6 m thick, upward-coarsening, sheet-like bedsets that extend laterally for several km to over 10 kilometers (Fig. 4, facies 1 and 2). Bedsets grade upward from richly bioturbated marine shales to thin beds of very-fine-grained hummocky cross-stratified sandstone. Each bedset records the progradation of sand and mud over many storm events into wave-dominated marine areas. Bedsets may be distal shoreface successions or more isolated wave-reworked shoals (Howard and Reineck 1981; Elliott 1986; Shanley and McCabe 1995). The ratio of sandstone to shale within these deposits varies both vertically and laterally within the lower Sego interval and within the underlying Buck Tongue. In cross sections (Fig. 3) this lithofacies association is divided to distinguish deposits composed completely or dominantly of shale from those richer in sandstone (Fig. 4F; note that this subdivision changes with the scale mapped in Fig. 3). Tidal Inclined-Bed Sets Tidal inclined-bed sets (6 to 12 m thick, 2 to 6 km wide, and probably tens of kilometers long) are composed of cross-stratified fine- to mediumgrained sandstones with varying degrees of shallow marine bioturbation (most notably, abundant Ophiomorpha). They have basal erosion surfaces and internal beds that dip at 2 to 15 degrees basinward parallel to paleocurrents defined by the smaller-scale cross stratification (to the southeast). These bedsets generally coarsen upward from heterolithic cross-stratified sandstones with abundant internal shale drapes and evidence of flow reversals (Fig. 3, facies 3), to more homogeneous cross-stratified sandstones that reflect dominantly ebb-directed paleoflows (Fig. 3, facies 4). Bedset

3 248 B.J. WILLIS AND S.L. GABEL FIG. 2. A) Map of Sego Member outcrops along the Book Cliffs in central Utah and Colorado. Locations of measured sections along the depositional-strike oriented southeastern Book Cliffs outcrop belt used to construct Fig. 3A are shown by squares. Along the northwestern, more depositional dip-oriented, Book Cliffs outcrop belt locations sections from Willis (2000) are show by triangles, and those from McLaurin and Steel (2000) are shown by circles. Rectangular boxes show areas studied in detail by Willis and Gabel (2001) near Thompson and Westwater Canyons, and rose diagrams summarize paleocurrents of tidal sandstones measured in these two study areas (based on 215 and 310 measurements, respectively). B) Regional cross section of deposits exposed in the Book Cliffs from Price, Utah, to Grand Junction, Colorado (from Willis and Gabel 2001). Diagram was constructed by combining simplified versions of Young s (1955) diagram of the Blackhawk Sego Member interval with Kirschbaum and Hettinger s (1998) diagram of the Neslen Blue Castle Tongue interval in the southern Book Cliffs. The correlation of units above the Buck Tongue from the southern Book Cliffs northward (landward) toward Price remains controversial (cf. Willis 2000, and McLaurin and Steel 2000; Fig. 9). FIG. 3. A) Stratigraphic cross section of the lower Sego Member interval along the southern Book Cliffs outcrop belt. The cross section was constructed by tracing bedding surfaces, and noting facies variations, between sections shown in Fig. 2A. Sections from Willis and Gabel (2001) and Van Wagoner (1991) were also used extensively during this mapping process. The map position of measured sections were projected normal to a line extending 70 east of north to define their position on the cross section (note that this procedure produced slightly different cross-sectional stratal geometries than that of Van Wagoner (1991), who constructed his cross section along a line zigzagging between sections, as well as that of Willis and Gabel (2001), who used projection lines of slightly different orientation). Detailed sedimentologic cross sections showing stratal variations near B) Westwater Canyon and C) Thompson Canyon were modified slightly from Willis and Gabel (2001) based on additional field work. Note that facies distributions are defined at a slightly coarser scale in part A relative to B and C; the former defines more broadly classified divisions of sandier and shalier beds within the wave-dominated facies, as well as a more generalized division of more homogeneous and heterolithic facies within tidal sandstone layers. Numbers (1 3) along the right sides of cross sections B and C show the three tidal sandstone layers that Willis and Gabel (2001) correlated between their two study areas and interpreted to reflect regional sequences. The broader cross section presented in Part A shows that these three sandstone layers do not correlate and are not continuous regionally. Erosion surfaces at the top of the two thicker tidal sandstones in each area correlate, whereas thinner tidal sandstones appear to reflect more spatially restricted progradation of tidal sands into the basin.

4 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 249

5 250 B.J. WILLIS AND S.L. GABEL tops can be intensely bioturbated (Fig. 3, facies 5). Sandstones are dominated by cross-strata where bedsets are thickest, and they generally become more bioturbated as bedsets gradually thin toward their lateral margin. These bedsets are broadly interpreted to be deposits of prograding tidal coastlines (Willis and Gabel 2001). Bedding relationships suggest deposition on large bars with steep (about 5 15 ) lee faces that accreted basinward, parallel to the dominantly ebb-directed paleoflows recorded by the cross sets of smaller-scale superimposed dunes. These downstream-accreted bedsets differ from deposits of tidal ridges found on modern marine shelves. Such tidal-ridge deposits typically have few thick mud drapes, more uniformly sorted sands, and maximum bedding dips of only a few degrees oriented highly oblique to paleoflow (Stride et al. 1982; Berné et al 1988; Huthnance 1982a, 1982b). Sego tidal bedsets are also different from those produced by the migration of bars within tidal channels, passing through the mouths of estuaries or across broad intertidal and subtidal flats. Tidal-channel deposits typically fine upward and contain low-angle lateralaccretion bedding (Klein 1977; Harris et al. 1992; Dalrymple 1992; Dalrymple and Rhodes 1995). More specifically, Sego tidal inclined-bed sets are interpreted to be deposits of tidal-channel mouth bars formed where sediments pass down the delta front from tide-reworked intertidal and subtidal flats into deeper-water areas (Willis and Gabel 2001; see similar deposits documented by Nio 1976, Mutti et al. 1985, and Nio and Yang 1991). Channel-Form Bedsets Channel-form bedsets are 3 to 6 m thick and a few hundred meters wide, and have upward-concave basal erosion surfaces. They typically contain lateral-accretion bedding and upward-fining internal facies. Coarser-grained examples contain abundant fossilized wood chips in basal lags, are dominated by decimeters-thick cross-strata sets uniformly dipping basinward (southeast), and locally can be found complexly amalgamated above a more regional erosion surface. Finer-grained examples have ebb-dominated paleocurrents in their basal parts, and they typically have more heterolithic cross strata with abundant internal reactivation surfaces and superimposed ripple cross lamination indicating flood-directed paleocurrents. These finergrained bedsets can be locally amalgamated, but they typically are found isolated within tidal sandstone layers. Channel-form bedsets are interpreted to be deposits of fluvial or tidal channels. Lateral accretion beds record the growth of channel bars and suggest that channels were 3 6 m deep and up to 100 m wide. Because laterally accreted beds typically constitute less than half the bedset width, these deposits mainly reflect the filling of low-sinuosity channels. Coarsergrained examples may record deposition in fluvial or fluvial-dominated distributary channels. Finer-grained examples reflect deposition in strongly tide-influenced fluvial channels or channels of solely tidal origin crossing subtidal flats. In either case, these channel-form bedsets probably record deposition in shallower areas than do the tidal inclined-bed sets (Willis and Gabel 2001). REGIONAL DISTRIBUTION OF LITHOFACIES ALONG STRIKE Regional exposures of the lower Sego interval in the eastern Book Cliffs are essentially perpendicular to paleocurrents indicated by both sole marks in