Origin and Age of Fluvioglacial Sediments on Staten Island NY and Implications for Meltwater Flow

1. Introduction

Staten Island lies strategically between New Jersey and Long Island, and is key to correlating geological formations between these locations. The southwestern portion of Staten Island is underlain by the Cretaceous Raritan Formation of the Atlantic Coastal Plain [1], which also crops out nearby in New Jersey [2], but lies deeper beneath Long Island [3]. The Pensauken Formation was deposited on the eroded surface of the Raritan Formation during the Pliocene as observed across New Jersey [4], as well as in a few locations on Staten Island [5], but not on Long Island [3]. Overlying this are deposits of outwash of Illinoian and pre-Illinoian age, capped by till, including the terminal moraine from the Wisconsinan glacial maximum [4,6].

Much of Staten Island is developed, leaving very few small exposures of the surface sediment, and these are usually only accessible for a maximum of a few years before construction takes place or thick vegetation regrows [7,8]. This research focuses on two small outcrops of sand and gravel exposed in a road cut and a construction site in Charleston, in the southwest of Staten Island (Figure 1). This and other temporary exposures in the area have been used by previous researchers to determine the age and origin of sediments covering the area to the north of the terminal moraine, beneath the surface ground moraine [6]. However, these studies have resulted in different interpretations, with some identifying sediments of the Pensauken Formation [5,9,10], while others identified pre-Illinoian fluvioglacial deposits [11,12].

The Pensauken Formation is a fluvial deposit from a Pliocene drainage system ancestral to the modern Hudson and Delaware rivers [4]. It consists of sands and gravels, deposited as valley fill, in a temperate climate. Gravels are predominantly composed of quartz and chert, with smaller amounts of granite, sandstone and mudstone [15,16]. Sands are mostly subarkosic, with some arkose and glauconitic sands [4,15,17]. Pollen samples yield an age of approximately 3.5 Ma [4], superseding earlier work which assigned ages as Pleistocene [18] or late Miocene [17]. The Pensauken Plain extended from the Long Island Sound to the Delmarva Peninsula, with the Pensauken River flowing towards the southwest (Figure 1), and the sediments reached a maximum thickness of up to 40 m [4]. There are no deposits identified as Pensauken Formation on Long Island, probably as a result of subsequent glacial erosion. However, it is possible that the lithologically similar Mannetto Gravel represents the northeastern extension of the Pensauken Formation [3,4]. Previous studies of the Pensauken Formation on Staten Island have identified outcrops in the Charleston area, close to the location of this study [5,9].

There is evidence for at least three glaciations in sediment deposits from New Jersey and New York [19]. The pre-Illinoian was the earliest, and represents the first expansion of the Laurentide ice sheet about 2.4 Ma [19,20]. The limits of this expansion have been mapped in New Jersey, but can only be inferred for Staten Island and Long Island (Figure 1). This pre-Illinoian event diverted the Pensauken drainage system, resulting in the Hudson River moving close to its present-day location [4]. The extent of the Illinoian glaciation has been mapped in New Jersey from moraine deposits (Flanders and Bergen tills), as well as fluvioglacial outwash (Lamington Formation) and glaciolacustrine sediments [19]. These deposits are thought to correspond to the Montauk Till, and fluvioglacial Jameco Gravel and Merrick Formation on Long Island [21]. They have all been dated as corresponding to marine isotope stage 6, between 191 and 130 ka [19,20]. The most recent glaciation was the late Wisconsinan expansion of the Laurentide ice sheet [20], which reached its maximum extent in New Jersey 25 ka and began a rapid retreat 24 ka [19]. New Jersey segments of the Wisconsinan terminal moraine include the Perth Amboy, Madison, Mountain Lake and Budd Lake moraines [22], while the Harbor Hill terminal moraine extends across Staten Island and western Long Island [3,12]. As the ice sheet retreated, a series of glacial lakes formed and drained, including glacial Lake Bayonne directly to the west of Staten Island along the Arthur Kill and Newark Bay [19,23]. This lake initially drained through the Richmond Valley spillway before moving south to the Perth Amboy spillway, both of which fed the outwash channels of the paleo-Hudson River, across the present-day Hudson Shelf [14,23] (Figure 1).

