Channel avulsion plays a significant role in building alluvial stratigraphy, including fluvial sand bodies with through-going erosion surfaces, called multistory sand bodies (MSBs); consequently, MSB characteristics may be useful for interpreting paleo-avulsion dynamics from fluvial deposits. Here, using a combination of literature review, geometric modeling, and field observations, we explore the degree to which MSB characteristics may be useful for interpreting paleo-avulsion dynamics from ancient deposits. We use published interpretations of MSB origins to identify characteristics that are exclusively associated with avulsion-related MSBs and find that vertical story stacking, irregular MSB bounding surfaces, and floodplain deposits that correlate with individual stories are attributes uniquely associated with avulsion-origin MSBs. To evaluate how avulsion patterns may affect MSB properties, we use a 2D geometric model to build synthetic stratigraphy and compare characteristics for MSBs formed under different avulsion patterns at a range of model-aggradation rates. Avulsion patterns include random (equal probability of a channel relocating anywhere across the model domain), compensational (channels avulse to the lowest elevation in the model domain at each time step), and clustered (where channel relocation is restricted to a zone near a previously occupied location). Model results show that at moderate to high aggradation rates, MSBs formed under clustered avulsion patterns can be differentiated from those formed via random and compensational avulsion patterns. Specifically, MSBs formed by clustered avulsion patterns have many stories and commonly contain remnant floodplain deposits. We apply these insights to the well-exposed lower Williams Fork Formation (Upper Cretaceous, Piceance Basin, Colorado, USA) to demonstrate how they can be used in practice. Fifty-four percent of lower Williams Fork MSBs contain clear evidence for avulsion reoccupation, 32% have characteristics suggesting intra-channel-belt origins, and the remaining 20% have ambiguous evidence or insufficient exposure to permit confident process interpretations. Overall this study demonstrates that despite inherent challenges in eliciting information about avulsion dynamics from the stratigraphic record, some MSB characteristics can be interpreted as clear signals of paleo-avulsion processes.


Alluvial stratigraphy is influenced by basin subsidence rate, sediment and water supply, and sedimentary process dynamics, including channel avulsion (e.g., Leeder 1978; Allen 1979; Bridge and Leeder 1979; Heller and Paola 1996). Isolating avulsion-specific signals from the sedimentary record is important for testing hypotheses about the effects of avulsion on alluvial stratigraphy, including the degree to which avulsion rates and patterns influence alluvial architecture (e.g., Leeder 1978; Allen 1979; Bridge and Leeder 1979; Mackey and Bridge 1995; Heller and Paola 1996). However, interpreting paleo-avulsion dynamics from ancient deposits is challenging because channel deposits reflect a mix of fluvial processes, including channel meandering and avulsion, bar migration, and erosion and deposition in response to variations in discharge or base level (e.g., Bridge and Diemer 1983; Miall 1985; Gibling and Rust 1990; Mohrig et al. 2000; Holbrook 2001; Gibling 2006; Labourdette and Jones 2007).

Channel-belt sandstones often contain through-going surfaces that define stratigraphic “stories” (Friend et al. 1979; Gibling 2006) and broadly indicate transient sediment deposition and erosion events within an ancient fluvial landscape. For example, channel threads switching within a channel belt (e.g., Johnson and Pierce 1990; Leleu et al. 2009), seasonal discharge fluctuations (e.g., Tunbridge 1981; Olsen 1989), and channel avulsions (e.g., Kraus and Gwinn 1997; Mohrig et al. 2000) can all generate channel complexes with internal erosional surfaces, called multistory sand bodies (MSBs). These amalgamated sand bodies contain important records of paleo-river dynamics, including avulsion processes, but it remains difficult to uniquely interpret processes responsible for forming MSBs.

Here we explore whether MSBs can be used to interpret paleo-avulsion dynamics from ancient deposits. To this end, we use published studies to evaluate the degree to which avulsion-derived MSBs can be uniquely distinguished from MSBs resulting from other processes, and we investigate how large-scale avulsion patterns may influence MSB formation and preservation with a simplified geometric basin-filling model. We apply these insights to the well-exposed lower Williams Fork Formation (Upper Cretaceous, Piceance Basin, Colorado, USA) to demonstrate how they can be used in practice. We find that despite inherent challenges in eliciting information about ancient depositional processes from the stratigraphic record, some MSB characteristics can be interpreted as clear signals of paleo-avulsion processes.