the wave deposits and cross strata in the tidal deposits (Fig. 2A). There is a vertical interlayering of wave-deposited and tidal sandstone facies and pronounced lateral changes in the abundance of tidal sandstones along depositional strike (Figs. 3A, 4G). The wave-deposited lithofacies, consisting entirely of very-fine-grained sandstones and shales, contrast with the thicker-bedded, fine- to medium-grained sandstones that dominate the tidal deposits. The tidal sandstones are concentrated in two areas along the regional cross section (Fig. 3A); an area to the east that includes the outcrops near Westwater Canyon (Fig. 3B), and an area to the west that includes outcrops near Thompson Canyon (Fig. 3C). This facies architecture indicates that deposition of tidal sands was both episodic and discretely located within the basin. The authors previously recognized three tidal sandstone layers in both Thompson Canyon (Fig. 3C) and Westwater Canyon (Fig. 3B) areas and interpreted each to record a regional progradation of shorelines into the basin (Willis and Gabel 2001). Shale-rich, wave-deposited strata vertically separating these tidal deposits were interpreted to record intervening periods of shoreline transgression followed by deposition during early stages of renewed regression. The regional cross section presented here (Fig. 3A) shows that none of the tidal sandstone layers extend across the 50 km separating the Thompson Canyon and Westwater Canyon study areas. There are, however, layers rich in hummocky cross-stratified sandstone that connect tidal sandstone layers between these areas along strike. Tidal sands may extend into varying water depths along the coastline, depending on local variations in the strength of tidal currents passing onto the coast. In contrast, where waves dominate deposition maximum current speeds and distances of basinward sand transport over many storms can be inferred to decrease relatively uniformly with increasing water depth away from the coastline. Thus for wave-dominated successions a progressive upward increase in hummocky cross-stratified sandstone relative to shale can be generally inferred to record shoaling. Where there are no tidal sandstones within the lower Sego interval (Fig. 3A, e.g., between km 40 and 60), there are two tens-of-meters thick, upward-coarsening successions: (1) from the middle of the Buck Tongue upward to an erosion surface in the middle of the lower Sego interval; and (2) from the dark shales overlying this erosion surface to an erosion surface that correlates with the top of the second thick tidal sandstone layer in the Westwater Canyon area. Thicker intervals rich in hummocky cross-stratified sandstone at the top of the first of these successions occur where there are overlying tidal sandstone layers. A third upward-coarsening succession starts at the base of the Anchor Mine Tongue and ends at the erosional base of the Upper Sego Sandstone. There are also several thinner layers rich in hummocky cross-stratified sandstones that occur out of sequence within these larger-scale upward-coarsening successions. The first thick layer of tidal sandstone in both the Thompson Canyon and Westwater Canyon areas terminates upward at the erosion surface that caps the first upward-coarsening hummocky cross-stratified succession. The second tidal sandstone layer at the east end of the Thompson Canyon outcrop (Fig. 3C) gradually thins to the east as it passes into a thin bench of hummocky cross-stratified sandstone that extends across to deposits near the base of the second tidal sandstone layer in the Westwater Canyon outcrop. This layer arguably continues farther as tidal deposits in the basal part of the second tidal sandstone layer in the Westwater Canyon outcrop, although these deposits are locally cut out by overlying tidal sandstones (e.g., note division of the second tidal sandstone in the Westwater Canyon area defined by a thin bed of dark shale with wave-rippled linsen sandstones; Fig. 3B, between km 11 and 12). A third tidal sandstone layer in Thompson Canyon is connected to the top of the second tidal sandstone layer in the Westwater Canyon outcrop by the hummocky cross-stratified sandstones in the upper part of the second upward-coarsening wave-dominated succession in the lower Sego interval. An isolated large-scale channel-shaped body of tidal sandstone with abundant oyster-shell lags is incised into this second thick hummocky crossstratified layer near Cottonwood Canyon in an otherwise fine-grained interval about halfway between the Thompson Canyon and Westwater Canyon outcrops. The third tidal sandstone in the Westwater Canyon area passes westward into a thin, discontinuous shell lag encased in shale, which could be traced only a few kilometers west of Westwater Canyon. It is possible that this tidal sandstone layer passes into a cryptic surface that extends farther west, or that it descends to join the top part of the hummocky cross-stratified bench adjacent to the top of the second tidal sandstone layer in Westwater Canyon. Neither of these relationships could be

6 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 251 FIG. 4. Details of facies recognized within the lower Sego Sandstone. A) Facies 1: Darkgray to medium-gray shale with variable bioturbation is interbedded with wave-rippled, lenticular to wavy bedded, very fine-grained sandstones. B) Facies 2: Decimeters-thick bed of hummocky cross-stratified sandstone records deposition under wave-generated currents. A low-angle dipping set of cross strata at the top of the bed and records a component of offshoredirected currents. C) Facies 3: Heterolithic cross strata overlying the basal erosion surface of a tidal inclined-bed set records dune migration under the influence of tidal currents; D) Facies 4: More homogeneous cross-stratified sandstone in the upper part of tidal inclined-bed set records more uniform offshore directed currents. E) Facies 5: Highly bioturbated sandstones (Ophiomorpha dominated) near the top of a tidal inclined-bed set. F) Coarsening-upward succession of wave-deposited bedsets (each a few meters thick) in the upper part of the Buck Tongue are truncated at the base of the first tidal sandstone layer of the lower Sego. G) Interlayering of finer-grained wave-dominated deposits and thicker-bedded tidal sandstones. demonstrated in the field because of breaks in the strike-oriented outcrop belt associated with side canyons cut into the Book Cliffs. The two thickest tidal sandstone layers partially erode into the top of adjacent upward-coarsening hummocky cross-stratified successions and partially aggrade above them. Thicker accumulations of hummocky crossstratified sandstone (i.e., beginning at lower stratigraphic levels) in close proximity to, and beneath, these thicker tidal sandstone layers indicate that areas where tidal sandstones occur consistently received a greater sand supply than intervening areas along the coastline. Variation in the amount of sandstone along strike is therefore related to changes in sediment supply along the coastline rather than to changes in initial basin bathymetry or subsidence rate. Strongly ebb-dominated paleocurrents and coarse grain sizes in tidal sandstones (relative to grain sizes within hummocky crossstratified beds) suggest that most of this sediment prograded into the basin,

