1 INTRODUCTION

Bounding surfaces separate different deposited sediments and sedimentary rocks, defined in terms of grain size, sedimentary textures (sorting and rounding), sedimentary structures and geometry at different times, representing different physical, chemical and biological processes. Such surfaces record and reveal the orientation of accretionary and sedimentary growth. It is well known that many depositional units accumulate during short intervals of rapid sedimentation separated by long intervals of time when little or no sediment is deposited (Miall, 2014). Sedimentary depositional packages are bounded by different hierarchies of boundaries that often represent these periods of non-deposition. A hierarchical system for fluvial bounding surfaces and depositional units was first proposed by Allen (1983) and developed by Miall and Bridge (Bridge and Mackey, 1993; Miall, 1996) and has been widely applied in these settings. There are also bounding surface systems for eolian, as well as turbidite environments (Brookfield, 1977; Kocurek, 1981; Kocurek, 1988; Miall, 1989; Prather et al., 2000), but hierarchical models have not been applied to coastal deposits. Two reasons may explain this. One is that many coastal workers have focused on coastal geomorphology and engineering rather than sedimentology (Woodroffe, 2002), where the main goal has been to study a range of water movements instead of depositional records. Another reason is that most Quaternary coastal barriers have mainly formed during the Holocene due to transgression since the last glacial maximum (Davis Jr, 1994). Bounding surface models developed for other sedimentary environments could also be modified and adapted for work in the coastal washover barrier environment.

The barrier deposits we target here are located on the north shore of Huangqihai Lake, North China (Fig. 1). Recently, the lake level evolution of Huangqihai Lake was determined using quartz optically stimulated luminescence (OSL) dating and sedimentologic and stratigraphic analysis of profiles constructed from trenching that runs from the east to west shores of the lake (Chen et al., 2008; Zhang et al., 2011, 2012). The relationship between lake-level fluctuations and this barrier evolution in the area was further explored using radar stratigraphy and sedimentary architecture analysis (Shan et al., 2015). This barrier is formed by washover and swash processes based on analysis of depositional record and GPR facies characteristics. These observations, together with a new understanding of the bounding surfaces that define this sediment body allow us to investigate its boundary hierarchy.

In this paper, Ground Penetrating Radar (GPR) and trenching data are reported from this barrier spit. These data reveal the record of bounding surfaces, and thus the sedimentary processes which allow these surfaces to form. A lacustrine coastal boundary hierarchy model is proposed in this study, based on sedimentologic observations from trenches and radar stratigraphy analysis of GPR data. By developing an appropriate hierarchical model we will be better equipped to interpret barrier deposits and understand what processes are responsible for their formation.

2 STUDY AREA

Huangqihai Lake (40°47′–40°55′N and 113°10′– 113°23′E) is a hydrologically closed lacustrine basin with a maximum length of 20 km and a maximum width of 9 km, and an area of 110 km² (Fig. 1a). The lake lies 1 262 m above sea-level having a drainage area of more than 4 500 km². The regional geomorphic landscape is featured by basaltic ranges of mountains and alluvial plain. The coastal sediments are mainly supplied by rivers such as Bawang River, Quyulin River and Dahewan River (Fig. 1b). This lake is in a semiarid temperate climate which is cold and dry in winter and warm and humid in summer. The mean annual temperature in the area is about 4.5℃ and the mean annual precipitation and evaporation are 362 mm and 1 927 mm respectively (Zhang et al., 2011, 2012). The coastal barrier system formed in an area where the gradient of the depositional coastlines is less than 2°. The lake level drops dramatically due to the combined effect of the semi-arid climate and reduced fresh-water input.

The barrier spit is about 1 km long and 150–175 m wide (Fig. 1c). The maximum thickness of the whole spit unit is more than 1.6 m (Fig. 2a). The barrier deposits are mainly sub-rounded to sub-angular granules and medium-coarse grained sands. The mineralogy of the barrier sediments is quartz and feldspar, derived ultimately from the erosion of granite, typical of the lithology of the surrounding mountains. The top of it has been altered by eolian processes as well as bioturbation by shrubs and grasses. This barrier spit was mainly formed by swash and washover processes, although it is above the lake level (Fig. 1b).