Avulsion Processes and the Origins of MSBS

Avulsion processes have long been known to play a significant role in alluvial basin filling, particularly channel-body architecture and distribution (e.g., Leeder 1978; Allen 1979; Bridge and Leeder 1979; Heller and Paola 1996). Channel avulsion can directly lead to MSB formation when channel belts avulse onto or “reoccupy” older channel deposits (e.g., Mohrig et al. 2000) (also called “annexation,” cf. Slingerland and Smith 2004) resulting in an amalgamated, composite sand body. It is therefore reasonable that fluvial systems associated with a high likelihood of avulsion reoccupation might produce many MSBs that record some aspects of paleo-avulsion dynamics (e.g., Mohrig et al. 2000).

In practice, using MSBs to reconstruct paleo-avulsion dynamics is difficult not only because multiple processes can produce MSBs, but also because defining and identifying MSBs is inherently qualitative. Formally defined by Friend et al. (1979) as “the vertical sequence of storeys within any sandstone body,” the definition of MSB has broadened to “a sand body of one cycle [that] is superimposed upon one or more earlier sand bodies” (Gibling 2006), including laterally stacked channel elements (originally termed “multi-lateral” bodies by Friend et al. 1979). These generic definitions allow stratigraphers to apply the term to many types of sand bodies at many scales. For example, different publications define stories as bar-scale deposits (e.g., Marzo et al. 1988; Labourdette and Jones 2007), channel-thread deposits (e.g., Matthews et al. 2007; Kjemperud et al. 2008), or channel-belt deposits (e.g., Mohrig et al. 2000; Arche and López-Gómez 2005).

Story-generating processes can be grouped broadly into three main categories: 1) intra-channel-belt processes, 2) avulsion reoccupation, and 3) incised-valley processes (Fig. 1). Intra-channel-belt processes is a catch-all grouping of many processes acting at small spatiotemporal scales within an active channel belt. Many common fluvial processes within a channel belt can generate scour surfaces that lead to story formation, including bar migration (e.g., Diemer and Belt 1991; Kumar 1993; Labourdette and Jones 2007) (Fig. 1A), channel-thread migration (e.g., Johnson and Pierce 1990), meander cut-off (e.g., Platt and Keller 1992; Corbett et al. 2011), and seasonal discharge fluctuations (e.g., Tunbridge 1981; Olsen 1989; Limarino et al. 2001).

Avulsion reoccupation, wherein an avulsion channel localizes atop a previously abandoned channel-belt deposit, can occur at a range of scales (Heller and Paola 1996). Small-scale channel avulsions can occur in or near an active channel belt (Slingerland and Smith 2004), but these are unlikely to be stratigraphically differentiable from other intra-channel-belt processes, like channel migration or meander-bend cutoff. At larger scales, channel-belt avulsions can have a significant impact on alluvial basin filling and may be relatively “local,” where channel relocation is limited to a given channel-belt reach, or “regional,” where the entire channel belt downstream of the avulsion site relocates (Mackey and Bridge 1995; Heller and Paola 1996) (e.g., Fig. 1B). This study focuses on identifying MSBs derived from channel-belt-scale avulsions, because they provide information on long-term avulsion patterns and may pose particular challenges for reservoir connectivity (Bridge and Mackey 1993).

MSBs can also form when a fluvial system is laterally confined, causing successive scour and reworking of underlying deposits (Fig. 1C). This type of lateral confinement often occurs in incised valleys, where base-level fall or increased water discharge causes downcutting, trapping fluvial systems in a container valley (e.g., Blum et al. 1994; Shanley and McCabe 1994). Faults, salt-walled mini basins, or other structural features can also alter landscape topography such that fluvial systems are restricted to one location (e.g., Arche and López-Gómez 2005; Banham and Mountney 2013). Fluvial systems in an axial basin position may be more likely to be laterally confined by existing topography, as opposed to transverse, distributive systems, which avulse across unconfined fluvial fans (Weissmann et al. 2010). Because incised valleys are important for sequence stratigraphy, a large body of literature exists about identifying incised-valley systems in the stratigraphic record (see Gibling et al. 2011 for a thorough summary and evaluation). These analyses, however, have not specifically delineated the differences between MSBs resulting from incised-valley and avulsion processes.

Published Field Evidence For Avulsion-Type MSBS

To distinguish which MSB characteristics are most useful for process-origin interpretations, and are unique to avulsion-type MSBs, we examined 50 peer-reviewed studies published between 1970 and 2011 that interpreted fluvial MSB origin from outcrop exposure. We included only studies that explicitly interpreted the process origin of MSBs and used the term “multistory,” although these interpretations were not always the central focus of the publication. In a few instances where an amalgamated sand body was formally termed “multi-lateral” but its internal components were described as “stories,” we considered the sand body eligible for this comparison. This is not a comprehensive review of all publications that interpret MSB origins, but it presents a representative selection of publications in which authors explicitly interpreted MSB process origin.