7 252 B.J. WILLIS AND S.L. GABEL and was not built up by flood currents transporting marine sediments landward. Given these depositional patterns, areas reworked by tides were probably associated at the regional scale with large-scale shoreline protrusions rather than embayments. The two sandier areas within the lower Sego Sandstone documented in the regional cross section (Fig. 3A) are interpreted to reflect deposition along the axis of different tide-dominated deltas (Willis and Gabel 2001), fed by major rivers flowing from the northwest. These rivers were apparently spaced about 60 km apart along the shoreline, and they presumably flowed obliquely into the foreland basin as they entered the shallow seaway. On the basis of the one fully documented example (Fig. 3A), sandy parts of deltas were about 40 km wide, and were separated by mudstone-dominated intervals about 20 km wide. While many students of ancient tidal sandstones focused on enhanced tidal currents within shoreline embayments (e.g., Dalrymple et al. 1992; Reinson 1992; Mellere and Steel 1996; Bhattacharya and Willis 2001), changes in the distance that tidal currents transport sand basinward of coastlines and tidal-current amplification are also expected in areas where the seaway rapidly shoaled and when changes in shoreline position caused tidal resonance within the basin. Van Wagoner et al. (1990) suggested that lateral changes in the amount of sandstone within the Sego interval reflect tectonic changes in subsidence rate related to faults that cut obliquely across the foreland basin. Subtle tectonic deformation within the foreland basin is quite likely to have controlled the traverse of rivers to the seaway and sandstone depocenters along the shoreline (Martinsen and Krystinik 1998; Bhattacharya and Willis 2001; Martinsen 2001). The lower Sego interval does not, however, change dramatically in thickness along strike, suggesting that lateral variations in subsidence rate were not the dominant control on tidal sandstone occurrence. The cross section shown in Figure 3A, illustrates that two layers can be correlated regionally across the lower Sego interval despite lateral variations of internal facies. Thinner tidal sandstone layers and associated sandier intervals within the wave-dominated deposits are not as distinct regionally, and they may reflect more localized variations in offshore sand transport along the coastline. Sequence stratigraphic divisions of the Sego Sandstone depend on identifying regressive and transgressive episodes. Because the strength of tidal currents within a restricted seaway and distances that sands are transported offshore can vary greatly along the coastline and through time, interpretations require delineation of regional changes in facies architecture and identification of erosion surfaces that record longer-term sediment bypass into the basin. Understanding the origins of the deep incisions within the lower Sego deposits is considered critical for interpreting the stratigraphy of these thicker tidal sandstone layers. NATURE AND ORIGIN OF EROSION IN THE SEGO SANDSTONE Erosion Surfaces within the Lower Sego There are many erosion surfaces within the lower Sego Sandstone: (1) surfaces at the base of individual cross sets; (2) surfaces at the base of depositional beds; (3) basal surfaces of individual tidal inclined-bed sets and channel-form bedsets; (4) basal surfaces of tidal sandstone layers; (5) surfaces at the top of tidal sandstone layers that also extend across the top of adjacent hummocky cross-stratified sandstone layers; and (6) high-relief surfaces cutting through tidal sandstone layers. The majority of these erosion surfaces can be related to local depositional processes such as: (1) erosion in troughs separating centimeter- to meter-high bedforms migrating under wave-driven or tidal currents; (2) short-term periods of erosion during individual storms, river floods, or periods of enhanced tidal currents preceding deposition of decimeter- to meter-thick sandstone beds; and (3) erosion between bars migrating in channels or across areas frequently influenced by strong tidal currents. Lags overlying erosion surfaces of different scale can be highly variable along strike, and in many cases surfaces of different scale can look quite similar in isolated vertical sections, making it difficult to objectively use the local character of the surface alone to distinguish its relative scale and stratigraphic significance. For example, although basal erosion surfaces of tidal inclined-bed sets are commonly marked by thick mud-chip lags, locally where a bedset is underlain by sandy deposits its basal erosion surface may not be overlain by a mud-chip lag. In such cases, bedset boundaries are defined only by the truncation of internal inclined beds, vertical changes in smaller-scale cross-strata set thicknesses, and abundances of internal shale drapes. Similarly, erosion surfaces of regional significance are defined locally by particularly thick or shell-rich lags. In most cases, however, regional erosion surfaces are recognized not by the character of the surface itself but rather by abrupt changes in facies or breaks in larger-scale vertical facies trends, by demonstrating that the surface continues for long distances laterally, and/or by erosional relief substantially greater than the basal surfaces of adjacent bedsets. Three types of erosion surfaces have lateral extent and/or relief substantially greater than that of adjacent depositional bedsets; these surfaces may have broader stratigraphic significance. Erosion Surfaces at the Bases of Tidal Sandstone Layers Vertical transitions from wave-dominated deposits upward to tidal sandstone layers are nearly always erosional, and they define pronounced changes in depositional processes that can extend laterally for tens of kilometers (greater than the several kilometers spanned by individual tidal inclinedbed sets). Van Wagoner (1990) interpreted these erosional transitions to be high-frequency sequence boundaries (as discussed in Previous Interpretations, below). However, erosional relief at the base of tidal sandstone layers, defined by incision of these surfaces into underlying layers of hummocky cross-stratified sandstone beds, is no greater than basal erosional relief of individual tidal inclined-bed sets. Thus, the depth of erosion defined by surfaces at the base of tidal sandstone layers is no greater than that which would be expected from the amalgamation of tidal bar and channel deposits observed within the overlying sandstone layers. Local discontinuities in erosion surfaces at the base of tidal sandstone layers (e.g., between sections 7 and 8, and between sections 10 and 11, on the basal tidal sandstone layer in the Westwater Canyon; Fig. 3B) further suggest that these surfaces may have formed by a broad-scale amalgamation of localized tidal scour around prograding tidal bars. On the basis of this observation, Willis and Gabel (2001) interpreted many of the transitions from wave- to tide-dominated deposits to be within stratigraphically conformable successions formed as sea level fell and/or as strengthening tides scoured the sea floor and transported sands into deeper-water areas. In modern tidal systems, sand deposition in shallow subtidal areas is commonly associated with greater than 10 m of local erosion (Johnson and Belderson 1969; Hori et al. 2002; Allison 1998; Harris et al. 1993; Harris et al. 1996; Kuehl et al. 1989; Kuehl et al. 1997). Furthermore, many recent studies have reported abrupt vertical coarsening of various types of shoreline successions where sediments rapidly prograded, particularly where shorelines are forced to shift basinward by falling sea level (e.g., Walker and Plint 1992; Posamentier et al. 1992; Helland-Hansen and Gjellberg 1994). If the Sego tidal sandstone layers are such sharp-based shoreline successions, vertical transitions from wave-dominated to tidal sandstones would simply record rapid sand progradation as tidal current influence increased (a regressive surface of erosion) rather than to the development of a lowstand surface of erosion (i.e., Exxon type sequence boundary). Erosion Surfaces at the Tops of Tidal Sandstone Layers Erosion surfaces capping tidal sandstone layers, commonly overlain locally by an oyster-shell lag, clearly truncate underlying tidal bedsets and mark an abrupt grain-size fining. These surfaces could have formed when the locus of sand deposition shifted to another location along the coast and tide reworking continued. However, these surfaces have very low relief and