Because much of the Wisconsinan terminal moraine remains at the surface, the extent of the most recent glaciation is well known across the region (Figure 1). However, this most recent glacial event removed evidence for previous glacial cycles in the New York City area [19]. The extent of the pre-Illinoian and Illinoian ice sheets are inferred to a limited extent from moraine deposits in New Jersey, but are essentially unknown on Staten Island [4,13,19]. However, their maximum extent is thought to be further north than for the Wisconsinan on Staten Island (Figure 1), so the study location was unlikely to have been glaciated during these events and would have experienced a periglacial environment. Potential deposits of pre-Illinoian and Illinoian outwash have been identified on Staten Island [12].

The goal of this study is to use a detailed analysis of grain size distributions, sedimentary structures and provenance of clasts, in conjunction with an interpretation of past topographic surfaces, to resolve the previous disagreement as to whether the sediments are Pensauken or fluvioglacial in origin, and to interpret glacial meltwater flow across Staten Island. The sediments were determined to be fluvioglacial, based on their texture and composition, and probably of Illinoian age. However, this does not refute the identification of nearby outcrops as Pensauken. An interpretation of the 3-dimensional structure of the area suggests that a history of erosional surfaces and valley fill has resulted in the juxtaposition of sediments of different ages and origins.

2. Materials and Methods

The study area comprises two small outcrops of sediments in Charleston, in the southwest of Staten Island (Figure 2). Outcrop 1 was exposed in a road cut around 2004 [10], and was logged and sampled in May 2013, before it became fully revegetated and inaccessible [7]. Outcrop 2, located 30 m from Outcrop 1, was exposed during construction in July 2022 [8]. There is some lateral variation within the outcrops, so two sections 10 m apart (Sections A and B) were logged in Outcrop 1, and three sections 5 m apart (Sections C, D and E) were logged in Outcrop 2. Samples were collected from every bed at each logged location. Because the outcrops are small, almost vertical cliffs, they represent a single flat surface, so it was not possible to collect any data showing the 3-dimensional orientation of the beds or cross-laminations. Unlithified and poorly lithified sediments comprise the bulk of the outcrop, so sampling of each bed was achieved by scooping the sands and gravels into sample bags using a trowel. The single partially lithified bed in Outcrop 1 yielded easily removable soft pieces of mudstone.

All samples were dried at 50 °C for one week, then separated into identical subsamples using the coning and quartering technique [24] for samples from Sections A and B, and using a Gilson Universal Mini-Splitter for Sections C, D and E. One half of each original sample was dry sieved for 15 minutes using a Gilson tapping sieve shaker model SS-8R. Sieve sizes ranged from -5 Φ (32 mm) to 4 Φ (63 μm) in 1 Φ intervals. Sediment finer than 4 Φ was collected in the pan and comprises all silt and clay size particles (mud). Each size fraction was then weighed to determine the proportions of mud, sand and gravel. Examination of each size fraction under a binocular microscope confirmed that all grains were fully disaggregated.

Data for each sample were processed using G2Sd [25]. The grain size descriptions reported here (mean, sorting, skewness and kurtosis) for each of the bulk samples, follow the methods of statistical analysis described by Folk and Ward [26]. Specific grain size classifications follow Folk [27]. All clasts larger than -2 Φ were observed under a binocular microscope to determine their lithology. All data supporting this work are deposited in CUNY Academic Works at location to be provided during review.

3. Results

3.1. Correlation of Beds Between Sections

The initial field site, where sections A and B were logged, had been investigated several years earlier (2004), when the underlying contact with the Raritan Formation was still visible [10]. By the time of this investigation (2013), this contact had been covered by a concrete barrier, but was known to exist less than a meter from the base of the observed outcrop. Sections C, D and E are at a topographically higher location by a few meters, and include a cap of glacial till that had been eroded from sections A and B (Figure 3). Overall, the combined sections include all sediments recorded between the bedrock Cretaceous (Raritan Formation) and the most recent Wisconsinan glaciation, a thickness of approximately 3 m.