For each publication we recorded the evidence used to make the process interpretation and categorized the interpreted origin as either intra-channel-belt processes (including bar migration, channel-thread switching, and seasonal discharge fluctuations), avulsion reoccupation, or incised valley. We assumed that interpretations in each publication were reasonable. Field evidence cited for each published process interpretation was grouped into a few broad categories, including MSB properties, MSB relationship with surrounding fine-grained lithofacies, and story properties (Fig. 2; Table 1; Supplement 1). For example, Aitken and Flint (1995) interpreted an incised-valley origin for MSBs in the Breathitt Group (Carboniferous, Kentucky, USA), and explicitly used the following field evidence and rationale for their interpretation: “It is the combination of (a) the degree of facies dislocation, (b) the marked contrast in grain size, (c) the degree of incision, (d) the contrast in depth of individual channel fills in relation to the total depth of incision and (e) the regional, mappable extent of the surfaces… [that leads to the incised valley interpretation].” In a different example, Jinnah and Roberts (2011) make an avulsion-origin interpretation for sheet sand bodies in the Wahweap Formation (Cretaceous, Utah, USA) citing pebble lags along undulating story contacts and the presence of channel abandonment deposits. They note that “upward-fining successions … record channel abandonment and pedogenesis in channel sediments” and that “reactivation surfaces were formed by channel migration and avulsion, leading to the formation of amalgamated channel sandstones.”

The resultant dataset reveals how commonly different types of field observations are used to interpret MSB process origins (Table 1). Half of the published field evidence is cited in multiple process-interpretation categories, suggesting that those characteristics are not useful for uniquely interpreting MSB origins. For example, channel-abandonment facies and evidence of subaerial exposure in MSBs intuitively seem like characteristics that would be associated with avulsion reoccupation, and are cited frequently by authors interpreting an avulsion origin for a MSB. However, these features can also result from common fluvial processes, such as temporary local abandonment due to channel migration or strongly seasonal discharge variations; consequently evidence of subaerial exposure at story tops is also commonly cited by authors as evidence of intra-channel-belt processes. This indicates that these observations are not directly useful for uniquely determining process origins of MSBs.

Some field evidence seems to be useful for uniquely interpreting the process origin of MSBs, especially characteristics of the MSB bounding surface and the relationship of the MSB and surrounding floodplain deposits (Table 1). A smooth MSB bounding surface (e.g., Fig. 2, right side) was cited only to interpret intra-channel-belt origin, whereas an irregular MSB bounding surface (e.g., Fig. 2, left side) was cited exclusively as evidence for avulsion-reoccupation origins of MSBs. Although smooth MSB bounding surfaces were not widely cited as evidence for intra-channel-belt origin (only 6% of intra-channel-belt papers), this characteristic is consistent with a simple cut-and-fill history, and also provides a logical contrast to irregular MSB bounding surfaces as evidence for avulsion-origin MSBs. Floodplain horizons that correlate with individual stories were also uniquely cited as evidence for avulsion reoccupation. In contrast, a high-relief, regionally extensive, and steep-sided MSB bounding surface and distinct floodplain or paleosol horizons that correlate with the top of the MSB were cited only in incised-valley-origin interpretations. These lines of evidence for incised-valley origin are consistent with the findings of Gibling et al. (2011), who note that “alluvial valleys are commonly deeply incised…” and that “mature paleosols commonly cover the areas between coastal valleys,” but warn that there are caveats and exceptions in both cases.

Overall, this literature comparison points to attributes likely to be useful for interpreting MSB origin, but we note that this result does not imply that non-unique characteristics could never be used for interpretations of process origin, or that sand bodies generated from multiple processes may not have a mix of characteristics. For example, specific attributes of story sedimentology may, in some cases, be clear indicators of one process or another. However, broadly speaking, studies aimed at interpreting avulsion or incised-valley processes from MSBs might best be served by focusing on MSB attributes that are likely to be uniquely diagnostic of those process interpretations. In addition, because these MSB-generating processes operate at different scales, they are not necessarily mutually exclusive. The useful attributes identified here— particularly the MSB bounding surface and correlation of floodplain horizons with the sand body—are generated by channel-belt-scale processes and are consequently most applicable to interpretations at that scale rather than, for example, reconstructing intra-channel-belt dynamics from MSB deposits.

Although these results indicate that MSB process origin can be uniquely interpreted from field evidence, some systems and datasets may not be amenable to making unique process interpretations. For example, studies that confidently interpreted avulsion origin were conducted in deposits with abundant preserved overbank deposits, and, in most cases, MSBs with well-exposed margins. This means that distinguishing avulsion-type MSBs in very sandy systems may be difficult. Confidently interpreting MSB origin from subsurface datasets may be similarly challenging, because, for example, most seismic data sets do not have high enough resolution to examine the shape of MSB bounding surfaces, and cores and well logs must be closely spaced and confidently correlated to provide insight about MSB bounding surfaces and floodplain correlations. Despite these interpretation difficulties in sand-dominated systems and in subsurface datasets, our results from this literature survey indicate that avulsion-reoccupation-generated MSBs can be uniquely interpreted in cases where the MSB margins are irregular and floodplain horizons correlate with story boundaries.