3 METHOD

Bounding surfaces of the barrier system on the north shore of Huangqihai Lake were identified in trenches in key locations, and by Ground Penetrating Radar (GPR) profiles. Trenches were chosen to show higher-order bounding surface features which are more microscopic. The GPR profiles show lowerorder bounding surface characteristics because GPR has lower resolution than trenches, but a greater macroscopic view. The correspondence between surfaces in trenches and radar stratigraphy in GPR profiles in this barrier has been verified by Shan et al. (2015). Topographic surveys were carried out and over 3 km of GPR profiles were collected. Washover processes are assumed mostly to be in shore-normal directions. The GPR lines were thus arranged normal and perpendicular to this direction across the barrier because in these directions the bounding surface characteristics are more evident.

3.1 Ground-penetrating radar

A GPR unit consists of a transmitter and a receiver, which moves over the ground surface. GPR utilizes propagating electromagnetic waves that responds to changes in the acoustic properties to detect shallow subsurface geology. The reflected acoustic wave was detected by a receiving antenna, and recorded. Water saturation, water type, mineralogical changes and porosity determine dielectric properties. In sedimentology, GPR is used primarily for providing high-resolution profiles to aid in determining sedimentary architecture, geometry and correlation.

The GPR system employed in this work was the TerraSIRch SIR 3000. A high-frequency 400 MHz antenna was used to provide higher resolution imagery to depths about 2.5 m, because the maximum depth of the barrier spit is~1.6 m which is less than that. Over 3 km of GPR profiles were collected (Fig. 1). Three radar profiles (~0.5 km) were selected in the work described in this paper to illustrate the bounding surfaces of this spit. Reflexw Software was used for editing and processing the GPR data (Sandmeier, 2014). Signal ringing, i.e. horizontal coherent noise (Kim et al., 2007), occurred in the GPR profiles because of thriving shrubs and grasses, caused by high ionbearing liquid concentrations in the roots (Lindhorst et al., 2008). All GPR data were post processed including frequency filtering, trace-to-trace stacking, constantvelocity migration and gain adjustment. The GPR wave velocity in spit sediments was determined by correlation of sediments in trenches to GPR data and via hyperbolic velocity analysis. The velocities for saturated sands and dry sands are~0.06 m/ns and~0.12 m/ns, respectively. A mean velocity of 0.09 m/ns was used for time-depth conversion. Topographic correction was made of GPR data along survey lines, collected using a global-positioning system (Trimble GEO XT 2008) with accuracy better than 1 m. Trace increment was set to 0.05 m. In this study, the GPR imaged this spit to a depth of 25–35 ns corresponding to a depth of 1.6–2 m.

Radar surfaces were determined based on principles of radar stratigraphy proposed by Bristow (2009) and Bristow et al. (2005). We have improved this methodology to determine the order of boundaries, based on identification of termination patterns. The relative chronology of successive radar sequences can be deduced by the laws of superposition and crosscutting relationships (Neal et al., 2003).

3.2 Trenching

We used 17 trenches (1.5 m long, 1 m wide and 1–2 m deep) to identify and determine the lower-order bounding surfaces of this barrier. In this study, four trenches were selected to show the lithology, sedimentary texture (i.e., grain size, sorting and rounding) and sedimentary structures. The locations of trenches T6, T9, and T10 were chosen using GPR survey lines to show internal structure of this spit and sedimentary features both above and below the bounding surfaces. The trench T17 was used to indicate the relationship between boundaries and grain size features (sorting, skewness, and mean grain size). The grain size analysis was conducted in China University of Geosciences Beijing by using a grain size analyzer Mastersizer 2000 and software GRADISTAT written by Blott and Pye (2001).

4 RESULT

The barrier spit in Huangqihai Lake is composed of building blocks formed during a series of events or by different sedimentary processes separated by hierarchically ordered boundaries. Boundaries within washover barrier deposits can be classified into a hierarchical order. We found that it is practical to develop a surface classification to a four-fold hierarchy, in order to consider wave current flow regime (3^rd and 2^nd order surfaces), barrier spit evolution (1^st order surfaces) and sea-level or lakelevel change (super bounding surfaces) simultaneously.

The mean grain size of washover units from trench T17 spans from 284.5 μm (1.813 phi) to 461.7 μm (1.115 phi). The sorting of these washover units is from 0.822 (moderately sorted) to 1.793 (poorly sorted). The grain size distribution of these sediments shows negative skewness. The washover lope deposits shows an overall coarsening upward trend and a decreasing sorting upward trend from 0.822 (moderately sorted) to 1.753 (poorly sorted) in this trench, with a mean grain size ranging from 461.7 μm to 454.1 μm (Fig. 3).