Most of the sediments are sands and sandy gravels, many of which display planar horizontal and cross laminations (Figure 3). Beds range from 6 to 40 cm in thickness. These sediments vary laterally, and beds can be correlated over short distances, but not across all five sections. However, there are also three distinct layers that can be correlated over larger distances. The lowest of these is a thin (5 cm) soft siltstone identified in sections A and B, which is stratigraphically lower than the base of sections C, D and E. The second is a thick gravel deposit (30–80 cm), that can be correlated across all five sections. Finally, there is the gravel deposit of the till, which caps sections C, D and E. These key beds have been used to divide the sands and gravels into groups to aid their description and interpretation; Group 1 below the siltstone bed, Group 2 between the siltstone and the thick gravel, and Group 3 between the thick gravel and the till (Figure 3).

3.2. Grain Size Analysis and Sediment Texture

Grain size analysis was completed for all samples of sand and gravel, excluding the partially lithified siltstone and the till. The till was excluded as boulders are too large to incorporate, so processing an accurately representative sample was not possible. One sandy gravel sample from Group 3 was also removed due to the sample size being too small to be representative. Sample texture classification follows the system developed by Folk [27] (Figure 4). Samples referred to as “sand” in Groups 1, 2 and 3 include both sand (<0.1% gravel) and slightly gravelly sand (0.01–5% gravel). Samples described as “sandy gravel” include both gravelly sand (5–30% gravel) and sandy gravel (30–80% gravel). The gravel bed referred to in figures as “gravel” also has a texture of sandy gravel, but with a higher gravel content than sandy gravels in Groups 1, 2 and 3. Mud content of all samples is low enough (1.4–7.6%) that none of the samples have a texture code that includes mud.

The cumulative frequency curves illustrate the differences among the groups of sediment. Sandy gravel sediments from all three Groups (Figure 3) show similar grain size distributions (Figure 5A-E). They have mean grain sizes of coarse sand to fine gravel, and are poorly to very poorly sorted. They are bimodal, with the modes being medium to coarse sand and pebbles. The sandy gravels are the most variable of the sediment classifications, but there is full overlap between Groups 1, 2 and 3. The sand samples from Groups 1 and 2 have similar grain size distributions to each other, but are distinct from the sands in Group 3 (Figure 5A). For Groups 1 and 2 (Figure 5C,D), the sands have mean grain sizes of medium to coarse sand, and are moderately to poorly sorted. Most are unimodal with a mode of medium sand. However, there is some variability, and a few have a second mode of granules. The three sands from Group 3 (Figure 5D) are almost identical to each other and unimodal, with both mode and mean grain size of medium sand. They are poorly sorted with more fine sand and silt than the sands from Groups 2 and 3, but with no gravel. The Gravel bed (Figure 5B) has a higher content of gravel clasts than the other sandy gravels and is much less variable. These sediments have a mean grain size of fine gravel, and are poorly to very poorly sorted. They are bimodal with modes of medium or coarse sand and pebbles.

Scatter plots of the grain size parameters assist with the interpretation of the environments in which the different sediments were deposited [26] (Figure 6). The sediment classifications described above clearly plot in distinct fields in the charts of standard deviation (which represents sorting) versus mean clast size (Figure 6A) and skewness versus mean clast size (Figure 6B). The sandy gravels of Groups 1, 2 and 3 show a linear correlation between mean grain size and standard deviation (sorting), as would be expected when higher fluid flow incorporates gravel in addition to sand. They follow the trendline for sorting in bimodal sediments, where the poorest sorting occurs at a mean grain size at the mid-point between the two modes (Figure 6A), typical of a fluvial environment [26]. Most are coarse to very coarse skewed, due to the presence of gravel, and follow the sinusoidal trend for fluvial sediments [26]. However, one sample from Group 2 is an outlier and is fine skewed (Figure 6B). The sands of Groups 1 and 2 cluster loosely in the medium to coarse sand size mode and have a lower standard deviation representing a higher degree of sorting (Figure 6A). Their skewness is also more variable, ranging from symmetrical to very fine skewed (Figure 6B). The sands of Group 3 cluster tightly on both plots (Figure 6). They have a finer mean grain size than the sands of Groups 2 and 3, but are more poorly sorted and very fine skewed due to the presence of fine sand and silt. As a result, they are more fine skewed than typical fluvial sediments, which would not normally have a positive skewness greater than the magnitude of negative skewness in samples from the same fluvial environment [26]. This suggests a different environment of deposition to the other sands and gravels. The Gravel bed samples form clusters on both charts, with one outlier (Figure 6). They represent the coarsest mean grain sizes and high standard deviation indicating poor to very poor sorting (Figure 6A). Skewness increases to become very fine skewed at the highest mean grain size, again not fully following the predicted trend for fluvial deposits [26] (Figure 6B), potentially suggesting a different environment of deposition.