Avulsion-Pattern Controls On MSB Formation and Preservation

At basin-filling scales, basin aggradation and channel-belt avulsion should influence the abundance and character of MSBs. Early models of alluvial architecture explored the principal controls on channel stacking patterns in avulsion-dominated basins (Allen 1978; Leeder 1978; Bridge and Leeder 1979), focusing on the effects of sedimentation rate and avulsion on channel-body distribution and establishing the fundamental connection between low aggradation rates and the potential for high sand-body-stacking density and interconnectedness in alluvial basins. Later geometric models of alluvial architecture explored the connection between sedimentation rate and avulsion frequency (Heller and Paola 1996), demonstrating that significant changes in channel-body stacking can result when avulsion frequency varies with sedimentation rate. Recent work has shown the potential for different characteristic avulsion patterns in alluvial basins (e.g., Sheets et al. 2007; Straub et al. 2009; Hajek et al. 2010), suggesting that even with similar long-term aggradation rates, alluvial architecture may differ depending on whether channel-belt avulsions move evenly or unevenly across a basin. Avulsion-derived MSBs may be particularly useful for reconstructing paleo-avulsion dynamics, including helping to constrain the range of avulsion patterns observed in natural systems and improving prediction of reservoir heterogeneity in fluvial deposits.

Here we use a simple 2D geometric model to explore how different basin-filling conditions (including channel incision rates and floodplain aggradation rates) and avulsion patterns influence the formation and characteristics of avulsion-origin MSBs. We analyze the shape, size, and preserved overbank content of model-generated MSBs constructed under different avulsion-pattern and aggradation conditions. Philosophically, we draw from the spirit of early models of alluvial architecture and do not attempt to resolve the physical processes that cause avulsions. Consequently, these heuristic models do not predict exact sand-body patterns or detailed MSB characteristics, but instead are used to explore how end-member basin-filling conditions influence gross architectural patterns and MSB formation. Specific channel and floodplain processes and lithofacies are not included in this simplified geometric model.

We use this model to test the influence of three end-member avulsion patterns—random, compensational, and clustered—under different aggradation conditions. A random avulsion pattern occurs when positions of channel relocations can be described by a uniform-random distribution (that is, an avulsion is equally likely to localize any place across a fluvial plain; e.g., Leeder (1978)). A compensational avulsion pattern occurs when avulsion channels localize in topographically low areas on the floodplain, thereby “compensating,” or filling in, basin topography (Bridge and Leeder 1979; Straub et al. 2009). Avulsion-driven compensational basin filling in alluvial settings has been observed and quantified in reduced-scale stratigraphic experiments (Sheets et al. 2007; Straub et al. 2009; Wang et al. 2011). In contrast, a clustered avulsion pattern occurs when avulsion channels are likely to localize somewhere close to a formerly occupied site (e.g., Jerolmack and Paola 2007) until a maximum thickness threshold is met, after which the channel avulses compensationally (to the lowest point on the floodplain). This generates stratigraphy with closely spaced “clusters” of channel bodies, which have been qualitatively observed in well logs and outcrop (e.g., Hofmann et al. 2011; Hampson et al. 2012) and statistically documented in outcrop (Hajek et al. 2010).

Model Design and Parameters

To explore the potential effects of avulsion pattern and aggradation rate on avulsion MSBs, we use an object-based basin-filling model that simulates a 2D basin cross section (Fig. 3). At each time step, a rectangular channel-belt-scale object (“channel element”) is placed in the basin following one of several avulsion-pattern rules. In the random case, element location is drawn from a uniform-random distribution of all locations in the model domain that are below a maximum elevation threshold. We impose a maximum relief condition of three times the channel element thickness to prevent topographic relief from growing continually throughout each model run. In the compensational case, element location is set as the lowest elevation in the model domain. In clustered cases, channel-element location is randomly chosen within a limited distance away from the previous channel location until a regional topographic-relief threshold is attained; when relief between the top of an active channel element and its local floodplain (within three channel-element widths) exceeds the thickness of the channel element, the subsequent avulsion locates in the lowest spot in the model domain (i.e., is compensational). The clustered avulsion zones presented here are two (narrow case) and four (wide case) times the channel width on either side of the active channel element.