8 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 253 are more likely to have formed as wave ravinement planed off tidal bedform topography after tidal currents stopped dominating sand deposition and basinward transport of sand decreased. Because erosion surfaces capping tidal sandstone layers are abruptly overlain by marine shales, termination of tidal deposition probably records transgression, restriction of sand deposition to more proximal areas, and overall deeper-water conditions. Estimates of the depth of erosion associated with transgressive wave ravinement vary from centimeters to tens of meters; maximum values for other shoreline sandstones in the Cretaceous Interior Seaway average from 10 to 20 m (Nummedal and Swift 1987; Bergman and Walker 1988; Posamentier and Chamberlain 1993; Bhattacharya 1993; Bhattacharya and Willis 2001). Ten meters of erosion during transgressions could have systematically stripped off paleosols and paralic facies deposited on delta tops or valley interfluves, and thus it is difficult know whether regression during the deposition of particular sandstone layers within the Sego Member was associated with subaerial exposure of areas now located in the Book Cliffs. Similarly, it should not be assumed that areas along the regional cross section (Fig. 3A) that contain only upward-coarsening hummocky crossstratified deposits were never influenced by tidal currents. It may be that the entire stretch of coastline spanned by this study was periodically influenced by strong tidal currents, but evidence of these tidal currents was preserved following transgressive ravinement only in areas where a sufficient thickness of tidal sandstone had accumulated. This latter scenario is supported by the channel-shaped tidal sandstone preserved near Cottonwood Canyon (Fig. 3A), which is isolated from the thicker tidal sandstone layers along the same stratigraphic interval exposed to the west and east. High-Relief Erosion Surfaces within Tidal Sandstone Layers There are a few erosion surfaces within the tidal sandstone layers that have relief significantly greater than the thickness of individual depositional bedsets; these erosion surfaces can incise entirely through tidal sandstone layers (Figs. 5, 6). These surfaces contrast with basal erosion surfaces of tidal inclined-bed sets and channel deposits, which typically have basal erosional relief similar to or less than the thickness of overlying sets of inclined beds deposited during bar migration. Such high-relief erosion surfaces cutting ancient shoreline successions are commonly interpreted to have formed by the incision of rivers during a sea-level lowstand. Basal erosion surfaces of ancient valley fills are generally depicted to rise along strike upward to an interfluve paleosol, and areas of deepest incision to be overlain by fluvial channel deposits. In many ancient shoreline successions, however, these relationships are not preserved. Estuarine tidal currents can rework lowstand fluvial deposits once the valley is flooded (Dalrymple et al. 1992; Allen and Posamentier 1994), and ravinement can remove deltatop paleosols and paralic facies during subsequent transgression (Nummedal and Swift 1987; Bergman and Walker 1988; Posamentier and Chamberlain 1993; Bhattacharya and Willis 2001). Although prograding shorelines are expected to leave vertical facies trends that record shoaling, facies trends within valley fills can vary greatly depending on the rate of shoreline flooding relative to sediment supply to the valley (Thomas and Anderson 1994; Willis 1997). If sediment supply is relatively slow, valleys are rapidly flooded and estuarine tidal currents are likely to rework lowstand fluvial deposits. In such cases, marine sandstones may fill the mouth of the valley, estuary bay shales can accumulate in the center of the fill, and small riverdominated deltas can prograde from the head of the estuary (Dalrymple et al. 1992; Zaitlin et al. 1994). If sediment is supplied faster, the valley becomes filled with lowstand fluvial-channel deposits during the initial stages of relative sea-level rise, and the deposits gradually fine upward to retrogradational bayhead delta deposits as the rate of transgression quickens (Thomas and Anderson 1994; Willis 1997). The character of a valley fill can also vary along depositional dip, because of changes in the rate of relative sea-level rise and/or sediment supply during transgression (Zaitlin et al. 1994). Because facies patterns can be complex within shoreline deposits cut by valley fills, interpretation of the origin of high-relief erosion surfaces in the Sego depends on documenting how the positions of these surfaces change within tidal sandstone layers in association with interpretations of larger-scale facies trends (see below, and also more detailed depositional interpretations of specific areas along these lower Sego Member outcrops in Willis and Gabel 2001). Relationship between Facies Variations and Erosion Surfaces within the Lower Sego In the eastern delta deposit (i.e., Westwater Canyon outcrop) erosion surfaces with the greatest relief penetrate downward from the top of the lower Sego interval; these surfaces locally cut entirely through the upper tidal sandstone layer into the layer below. A surface with significant relief also cuts down from the middle of the lower Sego near Prairie Canyon (i.e., from the top of the lower tidal sandstone layer; Fig. 3A). In the western delta deposit (i.e., Thompson Canyon outcrop), similar high-relief erosion surfaces descend from within the uppermost tidal sandstone layer, and one surface cuts entirely through the lower Sego Sandstone into the underlying Buck Tongue Shale. In this western area, other surfaces of somewhat lower relief ( 10 m) also incise downward from the middle of the lower tidal sandstone layer (e.g., near Sagers Canyon), but most of these surfaces define narrow channel-shaped bodies that contain laterally accreted bars of nearly the same scale as their basal erosional relief. In the western delta deposit the lower tidal sandstone layer grades upward above its erosional base from tidal inclined-bed sets, locally, to channel-form tidal deposits, and finally to sandstones completely bioturbated with Ophiomorpha. This succession has been interpreted to record prodelta tidal erosion during progradation of a tide-dominated delta front, shoaling and local preservation of delta top channels, and finally a reduction in sediment supply and marine bioturbation during the last stages of regression and the start of transgression (Willis and Gabel 2001). Facies trends are similar in the eastern delta deposit, except near Prairie Canyon, where a high-relief erosion surface cuts out the lower sandstone layer. Above this erosion surface are tidal inclined-bed sets overlain by channel-form bedsets. Locally a coaly paleosol horizon caps this incised fill; elsewhere overlying channel-form bedsets cut out this paleosol horizon. Thus, like the adjacent deposits, those above the high-relief erosion surface appear to shoal upward. In this tidal sandstone layer the shallowest-water facies are above the high-relief erosion surface rather than occurring on adjacent interfluve areas as would be expected for a lowstand incised valley overlain by a transgressive fill. A laterally continuous shale extending through the middle of the lower Sego sandstone records flooding of the delta top, and a shift in the locus of deposition landward (to the northwest). Because this shale extends across both areas containing tidal sandstones and across the top of the intervening hummocky cross-stratified succession as well, it is interpreted to record a regional shoreline transgression. Thin tidal sandstones above this transgressive shale, in most locations only a single tidal bar thick, are observed near the axis of each delta. A thin bench of hummocky cross-stratified sandstone connects these tidal sandstones across the area, separating the eastern and western delta sandstones. This layer may represent a minor regional regression during a more extensive transgression. More likely, however, this sandstone layer may reflect a temporal change in the intensity of tidal currents in the seaway. It may be that shifts in the position of the shoreline took the basin in and out of tidal resonance (Pugh 1987), and when the tides resonated sands could be transported greater distances basinward into deeper water areas. The tidal sandstone layer that constitutes the upper part of the lower Sego Member is thicker than those below, and it is cut by numerous deep incisions (Figs. 5, 6). Deposits below these incision surfaces coarsen upward overall, and these are interpreted to record delta progradation. The deepest incision into each delta is relatively narrow (Fig. 3A, arrows). Near

9 254 B.J. WILLIS AND S.L. GABEL