4.1 Super bounding surface 4.1.1 Description

In GPR profiles, above the super bounding surface, the overlying reflections are typically concordant or down-lapping on the underlying low amplitude horizontal or reflection free strata which are typical of lagoonal salt marsh peats (Fig. 2d, e, f).

In trenches, the super bounding surface is the surface which separates washover barrier sands and lagoon peats deposits (Fig. 1a, c). Both mean grain size and sorting show dramatic change across this super boundary (Fig. 3). A mean grain size of a salt marsh peat, 19.81 μm and 5.658 phi, is much smaller than an overlying washover unit (461.7 μm and 1.115 phi). The grain size distribution feature of this salt marsh deposit is polymodal and very poorly sorted, due to the large span of the grain size (1.126 μm to 227.1 μm). The sorting for the overlying washover unit is 0.822 (moderately sorted).

4.1.2 Interpretation

The super bounding surface is the most significant and important surface because it covers the largest area and marks the longest time interval among all the hierarchical surfaces in the coastal barrier. Barriers are typically 0.5–5 km wide and 1–100 km long (Bird, 2000), so super bounding surfaces have a greater length and width because the barrier will migrate as the sea-level and sediment supply changes. It represents the cessation of peat deposition in a lowenergy back barrier setting and the onset of washover laminated sands deposited in a high-energy setting. The underlying non-washover fine-grained sediments were eroded by the overlying washover sands (Sedgwick and Davis Jr, 2003). The prerequisite for occurrence of washover is that the wave run-up must exceed the height of the pre-dune induced by storm surges. A significant change across super bounding surface in mean grain size and sorting indicates a huge transportation and sedimentation mechanism change. The onset of washover deposition often represents a climate change to a wetter condition (storm season) due to intensified monsoon (Costas et al., 2006; Garrison Jr et al., 2010). As a result this surface which separates grey to black back-barrier saltmarsh and overlying washover laminated sands (Fig. 2a, b, c) often indicates a sea-level or lake-level rise caused by a wet climate driven by the strong monsoon. The super surfaces originated and evolved in response to external factors including climatic and sea-level change.

4.2 1^st order bounding surface 4.2.1 Description

In shore-normal GPR profiles A-1 and A-2, the washover lobe deposits are characterized by landwarddipping reflections (Fig. 2d, e). The 1^st order bounding surface is often downlapped by overlying landwarddipping reflections (Fig. 2d, e).

The grain size measurement result shows a small change in terms of the mean grain size, sorting, and skewness. Most commonly, one single unit often consists of cross-bedded sands and gravels at the base and a mud cap at the top (Figs. 2a, b and 4). In the proximal part of the washover deposits, the channel throat of the washover lobes may produce normal graded sands at the base of the unit, whereas in the distal part, the lobe may characterized by tabular cross-bedded sands and undifferentiated sands (Leatherman and Williams, 1977; Sedgwick and Davis Jr, 2003).

4.2.2 Interpretation

The 1^st order bounding surface often separates two washover units which are formed by two different storm events. One washover unit, bounded by two 1^st order surfaces, always shows an overall finingupward trend due to the waning wave current energy (Fig. 2a, b, c) covering periods of tens of hours, although the interval between two storms can be more than tens of years (Davis Jr and Fitzgerald, 2009).

No evident change across 1^st order in terms of mean grain size and sorting indicates this washover lobe and overlying unit were formed by similar sedimentary processes. The washover current can flow landward at speeds in excess of 2 m/s decelerating with distance landward (Holland et al., 1991). As a result along 1^st order surfaces from proximal to distal, the overlying lithofacies changes from laminated sands to undifferentiated sand. In proximal parts of washover barriers, dune sets deeply truncate into previously underlying deposits (Blakey and Middleton, 1983), and the shape of the surface tends to be “channelform” in a parallel direction (Fig. 2f). However, in the distal part, the basal surfaces of each washover fanshaped deposit tend to be flattened. Below the washover deposits, in the shore-parallel direction, this surface extends landward from tens to hundreds of meters and several to tens of meters wide (Fisher et al., 1974; Leatherman, 1979; Dolan and Hayden, 1981).

4.3 2^nd order bounding surface 4.3.1 Description

Within one single phase washover deposit, an overall fining-upward trend from cross-laminated/ graded sands to muds is always found. 2^nd order surfaces often separate these two different lithofacies (the lower cross-laminated/normal graded sands and overlying fine-grained clays and silts) representing changes of flow conditions.

4.3.2 Interpretation

The fining upward trend within a lobe indicates a waning current energy, which also represents the end of a storm surge.