Channel-element and floodplain-aggradation rules are set independently and are constant throughout each run. At each timestep a channel element is placed in the model domain at a location determined by aforementioned avulsion-pattern rules. The channel element incises a set proportion of its thickness (0–100%) into the existing stratigraphy (Fig. 3). At every timestep the floodplain aggrades at a set proportion of the channel-element thickness (0–50%) at all locations outside the active channel element, although floodplain aggradation is not permitted to exceed the elevation of top of the active channel element anywhere in the model domain. The ratio of channel incision to floodplain aggradation controls total aggradation for model runs, or mean model-aggradation rate, which is reported as a fraction of the channel-element thickness per timestep. Figure 4 shows example stratigraphy built in model runs with varying channel-element incision and floodplain aggradation for all avulsion patterns.

MSB properties are calculated from the output of each model run using the image processing toolbox in Matlab. We define model MSBs as any contiguous channel elements, no matter the orientation or arrangement (i.e., if two or more elements touch, they define an MSB; e.g., Fig. 3). MSB width-to-thickness ratio is defined by the maximum cross-basin width and stratigraphic thickness of the MSB area. When remnant pockets of floodplain deposits are fully enclosed by channel elements, the MSB is marked as “floodplain-enclosing” (Fig. 3). Story preservation is the fraction of the original channel element preserved at the end of the model run (Fig. 3).

In the model runs for this study, channel-element dimensions are 10 model units wide by 2 model units thick, yielding a channel element W:T ratio of 5. The active model domain is 55 times the channel element width (550 model units), plus a buffer zone of one channel width on both sides of the model domain. Each run is 500 iterations. Results presented here reflect 624 model runs: 156 per avulsion type, spanning incision rates from 0 to 90% of the channel-element thickness at 10% increments, and floodplain aggradation rates from 2.5 to 50% of the channel-element thickness at 2.5% increments.

Model Results

Here we compare MSBs generated under different avulsion patterns at a range of channel-incision and floodplain-aggradation rates. When basin filling is accomplished exclusively through channel-element aggradation (relative channel incision ranging from 10 to 100% and floodplain aggradation of 0%), model outputs constitute a single MSB. Mean channel-element preservation increases with decreasing channel incision and does not significantly vary with avulsion pattern (Fig. 5). When floodplain deposition is included in model runs, mean basin-aggradation rate is a function of the balance of floodplain aggradation and channel incision. Generally, when model aggradation is low (i.e., low floodplain aggradation and high channel incision), model stratigraphy is channel-dominated, with clustered avulsion patterns generally resulting in stratigraphy with lower preserved channel fraction than compensation or random avulsion patterns (Fig. 6).

Model runs with mean aggradation rates below 10% generate stratigraphy with a small number of MSBs, each containing many stories and extending across the model domain (Fig. 4A). At these low model aggradation rates, all avulsion patterns yield MSBs with a similar number of stories per sand body (Fig. 7), similar width-to-thickness ratios (Fig. 8), and MSBs that commonly fully enclose floodplain deposits (Fig. 9). Consequently, MSBs produced by different avulsion patterns are indistinguishable from one another at low mean-aggradation rates.

When mean model-aggradation rates exceed 10%, avulsion pattern influences MSB properties. For example, the mean number of stories per sand body is higher for clustered avulsion patterns than for random and compensational patterns, and narrow-zone clustered runs have more stories per sand body than the wide-zone clustered runs (Fig. 7). Likewise, clustered avulsion patterns are much more likely to generate stratigraphy with a high proportion of MSBs; Figure 10 shows that the proportion of all sand bodies that are multistory (as opposed to single story) decreases with increasing aggradation rate, but stabilizes around 70% for narrow-zone clustered avulsion patterns and 40% for wide-zone clustered patterns. Random and compensational avulsion patterns generate MSBs that are generally one or two stories (Fig. 7), but clustered avulsion patterns generate MSBs that have over four stories much more frequently (only 0–18% of MSBs in random runs and 0–10% of MSBs in compensational runs, compared to 15–44% and 24–74% of MSBs in wide-zone and narrow-zone clustered runs, respectively; Fig. 7). Overall, this suggests that MSBs with many stories and stratigraphy with a high proportion of MSBs may be indicative of clustered avulsion patterns.

Avulsion pattern also influences the percent of floodplain-enclosing MSBs (Fig. 9) and mean-story preservation (Fig. 11). Only clustered-avulsion patterns produce MSBs that completely enclose floodplain deposits, with 10–15% of all MSBs generated in narrow-zone clustered runs containing preserved floodplain deposits. In contrast, fewer than 2% of MSBs generated in random and compensational avulsions runs enclose remnant floodplain deposits (Fig. 9). Additionally, clustered-avulsion patterns generate MSBs with the lowest mean story preservation (Fig. 11), in part because clustering results in a tendency to reoccupy and “rework” previously deposited channel elements, generating stratigraphy with lower overall channel fractions (Fig. 6). Compensational- and random-avulsion patterns have similar story preservation at all mean aggradation rates (Fig. 11) (and therefore similar overall channel∶floodplain ratios; Fig. 6).