10 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 255 their axes of incision these surfaces are overlain by coarse-grained channelform bedsets that are dominated by southeast-dipping, angle-of-repose cross strata. Because these channel sandstones show little evidence of tidalcurrent reworking, contain abundant plant fragments, and lack shell fragments, they may be of fluvial origin. Fills above shallower incisions are quite variable; some coarsen-upward from more heterolithic to more homogeneous tidal deposits, whereas others fine upward from tidal bar deposits to more heterolithic channel deposits. Except for the two most deeply incised examples, deposits in these fills are composed mostly of tidal inclined-bed sets and they could be interpreted to progressively shoal upward as bedsets become sandier or more channel-form in character (in many cases, however, changing relative water depths responsible for particular vertical facies trends are ambiguous). Thus, like the one example in the underlying tidal sandstone layer, incision fills do not reflect simple transgression of facies upward to capping marine shales, as might be predicted for fills of transgressed river valleys. An isolated 10-m-thick incision, filled with tidal sandstones, in the finergrained interval between the two delta deposits indicates that the rate or extent of tidal sand accumulation was generally higher during deposition of this second tidal sandstone layer. A final thin tidal sandstone layer within the Anchor Mine Tongue exposed in the Westwater Canyon area appears to be confined to the axis of the eastern delta, and, like the thin tidal sandstone above the first thick tidal sandstone layer, it may record a temporary increase in tidal-current strength associated with changes in basin shape during a longer-term transgression. Therefore, this final tidal sandstone layer probably does not record the same magnitude of change in water depth or shoreline position as the two thicker tidal sandstone layers in the underlying lower Sego Sandstone. SEGO STRATIGRAPHIC MODELS Previous Interpretations High-Frequency Sequences Van Wagoner (1991) postulated that erosion surfaces at transitions from wave-dominated to coarser-grained tidal deposits record abrupt sea-level falls that shifted depositional systems tracts basinward and allowed braided rivers to incise into offshore marine deposits. He further suggested that regressive shoreline deposits were not preserved because sea level fell very rapidly and thin regressive deposits were completely reworked during subsequent transgression. Sego tidal sandstones were inferred to be lowstand estuary deposits confined to incised valleys filled during early stages of sea-level rise. On the basis of these interpretations, and his lateral tracing of facies and erosion surfaces between measured sections spanning about 40 km of the eastern part of the Book Cliffs outcrop belt, he suggested that there are seven regional sequences within the lower Sego through Anchor Mine Tongue interval (i.e., the interval shown in Fig. 3A). Recognition of regional erosion surfaces that define his sequences depended critically on observing transitions from wave-dominated to coarser-grained tidal facies and tracing surfaces marking these transitions laterally through areas where layers of tidal deposits are amalgamated. Although the interpretation that the Sego Sandstone contains numerous high-frequency sequences provides an explanation for the interlayering of tide- and wave-dominated deposits and the presence of multiple high-relief erosion surfaces, this hypothesis has several weaknesses. The mechanism to produce high-frequency sea-level variations of a magnitude that can produce valleys 30 m deep in clastic systems during a nonglacial greenhouse time like the Cretaceous is not well documented. The idea that sea level fell so fast that there was no time for deposition during shoreline regression is not supported by the numerous recent studies reporting the significance of falling-stage deposits as components of sandstone wedges in the Cretaceous Western Interior and in clastic wedges that prograded into similar shallow seas elsewhere (see reviews in Plint and Nummedal 2000; Posamentier and Morris 2000). While it is quite possible for sediment bypass to occur during the regression of local areas, it seems less likely that this occurred uniformly over 100 km along strike during many different regression transgression cycles. The idea that regressive shoreline deposits were uniformly reworked into estuarine embayments as incised river valleys were flooded is not supported by the dominance of ebb-oriented cross strata and ebb-dipping inclined beds within bedsets throughout tidal sandstone layers. Further, there is no evidence that the various erosion surfaces defining these high-frequency sequences pass upward into paleosols or other types of deposits indicating subaerial exposure of interfluves during lowstands. Rather the shallowest-water facies within the Sego Sandstone appear to overlie the deepest incisions into tidal sandstone layers. Sequence stratigraphic models show a single valley incised into a prograding delta during falling sea level (e.g., Posamentier and Vail 1988; Van Wagoner et al. 1990). Van Wagoner s (1991) interpretation of the Sego Sandstone suggests that each sequence boundary is defined by multiple locations of deep river incision. Van Wagoner et al. (1990) explained similar sets of multiple valleys incised into the top of the Castlegate Sandstone by suggesting that individual channels in a braided river incised different valleys. This explanation seems unlikely. Braid bars are formed and reworked in equilibrium with the full river discharge, and thus they are unlikely to provide sites of permanent flow division during river incision. Because flow switches readily between braided channels, one channel within the river would be expected to rapidly capture all the discharge during incision where adjacent braided channels could not together sweep laterally to carve a single valley. It is also unlikely that each incision along a high-relief erosion surface represents a different river, inasmuch as large rivers are not typically spaced every few kilometers along a coastline. Most sediment is delivered to foreland basins from drainages within the adjacent orogenic belt, and only rivers draining relatively large areas can maintain their course through the rising thrust belt (Gupta 1997). Thus it is expected that the main sediment conduits to a foreland basin will be moderately large rivers spaced many tens of kilometers apart along the mountain front. Interpreting each highrelief incision into the lower Sego Member to be a different valley cut by a single river channel would suggest that these deposits comprise many FIG. 5. Architecture of lower Sego deposits exposed along selected segments of the Book Cliffs outcrop. A) Outcrop just east of Westwater Canyon where a high-relief erosion surface cuts down into a relatively sand-rich part of the lower Sego sandstone and is overlain by relatively finer-grained tidal deposits (see Fig. 3B, between km 7 and 9). The lower Sego is composed of two tidal sandstone layers. The upper tidal sandstone layers is an upward-coarsening stack of five tidal inclined-bed sets capped by an erosion surface on the left side of the set of photographs. The capping erosion surface gradually cuts down to the left through this tidal sandstone layer. Deposits above this erosion surface are more heterolithic and thus have a more recessive weathering style. The lower tidal sandstone layer thins dramatically to the right as the basal tidal bedset terminates and the layer becomes a thin bioturbated red sandstone (Fig. 6A, B). B) Outcrop just west of Thompson Canyon where a high-relief erosion surface cuts down into the Buck Tongue from the middle of the upper tidal sandstone layer in the lower Sego interval (Fig. 3C, between km 1 and 2; Fig. 6C, D). On the left side of the photo the lower Sego interval is composed of two distinct tidal sandstone layers separated by a coarsening-upward succession of hummocky cross-stratified deposits. A high-relief erosion surface at the base of the upper layer descends to the right, cutting out the lower layer. There deposits above the erosion surface are homogeneous channelized sandstones that pass upward into more heterolithic tidal sandstones. C) Two tidal sandstone layers exposed in San Arroyo Canyon (Fig. 3B, between km 13 and 14). A high-relief erosion surface in the upper tidal sandstone layer is highlighted by a laterally continuous cement bed. This surface descends beyond the photograph to the right, where it cuts out the lower sandstone layer.