3.3. Clast Composition

All pebble clasts were identified for each sample, and combined into representative summaries for Group 1, 2 and 3 sands and gravels, the gravel bed and till (Table 1). The most common clasts overall are red siltstone and sandstone, followed by other siltstone and sandstone and quartz. Minor constituents include chert, conglomerate, basalt, granite and gneiss, in descending order of prevalence. Gravels from Groups 1 and 2 contain more quartz than those of Group 3 or the gravel bed, while the till contains very few quartz pebbles. The highest concentrations of red sandstone and siltstone are seen in Group 3 sands and gravels and the till. Quartz pebbles are most likely to represent reworking of sediments from the Pensauken Formation, while red sandstone and siltstone represent a source area in the Newark Basin [13]. Hence, this indicates a changing source area for the sands and gravels over time.

3.4. Facies Analysis

The characteristics of the sediments allow each of the beds to be associated with one of six sedimentary facies (Table 2). These facies all relate to fluvioglacial or glacial environments. Specifically, the fluvial sands and gravels (FSG) and fluvial sand (FS1) are deposits of glacial outwash and are typical of a braided river environment [29,30]. The siltstone (S) is an overbank deposit. The sequence that includes FSG, FS1 and S is similar to a “Donjek Type” braided stream profile [31]. The gravel bed is an outwash flood gravel (OFG), representing a change from the braided river environment. It was probably deposited during a significant meltwater flood event, such as the breaching of an ice or moraine dam retaining a glacial lake as the ice retreated. The deposit is typical of a flow with high sediment content, which typically occurs if there is a long time interval between floods [32]. The OFG bed is overlain by fluvial sand (FS2) in Sections B, C and D, which is texturally different to FS1. It represents the dewatering of the underlying OFG sediments, which then remobilized and deposited sediment of a finer grain size, as it seen in present day glacial outwash floods [32]. The gravel capping the outcrops of Sections C, D and E represents a glacial till, from a more recent glaciation than the fluvioglacial sediments below [12,33].

4. Discussion and Concluding Remarks

The facies analysis of the sediments strongly suggests that these two outcrops represent a series of fluvioglacial outwash deposits between the Cretaceous Raritan formation below and the overlying Wisconsinan till, a total thickness of 3 m. They do not represent a temperate climate fluvial deposit like the Pensauken Formation. There is a complete absence of any plant material in the sediments collected, including pollen that could potentially be used for dating. The fluvioglacial interpretation is also supported by the lithologies of the pebble clasts present in the sandy gravel and gravel beds. These are dominated by sandstone and siltstone, with lesser amounts of quartz, chert, conglomerate and other rock types. This is similar to other fluvioglacial sediments described on Staten Island [11,13]. In contrast, the Pensauken Formation is predominantly composed of quartz, quartzite and chert, with lesser amounts of conglomerate, sandstone, siltstone and other rock types [5,17]. Silty sands in the Pensauken Formation contain up to 50% mud [5], whereas none of the sand beds in this study contained more than 8% mud (Figure 4). Sand and gravelly sand is often cemented in the Pensauken Formation [17], while similar beds in these outcrops were not lithified.

However, this interpretation needs to be reconciled with previous reports of outcrops of the Pensauken Formation nearby. Due to the developed nature of the area, none of these outcrops are currently accessible, so an assessment of the limited available data was undertaken. The two outcrop descriptions and sediment analysis by Bowman seem compatible with being assigned to the Pensauken Formation [5]. The third outcrop from a field guide did not include a lithological description [9].

The next objective was to assess where the Pensauken Formation could potentially be present on Staten Island. The geographic extent of the Pensauken Plain was combined with the estimated altitudes of the base (15 m) and surface (55 m) of the plain [4]. To achieve this, elevation data [35] was brought into a geographic information system to select locations between 15 m and 55 m (Figure 7). Surficial geology data [36] was used to identify the locations of till, ground moraines, and artificial fill. Due to the altitude of the Fresh Kills landfill site, the artificial fill designation was used to remove incorrectly identified Pensauken Formation locations. Based on this assessment, all previously reported Pensauken Formation exposures [5,9], the location of the pre-Illinoian sediments [11,12] and the present study location could potentially lie within the Pensuaken Formation (Figure 7). However, this initial assessment does not consider that the Pensuaken Formation was a valley fill, deposited on a topographic surface eroded into the underlying Cretaceous sedimentary rocks [4]. The true extent of the Pensauken Formation is therefore less that predicted by this calculation.