The influence of avulsion pattern on MSB width-to-thickness ratio varies with model aggradation rate (Fig. 8). Random and wide-zone clustered avulsion patterns generate MSBs with the lowest W∶T ratios, which do not change significantly with increasing aggradation rate. In contrast, narrow-zone clustered avulsions and compensational avulsion patterns generate MSBs with the highest W∶T ratios, which decrease with increasing model aggradation rate. Because the mean W∶T of wide-zone clustered MSBs is similar to random, and narrow-zone clustered W∶T is similar to compensational, this metric is not useful for broadly distinguishing clustered-avulsion patterns.

Implications of Model Results

Model results suggest that in systems with avulsion-origin MSBs, relatively high amounts of fine-grained lithofacies are necessary to reliably interpret paleo-avulsion pattern from MSB properties, because at low mean aggradation rates, all avulsion patterns generate MSBs with similar characteristics. Once overbank deposits comprise at least 60% of model stratigraphy (at mean model aggradation rates of 10% and higher), only clustered avulsion patterns regularly produce MSBs with more than four stories, and only clustered avulsion patterns result in MSBs that commonly contain preserved floodplain deposits. This suggests that these features may be useful for differentiating fluvial systems characterized by clustered paleo-avulsion patterns from those with random or compensational paleo-avulsion patterns in natural stratigraphy. At all aggradation rates, random and compensational avulsions generate MSBs with the same properties, suggesting that these avulsion patterns may not be differentiable based on MSB properties alone.

Model results indicate that in sand-dominated deposits, avulsion pattern cannot successfully be interpreted from MSB characteristics. However, even in these successions, story preservation, which is controlled primarily by channel incision, could be used to approximate paleo-channel incision rates (Fig. 5). Additionally, although exceptionally well-preserved bar forms have been hypothesized to be an indicator of compensational avulsions (Hajek and Heller 2012), these model runs suggest that compensational avulsion patterns do not generate MSBs with noticeably higher story preservation than random avulsion patterns.

Finally, although W∶T is used to categorize channel-body type in the fluvial literature (Gibling 2006), these results suggest that it is not a useful measure of avulsion pattern, even in mud-dominated model runs. This implies that for interpretations and exploration of paleo-avulsion patterns in systems with avulsion-origin MSBs, constraining sand-body dimensions is not an essential first step, and that constraining MSB W∶T is not a useful predictive or interpretive tool. Rather, the number of stories per MSB, the preservation of floodplain facies inside MSBs, and the proportion of all sand bodies that are multistory might be useful for identifying ancient systems characterized by clustered paleo-avulsion patterns when avulsion is the dominant control on MSB formation.

Case Study: The Williams Fork Formation

Identifying MSBs produced by avulsion reoccupation in natural deposits is a critical step for testing hypotheses about avulsion behavior and stratigraphic organization (e.g., Jerolmack and Paola 2007; Hajek et al. 2010). Here we apply insights from our literature review and modeling to an Upper Cretaceous fluvial succession as an example of how MSB process-origin interpretations can be made in outcrop exposures. The results from our literature review suggest that avulsion-type MSBs can be identified in mud-rich deposits where sand-body margins are exposed, and our model results suggest that some inferences about paleo-avulsion pattern can be made from avulsion-type MSB properties, including the number of stories per MSB and the presence of intra-MSB floodplain deposits. To demonstrate a field application of these findings, and to explore sources of interpretation uncertainty, we analyze MSBs of the fluvial Williams Fork Formation in western Colorado.

Study Area

The Williams Fork Formation is an Upper Cretaceous (mid-Campanian through early Maastrichtian) alluvial formation in the Piceance Basin of western Colorado (Fig. 12; Cole and Cumella (2003). It was deposited in eastward-flowing lowland rivers that were sourced from the Sevier thrust front (Cole and Cumella 2003). The Williams Fork ranges from 300 to 600 m thick, and comprises two unofficial members; the mudstone-dominated lower member is 30–60% sandstone and the sandstone-dominated upper member is 50–80% sandstone (Cole and Cumella 2003; Pranter et al. 2008). Because the Williams Fork Formation is coal- and natural gas-bearing, it has been the subject of several paleogeographic and reservoir studies (Cole and Cumella 2003; Pranter et al. 2009; Hofmann et al. 2011; Pranter and Sommer 2011). These studies have identified both single and multistory channel bodies in the lower Williams Fork, and generally interpret the sandstones as having formed in meandering channels in a coastal-plain setting.