11 256 B.J. WILLIS AND S.L. GABEL FIG. 6. Two sets of sedimentologic logs show that contrasting facies variations occur across high-relief erosion surfaces (see location of logs on photos in Fig. 5). From log A to B a high-relief erosion surface rises from the base to the top of the upper layer of tidal deposits. Deposits filling the incision above the erosion surface are far more heterolithic than those cut out by the erosion surface. The basal tidal inclined-bed set in the lower tidal sandstone layer ends laterally from A to B, and this layer continues farther to the east (right) as a thin bed of bioturbated red sandstone. There is a discontinuity in the basal erosion surface of the lower Sego where this bedset terminates. A high-relief erosion surface descends from the base of the upper tidal sandstone layer into the Buck Tongue between logs C and D. Where the erosion surface is incised deepest, it is overlain by relatively homogeneous channelized sandstones that are interpreted to be fluvial. These deposits pass upward into tidal inclined-bed sets, and local capping channel-form tidal sandstones. tens of sequences (i.e., one sea-level fall to produce each of the deep incisions observed to cut tidal sandstone layers). Alternative Origins for Deep Incisions In the lower Sego interval, there are many deep incisions cut into each tidal sandstone layer. Three hypotheses to explain multiple incisions into individual prograding tide-dominated delta sandstones are discussed: (1) incision of distributary channels (Fig. 7A); (2) shifting positions of incision related to upstream river avulsion (Fig. 7B); (3) tidal erosion of shorelines (Fig. 7C). Although these hypotheses are not mutually exclusive, each is initially presented below as an end-member case. In large prograding delta systems, the feeding river can begin to divide into distributaries many tens of kilometers upstream of the shoreline, starting where the channel passes onto the relatively flat delta top. With sealevel fall, channel incision occurs first at the knickpoint where depositional slopes abruptly increase on the newly exposed edge of the delta front (Butcher 1990; Posamentier et al. 1992; Koss et al. 1994; Burns et al. 1997;

12 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 257 Blum and Törnqvist 2000). Above knickpoints, the distributaries remain unaffected by sea-level fall and shoreline regression. Erosion of knickpoints into the former delta front edge associated with each distributary channel migrates landward with time. When knickpoints of individual distributaries migrate upstream to a distributary channel bifurcation, the more deeply incised distributary captures the combined flow. Only when a knickpoint migrates to the head of the delta top, where the first distributary bifurcation occurs, does the entire flow of the river pass down a single incised channel. This incised channel then continues to bypass sediment farther into the basin during continued sea-level fall. This hypothesis predicts that the number of deep incisions increases toward the exposed delta front edge, and that the longest-lived (thus presumably the widest) incision cuts deepest into underlying delta deposits and extends the greatest distance landward. The second hypothesis (Fig. 7B) suggests that the river can continue to avulse during early stages of sea-level fall. It is similar to the first hypothesis in assuming that valley incision begins first at the newly exposed delta front and that erosion of knickpoints takes time to migrate upstream from this initial incision. It differs in suggesting that the main river channel continues to avulse in areas landward of this knickpoint, and thus multiple incisions are assumed to form sequentially in response to successive river avulsions. In this case, the duration of erosion forming each incision need not be related to its relative age, but rather would reflect the time between random river avulsions. Younger river incisions, formed closer to the time of maximum sea-level lowstand, may incise more deeply, but they need not be the longest-lived. Thus in this case the depth of river incision near the delta front may not be related to incision width or length. The final hypothesis (Fig. 7C) suggests that tidal currents can significantly incise the delta front, either by widening preexisting distributary channels or simply by eroding areas where tidal currents are enhanced by shoreline geometry and sea-floor bathymetry (Harris et al. 1993; Harris 1994; Baker et al. 1995; Dalrymple 1998; Reynaud et al. 1999b; Hori et al. 2002). Modern tide-dominated river deltas have incisions greater than 10 m deep into prograding delta fronts that have formed during the recent sea-level highstand (Fig. 8). For example, distributary channels in the Fly River Delta (about 5 7 m deep) are eroded to depths of over 30 m at their mouth through tidal scour after they are abandoned (Baker et al. 1995; Harris et al. 1996; Dalrymple et al. 1990; Dalrymple personal communication, 2000). Similarly, delta fronts of the Ganges and Yangtze rivers have tide-enlarged distributaries over 10 m deep in areas not currently receiving active sediment supply (Hori et al. 2002; Kuehl et al. 1989; Kuehl et al. 1997). While tidal enlargement of distributary channels can clearly produce deep incisions into the delta front, these incisions probably would not extend as far inland as river-cut lowstand valleys. Tidally enlarged distributary channels can fill as a consequence of river reoccupation and continued delta progradation. Thus, unlike the two previous hypotheses, tidal erosion and subsequent filling of incisions could occur at any time during delta progradation and retreat, and formation of deep incisions into the delta front need not be restricted to times of sea-level fall and lowstand. The deepest tidal incisions into a delta front would probably reflect times when tidal range was greatest and/or when distributaries were abandoned for longer periods of time. Changes in basin shape during the progradation and transgression of shorelines can change the extent to which tides are enhanced by interaction with the sea floor or amplified by resonance within the basin. Tidal resonance is greatest when the basin width matches a harmonic of tidal wavelength, a condition that is not related to specific times during regression transgression cycles. Thus, changes in tidal-current strength need not be related to sequence stratigraphic divisions. Tidally resonant basin dimensions are most likely to occur during times when shoreline positions are changing most rapidly: either times of most rapid relative sea-level change or during lowstands when distal, low-slope parts of the basin floor are being regressed or flooded. Enhanced influence of tidal currents may also be more likely during lowstands when seas are shallow and lowered accommodation distributes FIG. 7. Three hypotheses to explain multiple incisions into individual prograding tide-dominated delta sandstones: A) incision of distributary channels; B) switching of the valley position related to upstream channel avulsion; C) tidal erosion of the delta front. Each of these mechanisms could produce high-relief erosion surfaces that are not part of a regionally continuous sequence boundary.