A more detailed assessment of reported contact altitudes was necessary to evaluate the past topographic surface of the Raritan Formation, and how this may have influenced the deposition of Pensauken sediments. The locations of the contacts between the Raritan Formation and overlying sediments [5,7,9, and this study], were identified on a present-day topographic profile (Figure 8). From this information, an estimate of the surface of the Raritan Formation could be interpolated. This illustrates that the outcrops of the Pensauken Formation (P1 and P3) represent topographically lower positions, where more sediment would have been deposited, and these deeper sediments would also have been protected from later erosion. The fluvioglacial sediments were then deposited on this new topographic surface, directly overlying the Raritan Formation (Figure 8). This location was not ice covered during pre-Illinoian and Illinoian glaciations, and so would have been on the outwash plain (Figure 1).

This fluvioglacial deposit most likely correlates with Illinoian glacial meltwater deposits (Lamington Formation, Drakes Brook and Drainards outwash) in northern New Jersey and the Merrick Formation on Long Island. The surface altitude of the Drakes Brook outwash is between 198 and 201 m and the Drainards outwash altitude is between 122 and 146 m [22], while the Merrick Formation is seen in boreholes at a depth of between 15 and 20 m below present day sea level [21]. The outcrop height of 25 to 27 m in in this study fits well with a meltwater flow from the uplands of New Jersey towards the Hudson Shelf. Potentially, the outcrop on the southern shore of Staten Island [12,13] (SG2 on Figure 7), beneath the terminal moraine, is a more distal example of sediments on this outwash plain.

The sand (facies FS1) and sandy gravel (facies FSG) are typical of braided river deposits on an outwash plain [29,30,31], while the gravel (facies G) and overlying sand (facies FS2) represent a flood deposit, probably due to the rapid release of sediment-laden meltwater from a glacial lake, due to the breaching of an ice or moraine dam [30,32]. The change in clast composition from the base to the top of the outcrop is also significant for the interpretation of its provenance (Table 1). Pebble clasts from beds in Groups 1 and 2 contain a higher proportion of quartz and chert (~25–30%) than those of Group 3 (~14%) or the Gravel bed (17%). This suggests that much of the material deposited initially was reworked from the Pensauken Formation, which would have been exposed nearby, to the south of the ice sheet. Then as the ice retreated, there was a greater input of sediment from the Triassic age sedimentary rocks of the Newark Basin further north in New Jersey. The few granite clasts present are less weathered than those reported in pre-Illinoian sediments [12,19], confirming an Illinoian age. It is unknown how much more sediment was deposited before erosion continued, including during the expansion of the Wisconsinan ice sheet which covered the area [19]. However, this deposit remains, capped by a layer of ground moraine till (Figure 8). As the ice retreated, meltwater became dammed by the terminal moraine to the south. This led to the formation of glacial Lake Bayonne in the present-day Arthur Kill, one of many such lakes in the Hudson Valley and New Jersey [23]. The surface of this lake reached a height of at least 15 m above present-day sea level [4,23], and would have flooded the valleys to the north and south of the study area, with this topographic high potentially forming an island. The lake drained initially through the Richmond Valley spillway, probably through the southern valley in Figure 8. As the terminal moraine was eroded, the lake level dropped and drainage moved to the Perth Amboy spillway in the Arthur Kill [4,14,23], leaving behind the topography seen today.

In conclusion:

• The sediments investigated in this study of a small outcrop in southwestern Staten Island are fluvioglacial in origin, most likely of Illinoian age.
• These fluvioglacial sediments were deposited on an outwash plain in a low-lying area of a topographic surface, between older preserved sediments of the Pensauken Formation.
• Many sediments in this area were eroded by the later Wisconsinan ice sheet and the flow of meltwater through the Richmond Valley spillway.
• The few outcrops that are occasionally exposed in this densely populated area can reveal the complex sedimentary history.