In this study, we focus on the mudstone-dominated lower member, which contains fluvial sand bodies separated by meters of relatively homogeneous gray floodplain deposits, including intermittent crevasse-splay sandstones. The sand bodies measured in this study are exposed in Coal Canyon, a five-kilometer-long canyon along the Colorado River, about five miles east of Palisade, Colorado (Fig. 12). This exposure is primarily oriented perpendicular to the eastward paleoflow direction (Pranter et al. 2009) and exposes the entire thickness of the lower Williams Fork, from the basal contact with the Iles Formation (Cretaceous shoreface deposits), to the upper contact with the sandy upper Williams Fork Formation.


The architecture of 34 lower Williams Fork Formation MSBs was evaluated in the field and using terrestrial lidar scans (3D surface-topography and reflectance-intensity scans collected with a Riegl VZ1000). Approximately 40% of the sand bodies in the measured interval are multistory and 60% are single story. For every MSB, the maximum sand-body thickness, total number of stories, maximum story thickness, and story grain sizes were measured, and the bounding-surface margin was characterized as smooth or sawtooth. For sand bodies that could not be safely reached in the field, these properties (except story grain size) were measured on the lidar point cloud (vertical accuracy to within 10 cm). Following the results of the literature review, MSB origin was interpreted based primarily on (1) the geometry of the sand-body margin and (2) the correlation of paleosol horizons between a MSB and its surrounding fine-grained (floodplain) facies. The process origin of each sand body was interpreted as either intra-channel-belt, avulsion reoccupation, or incised-valley, when possible; cases where MSB margin and surrounding floodplain horizons were insufficiently exposed to confidently make a process interpretation, MSB origin was classified as indeterminate.

To assess the confidence of these interpretations, we applied a qualitative uncertainty ranking to each MSB. “A” quality MSBs had good exposure of MSB margin on both sides of the sand body and/or distinct, mappable floodplain horizons. “B” quality MSBs had one well-exposed margin, but parts of the sand body may have been covered or inaccessible. A few correlatable floodplain horizons may be visible, but possibly only exposed on one side of the sand body. MSBs with a “C” quality had poorly exposed margins and/or difficult-to-correlate floodplain horizons, but a geologist can use other MSB properties, such as story scale or channel abandonment preservation, to make a tentative interpretation. Cases when sand-body margins were not exposed, floodplain horizons could not be correlated, and the scale of story features was difficult to determine, the MSB origin received a “D” ranking, and the process origin was indeterminate.

Results and Interpretation

Lower Williams Fork MSBs generally display consistent sedimentology and architecture throughout the measured interval (Table 2). Channel grain size ranges from upper-very-fine to upper-fine sand, and total MSB thickness ranges from 2.8 to 13 meters, with the 25th and 75th percentiles at 5.0 and 9.5 meters, respectively. Nearly half (48%) of MSBs had at least one well-exposed sand-body margin and were classified as “B” quality, although only three MSBs (9%) had both margins exposed (“A” quality). Overall, MSBs had between two and four stories, which ranged in thickness from 0.5 to 8 m, with a mean thickness of 2.9 m. Generally sand bodies had similar overall sedimentology and architecture, which suggests that depositional conditions were fairly consistent throughout the measured interval in the lower Williams Fork.

Of the 34 observed MSBs, 53% were avulsion origin, 32% were intra-channel-belt origin, 0% were incised-valley origin, and 15% were indeterminate. Avulsion-origin MSBs had at least one exposed irregular margin (generally of a sawtooth form) (Fig. 13C) and two to four stories, both of which are properties that are consistent with the literature review results. In the outcrop face shown in Figure 13A, avulsion-origin MSBs make up 25% of the total observed sand bodies (including single story). The relatively low number of stories per MSB, and the small number of avulsion-origin MSBs relative to the total number of sand bodies, suggests that the paleo-avulsion pattern was not clustered in the same sense as the model. A high percent (78%) of avulsion-origin MSBs had an A or B confidence ranking (Table 2).

In addition to avulsion-origin MSBs, 32% of MSBs were interpreted to have resulted from intra-channel-belt processes (Fig. 13B). These intra-channel-belt MSBs had at least one smooth margin and two or three stories (Table 2). No intra-channel-belt MSBs had an A confidence ranking, 54% had a B-grade confidence ranking, and 46% had a C ranking. These confidence levels are lower than the avulsion-origin interpretations, possibly because “smooth” margins are harder to confidently identify because of uncertainty associated with incomplete exposure.