13 258 B.J. WILLIS AND S.L. GABEL FIG. 8. Maps of bathymetry on three modern tide-dominated deltas shown at the same scale. These deltas have all prograded tens of kilometers since being transgressed during the last Holocene sea-level rise. Surface topography reflects tidal erosion of subtidal delta-top areas, tens of kilometers wide. A) Ganges Brahmaputra River delta shows tidal channels, 5 10 m deep, in the lobe that is being actively supplied by sediment from the river (to the right), and slightly greater erosional relief along the coast on the abandoned delta lobe with lesser sediment supply to the left (simplified from Kuehl et al. 1997). B) Yangtze River delta also shows m deep tidal distributaries cut into the subtidal part of the delta top. Also note the more extensive tidal erosion of shoreline areas to the north where there is less active supply of sediment (simplified from Hori et al. 2002). C) Fly River delta deltaic deposition over wider areas. Shallow seas would dampen waves, avulsion of distributary channels may take longer to reoccupy specific areas of broad lowstand deltas, and minor basin-floor topography may locally focus tidal currents (Bhattacharya and Willis 2001; Martinsen 2001). Tidal erosion may be particularly pronounced during the initial inundation of flat delta tops when wider shallow areas become susceptible to tidal reworking. Continued transgression increases accommodation in more proximal delta areas, leading to longer times before the main river channel reoccupies areas with tidally enlarged, abandoned distributaries. Thus, it is possible that tidal incisions would be more common at the lowstand delta edge relative to the fronts of deltas that prograded during highstand and fallingstage times. That said, the pronounced tidal enlargement and deepening of distributaries reported on the modern Fly (Baker et al. 1995; Harris et al. 1996; Dalrymple personal communication, 2000), Ganges (Allison 1998; Kuehl et al. 1989; Kuehl et al. 1997) and Yangtze (Hori et al. 2002) river deltas occurred during Holocene times as these tidal systems prograded many tens of kilometers into their respective basins. The presence of multiple high-relief incisions into the Sego deltas has important implications for recognizing sequence boundaries within these prograding tide-dominated deltaic successions. All three end-member cases described above suggest that deep incisions cut into individual delta-front deposits may not reflect a single regionally continuous lowstand erosion surface (i.e., a single allostratigraphic sequence boundary ). The first two cases imply that distributary piracy or upstream river avulsion limit the time available for upstream migration of individual knickpoints. Thus, different incisions observed in a cross section of the delta deposit may not be physically connected or contemporaneous. In the case where rivers are assumed to avulse landward of the incision, it is possible to have subsequent incisions cut into the lowstand delta deposited basinward of a previous incision with continued sea-level fall: This might result in local superposition of different erosion surfaces formed during the same episode of sea-level fall. In the third case, tidal erosion could leave shoreface incisions unrelated to relative sea level falls, and thus stratigraphic juxtaposition of these incisions also would reflect only local erosion of shorelines rather than multiple episodes of sea-level fall. Because knickpoint incision and tidal erosion are both most likely to occur at the edge of the delta front, where depositional slopes rapidly steepen basinward, it is expected that deep incisions predicted by all three hypotheses would cut down from the top of prograding delta deposits (i.e., horizons recording delta top pedogenesis and/or transgressive ravinement and flooding). Because erosion surfaces of different origin are expected to descend from essentially the same stratigraphic horizon, it may be difficult to distinguish different types of incised fills in the ancient record. In the first two cases deep incisions into the delta front would match standard criteria to define a sequence boundary: a lowstand erosion surface associated with a basinward shift in deposition (Van Wagoner et al. 1990). In both cases significant sandstone deposition is expected basinward of individual valleys, reflecting both the seaward transport of sediment eroded during valley incision and the bypass of sediment basinward through distributary or fluvial channels. Blum and Törnqvist (2000) indicate that there is typically greater than an order of magnitude more lowstand sediments than would be required to fill the associated valley. Therefore valleys typically bypass far more sediment to lowstand areas than is generated solely by river incision. With the first case, that of incised distributaries, the thickest accumulation of bypassed sediment is expected seaward of the longest and deepest incision. With the second scenario, that of multiple incisions formed due to avulsion, the most deeply incised valley need not be asso- shows a number of deep tidal incisions into its prograding delta front (simplified from Harris et al. 1993). D) Map of lower Sego exposures in the Book Cliffs (simplified from Fig. 2A) is shown here at the same scale as the delta maps.