However, 2^nd order surfaces are not imaged in GPR profiles because the thickness of overlying finegrained deposits is too thin. We interpret the surface, between two different lithofacies, as a 2^nd order surface because the current regime changes across this surface from washover processes to suspension.

4.4 3^rd order bounding surface 4.4.1 Description

3^rd order surfaces are defined as cross-bed set bounding surfaces. These surfaces may truncate underlying cross-bedding at a higher angle (Fig. 2a, b, c), because occurrence of these surfaces indicates higher energy of washover currents, which tend to create steep slopes.

4.4.2 Interpretation

3^rd order surfaces indicate continuous sedimentation of a similar bed type, minor erosion still exists because of an abrupt change in flow direction or flow velocity. These 3^rd order surfaces are actually reactivation surfaces, which represent minor scouring and reorientation during sedimentation.

In each sedimentary unit bounded by 1^st order surfaces, the unit may contain many 3^rd order surfaces, the frequency of which reflects the discontinuity in the lee face sedimentation.

5 DISCUSSION

The sandstone bodies can be separated by a hierarchically ordered set of bedding contacts and it has been exploited sedimentologically for many years (Allen, 1983). However, the washover boundary hierarchy idea and model has not been reported. The ideas and the model presented here are derived from observation of trenches and interpretation of GPR profiles (Fig. 4). We suggest that this hierarchy model can be applied to other washover deposits. The architecture records in the GPR and trenches shows the growth of this spit and washover features. Washover bounding surfaces similar to those described in this study have been described previously (Schwartz, 1982; Neal et al., 2003; Switzer et al., 2006; Switzer and Jones, 2008; Nebel et al., 2011; Nutz et al., 2015; Shan et al., 2015), although in those researches, no hierarchy model was proposed.

The orientation of a washover lobe is sometimes oblique to the paleo-shoreline as observed in trench T6 (Figs. 2a and 4) based on the evidence that two beds dip in opposite directions, which indicates two directions of landward moving sediments in the two lobes. Washover lobes are recognized based on identification of 3^rd order high-angle landward dipping bounding surfaces (Fig. 4). Such lobes develop when the high-wave energy overwash moves landward into a significant body of standing water, i.e. a lagoon or a flooded marsh (Neal et al., 2003). The elevation of the water level is a prerequisite for washover processes to occur. When run-up fails to climb the crest, it results in swash processes and associated swash laminated sands at the stoss side of the barrier (Shan et al., 2015). The boundary underlying the swash laminated sands is also a 1^st order surface (Fig. 4), because it represents another beginning of a storm deposition. The record of 2^nd order surfaces is uncommon, due to the fact that washover deposits are often eroded by subsequent storms as identified in Fig. 2c and suggested by Nutz, especially in forced regression conditions (Nutz et al., 2015). The 3^rd order surfaces described in this study are comparable to the 1^st order and 2^nd order surfaces defined by Miall (1996), representing sedimentation of similar bedform type. The actual washover occurrences are significantly related to local geomorphologic conditions other than storm scale and frequency. The storm frequency has a close relationship with the frequency of 1^st order bounding surfaces because the unit between the 1^st order surfaces is a single storm washover deposit.

The storm scale and frequency are not necessarily related to the order of the bounding surface. Sallenger Jr (2000) proposed a storm impact scale based on relations between current run-up and dune elevation, including swash regime, collision regime, washover regime and inundation regime. Although these different current regimes result in individual deposits, they are all deposits formed during a single storm. Thus the bounding surface enclosing these deposits belongs to the 1^st order.

Swash bars are landward migrating, shore parallel sediment bodies (Fitzgerald et al., 1984; Greenwood et al., 2004). They are also the common accretionary architectural elements of the barrier (Lindhorst et al., 2008). Swash-bars are bounded by erosional surfaces triggered by storms. This kind of basal erosional surface is also a 1^st order boundary because it represents the action of another storm.

6 CONCLUSION

The four-fold hierarchy bounding surface model proposed in this paper is based on interpretation of trenches and GPR profiles. This new model differs from previous hierarchy systems by showing GPR characteristics, grain size distribution features, genesis and shape in a lacustrine coastal barrier. Four orders of boundary are defined with each representing a unique suite of formation processes and scales of impact. This model can be used better to interpret washover barrier deposits in modern and ancient examples.

7 ACKNOWLEDGEMENT

Clift thanks the Charles T. McCord Chair in Petroleum Geology for support.