Only 15% of lower Williams Fork MSBs were of indeterminate origin (receiving a D ranking), largely because the lower Williams Fork sand bodies are generally well exposed. These indeterminate MSBs had no exposed margins, and contained two to four stories. Most of these indeterminate MSBs were exposed only in gulches, and sand-body margins were scree-covered along both sides of the gulch.

Overall, these results suggest that lower Williams Fork MSBs were formed primarily through avulsion reoccupation and secondarily via intra-channel-belt processes. Previous papers about the Williams Fork identify intra-channel-belt origins of some MSBs (Cole and Cumella 2003; Pranter et al. 2009), but applying our suggested framework for interpreting MSB origins reveals the importance of avulsion reoccupation in Williams Fork sand-body architecture for the first time. The low number of stories per sand body and the relatively low proportion of MSBs at this locality suggest that Williams Fork rivers avulsed with random or compensational avulsion patterns. This contrasts with results from Hofmann et al. (2011), who found that in a different part of the Piceance Basin (in the northern region) Williams Fork channel deposits are clustered in the subsurface, which suggests that avulsion patterns may vary regionally in contemporaneous deposits. These results can be further tested by measuring basin-filling patterns using statistical methods (e.g., Hajek et al. 2010; Wang et al. 2011).


In the literature-review, modeling, and field-application portions of this study, it is clear that the presence of fine-grained lithofacies is necessary to identify avulsion-reoccupation-origin MSBs with high confidence, and to interpret paleo-avulsion patterns from MSB characteristics. The two high-confidence literature-derived criteria for avulsion origin, an irregular MSB bounding surface and correlative floodplain horizons, necessitate preserved overbank deposits for field identification. In addition, our model results suggest that a clustered avulsion pattern can be definitively interpreted from avulsion-derived MSBs only in successions with appreciable mudstone accumulations, because in channel-dominated deposits, all avulsion patterns generate MSBs with similar properties. Fine-grained lithofacies with differentiable horizons, such as distinctive paleosol layers, would improve the possibility of correlating floodplain horizons with individual stories, and thus confidently interpreting avulsion origin.

The results presented here may help improve prediction of subsurface reservoir quality and compartmentalization. Sandstones extending laterally into surrounding floodplain deposits—common in avulsion-origin MSBs with stepped or sawtooth margins (Figs. 1, 11)—may act as thief zones, potentially reducing sweep efficiency relative to MSB reservoirs with smooth margins. Avulsion-origin MSBs formed by clustered paleo-avulsion patterns may be particularly prone to having internal baffles and barriers to fluid flow given the tendency for remnant floodplain deposits to be preserved under these conditions (Fig. 9). Additionally, given that aggregate story-preservation statistics may be useful for reconstructing paleo-incision ratios in avulsion-dominated successions (e.g., Fig. 5), the type and abundance of fine-grained lithofacies preserved in MSBs (e.g., Lynds and Hajek 2006) may be predictable.

Although this study confirms that avulsion-type MSBs can be confidently interpreted in the field and may have predictable properties, questions remain about MSB process origins and characteristics. More detailed process models are required to address outstanding questions about MSBs generated from other process origins, especially intra-channel-belt processes. For the purposes of this study, the catch-all intra-channel-belt category combined many distinct fluvial processes into one category; morphodynamic models could elucidate how some of these different processes (for example, bar migration versus channel-thread switching) influence MSB formation and characteristics.


Understanding avulsion controls on MSBs is important for interpreting paleo-avulsion histories and for using alluvial stratigraphy to study long time-scale avulsion processes. Combined, the literature analysis, object-based modeling, and field case study presented here show that avulsion-origin MSBs can be confidently interpreted in alluvial stratigraphy, especially in outcrops with abundant fine-grained lithofacies and well-exposed sand-body margins. Analysis of published MSB interpretations shows that irregular (stepped or sawtooth) bounding surfaces and floodplain horizons that correlate to individual stories are characteristics uniquely associated with avulsion-origin MSBs. Additionally, our model results show that MSBs with many stories and preserved remnant floodplain deposits are diagnostic of clustered paleo-avulsion patterns. Overall, these findings suggest that alluvial successions may preserve valuable information about paleo-avulsion processes and also that subsurface reservoir characteristics may be predicted with greater accuracy by carefully considering the role of avulsion processes in MSB formation.

Supplemental Material

Two supplemental files are available from JSR's Data Archive: http://sepm.org/pages.aspx?pageid=229.

This work was supported by National Science Foundation-EAR #1024710. Partial funding was also provided by the American Association of Petroleum Geologists Foundation and the Geological Society of America Foundation. We thank Sheila Trampush, Foster Chamberlin, and Rebecca Beichner for help in the field. We are grateful to G. Weissmann and an anonymous reviewer for thoughtful reviews that improved the clarity of the manuscript.