14 STRATIGRAPHY OF SEGO TIDE-DOMINATED DELTAS 259 ciated with the thickest lowstand wedge of sediment. For the third case, that of tidal incisions, associated sediment accumulations lying basinward of incisions may be far less significant than in the first two cases. The three hypotheses also lead to different predictions for the character of incision fills. The first two cases suggest that the main river channel is trapped and the youngest incision inundated during the start of transgression. This incision receives all the sediment discharge of the river during the initial stages of relative sea level rise, and thus it has the thickest accumulation of basal, coarse-grained, lowstand fluvial deposits. The first case further suggests that coarse-grained fluvial deposits are found in the deepest, widest, and longest incision into the delta, whereas in the second case this coarse-grained incision fill need not be the largest. Such coarsegrained incision fills probably fine upward in either case, reflecting progressively more rapid sea-level rise, the vertical widening of the river valley, and eventual loss of sediment supply to other areas on the delta top during transgression. Other incisions, carved by distributaries before their upstream source was captured or abandoned by previous upstream avulsion, respectively, flood rapidly during early stages of sea-level rise. These areas receive sediment only after the main river avulses following transgressive filling of its incision, unless marine currents are strong enough to rework sediment landward into their mouth. Without a supply of sediment from upstream, any lowstand fluvial deposits in these fills are likely to be reworked by estuarine processes during flooding. These incision fills may coarsen upward, reflecting increased sediment supply across the delta top after the youngest incision becomes filled. By this time, however, the delta top may be largely flooded and fills of younger incisions may be composed only of small bayhead delta deposits and marine reworked facies. In the third case, deposits may generally coarsen upward, reflecting the reoccupation of tidally enlarged distributary channels on a prograding delta. However, distributary channels never reoccupied before significant transgression of the delta may fill with marine deposits or remain unfilled in deep waters areas (Harris et al. 1993). Basal lags may be dominated by shell material formed by the reworking of open marine delta-front deposits rather than clasts associated with fluvial or estuarine erosion. Variability within fills would reflect changes in the strength of tidal currents both spatially along the coastline and through time, as well as the rate at which incisions are reoccupied by the switching distributaries. Fills of channels formed during rapid delta progradation may be generally richer in sandstone and may overfill, recording gradual subsidence between the time of tidal incision and distributary reoccupation. Those formed as the delta backsteps may be finer grained and may be incompletely filled because of the gradual landward shift in deposition. In ancient deposits, it may be very difficult to distinguish those incisions carved by the lowstand erosion of rivers that were subsequently tidally scoured during transgression from those of wholly tidal origin, unless differences in their landward extent or the thickness of lowstand wedges can be recognized. Valley Fills and Tidal Erosion in the Sego Sandstone Incisions and their fills within the lower Sego deltas can be interpreted in the context of the three hypothetical cases described in the previous section. Only one incision into each delta has basal deposits interpreted to be fluvial channels (in the western delta near Crescent Canyon, and in the eastern delta near West Salt Creek; Fig. 3A, arrows). These incisions each cut deeper than any other within their respective deltas, and are interpreted to record a maximum relative sea-level lowstand. The interpretation that fluvial channel deposits occupy the bases of the deepest incised fills implies that at least some incisions are river valleys. The two most deeply incised fills are fairly narrow, and clearly do not reflect the volumetrically dominant incision into their respective deltas. This observation goes against the idea that incisions reflect downcutting by multiple distributaries, inasmuch as this hypothesis also implies that the largest and longest-lived river valley would also be the most deeply incised and have the coarsest-grained fill. In the eastern delta, the numerous deep incisions cut into the top of the lower Sego interval could arguably be bounded by a single, continuous erosion surface (e.g., Willis and Gabel 2001). It could, alternatively, be argued that the margins of some of these fills overlap (depending on the relative significance assigned to surfaces at the bases of different bedsets in the upper part of the tidal sandstone layer). A continuous capping oystershell lag is observed where there are no deep incisions cut into the top of the sandstone layer, suggesting that interfluve paleosols may have been eroded. Thus, it is possible that all these incisions record river valleys cut during a single lowstand as non-incised upstream segments of the river continued to avulse. However, it is also possible that in most cases these incisions are tidally enlarged distributaries, cut and filled as the delta prograded. This latter interpretation is supported by those fills that shoal upward from tidal inclined-bed sets to channel-form bedsets. Understanding the origin of the deepest high-relief incisions depends critically on the interpretation that coarse-grained channelized bedsets in the deepest fills are river deposits and not ebb-dominated tidal current lags, an interpretation that cannot be proven in this case. In the western delta, several high-relief erosion surfaces incise locally from stratigraphic layers higher than that of the deepest incision. These stratigraphically higher erosion surfaces can be followed along the bases of various tidal bedsets stacked within the sandstone layer, supporting the interpretation that they are all associated with a second surface of lowstand erosion (Willis and Gabel 2001). These high-relief erosion surfaces are not distinctive away from areas where they are most deeply incised, however, and there it becomes somewhat arbitrary to correlate them laterally across the entire area shown in Figure 3. A more likely interpretation is that incisions higher in the sandstone layer represent more localized tidal enlargement of distributaries as the delta backstepped. Above these upper incision surfaces, the lack of basal fluvial deposits, abundance of shell lags in the heterolithic lower parts of incision fills, and dominantly ebb-dipping cross-strata in the sandier deposits higher in fills support this latter interpretation. The fine-grained basal deposits within these fills may reflect slower reoccupation of abandoned distributaries as the delta gradually backstepped and less sediment was transported basinward. If the deepest incisions are floored by fluvial channel deposits (as interpreted above), their depth of stratigraphic downcutting implies that the whole lower Sego interval in the southern Book Cliffs was subaerially exposed during maximum lowstand (including locations that now contain only vertical successions of marine shales interbedded with hummocky cross-stratified sandstones). While there is clear evidence of transgressive erosion at extensive oyster-shell lags locally capping tidal sandstones of the lower Sego interval, evidence of deep transgressive ravinement is less obvious above the hummocky cross-stratified sandstone layers away from the delta depositional axes. The local preservation of a paleosol horizon above the first thick tidal sandstone in the lower Sego interval, and thicker, more extensive tidal sandstones cut by deeper incisions in the upper part of the lower Sego interval, support an overall regressive pattern of deposition. Thus, it seems likely that there was extensive ravinement at the top of the lower Sego Sandstone that removed capping paleosols, even in locations where ravinement lags are not observed on this surface. Regional Stratigraphic Patterns Two recent studies examined depositional changes along the Sego interval up depositional dip into fluvial-dominated deposits along the western Book Cliffs. The first study (reported in Yoshida et al. 1996, Willis 2000, and Yoshida et al. 2001) suggested that the Sego Sandstone is underlain by a major sequence boundary that defines an angular unconformity with respect to the deposits of the Buck Tongue (Fig. 3A). Their interpretations supported those of Van Wagoner (1991) in suggesting that shorelines moved so rapidly into the basin that few regressive deposits formed and none were preserved following lowstand and transgressive reworking. The

15 260 B.J. WILLIS AND S.L. GABEL FIG. 9. Two interpretations of the landward correlation of the Sego Member and adjacent units along the western segment of the Book Cliffs outcrop belt. A) Interpretation simplified from Willis (2000) indicates that the Buck Tongue shale is truncated by an angular unconformity beneath the Sego Sandstone Member. A finegrained, tidally influenced interval in the fluvial succession near Price, Utah, is correlated with transgressive, tide-influenced fluvial and coastal-plain deposits of the Neslen Formation in the southern Book Cliffs. B) Interpretation simplified from McLaurin and Steel (2000) suggests that the Buck Tongue correlates with the fine-grained interval near Price, and that the Neslen Formation is dominantly a regressive succession. Sego Sandstone in the southern Book Cliffs was therefore interpreted to be a predominantly transgressive succession, passing upward from amalgamated sandy lowstand estuarine valley fills to fine-grained, tidally influenced fluvial deposits. An interval of tidal deposits within the dominantly fluvial succession at the northern end of the Book Cliffs was related to maximum shoreline transgression associated with a fine-grained interval within the Neslen Formation. They suggested that the smaller-scale stratigraphic variations mapped within the Sego Sandstone by Van Wagoner et al. (1990) and Van Wagoner (1991) do not extend very far landward because increased subsidence rates toward the mountain belt did not allow shoreline incision to extend into proximal areas of the basin. Their stratigraphic cross section, however, does not show appreciable thickening of the Sego Sandstone interval toward the landward end of their cross section, as would be expected if subsidence rates increased landward.