Sedimentary texture of crevasse splays formed by present-day and palaeofloods against the background of floodplain geomorphology and lithofacies exposed in channel cut banks (in the Vistula River valley between Warsaw and Płock, Poland)

A floodplain (different definitions given by Lóczy, Pirkhoffer & Gyenizse 2012) is an area: (i) built on alluvial deposits – the geological definition, (ii) that forms a nearly flat surface similar to a river longitudinal slope – the morphometric definition, and (iii) reveals the dominance of alluvial stratified soils (Luvisols). Lóczy, Pirkhoffer & Gyenizse (2012) also refer to hydrological and geomorphological definitions, where a floodplain is inundated by a flood and shaped by fluvial erosion or deposition, which is written in specific landforms.

In our study, we delve deeper into the geological definition; thus, we focus on alluvial deposits. We use sedimentological analysis of lithofacies as a tool to reconstruct the dynamics of the flood events that created the overbank deposits (Paola 2003, Smith et al. 2010); however, we link the sedimentological structures with the geomorphological features of the floodplain (Miall 1985, Korus & Joeckel 2023). We perceive a floodplain as an area of fluvial deposition, where the grain size of alluvium gradually decreases as a result of sedimentation from the overbank flow with its gradually decreasing velocity on its way from the channel to the floodplain margin (Fig. 1). The most dynamic deposition occurs in the proximal zone of the floodplain (Fig. 1), where several natural levees compose a convex landform of alluvial ridge (Kiss, Sipos & Vass 2022). Along an alluvial ridge, there are some locations where overbank erosion tends to appear repetitively, which induces the development of many crevasses or crevasse channels (Balogh et al. 2020).

Figure 1.

In such a location, with the predominance of crevasse channels in the alluvial ridge, we aim: (i) to compare the textural features (grain size) of two crevasse splays, formed by overbank flow, in the same location but in two different settings – a splay developed during present-day flooding as a result of an artificial levee (dike) breach and a splay formed on the floodplain before the construction of artificial levees; (ii) to contextualise the location of both splays in local floodplain geomorphology; and (iii) to analyse lithofacies of deposits exposed in cut banks of the adjacent river reach in order to determine the complexity of the sedimentological archive.

The research was conducted on a floodplain (Fig. 2) of the Lower Vistula River valley, located 70 km downstream from Warsaw city centre and few kilometres upstream from the proximal part of the Włocławek Reservoir near the city of Płock. The study area is adjacent to the Last Glacial Maximum (LGM) margin (Tylmann et al. 2022, Forysiak et al. 2023) and was a part of the LGM ice-marginal valley (Wierzbicki et al. 2020), which resulted in the high supply of fluvioglacial sand at the end of the Pleistocene.

Figure 2.

We used a combination of fieldwork, GIS & RS methods and laboratory techniques to achieve three different objectives: (i) determination of the textural features of splay sediment using statistical analysis; (ii) geomorphological mapping; and (iii) mapping of the distribution of lithofacies.

Objective (i): The sampling material was taken from the crevasse splays. Samples of the present-day flood splay were taken both from the surface and the inner part of sand bodies penetrated by excavations. The boundaries of the splay were easily identifiable, as they were clearly visible in the field (Wierzbicki et al. 2018). Sampling of the palaeoflood splay was possible after sophisticated remote sensing of the landform boundaries (Wierzbicki et al. 2013) and the use of mobile GIS techniques (Wierzbicki, Ostrowski & Falkowski 2020) to detect the boundaries in the field. A bias may occur in sampling on the palaeoflood splay because the material originated from boreholes; however, in both splays the thickest parts of sand bodies were sampled.

The samples were analysed in a sedimentological laboratory (Warsaw University). The laboratory work included grain size analysis on sieves for coarser samples, and by the aerometric method of Casagrande with the Prószyński modification for finer samples (Myślińska 2010, Prószyński 1949). Using a spreadsheet and simple computer program developed by Błażej Stankowski (Adam Mickiewicz University, Poznań), we recalculated the results of the grain size analysis into grain size composition parameters – statistical approach, according to Folk & Ward (1957).

Objective (ii): Using topographical and geological maps, remote sensing, GIS and field mapping with mobile GIS, we studied the floodplain geomorphology in order to draw a geomorphological map comparable to the model (Fig. 1). Detailed methods are described in previously published papers – Wierzbicki et al. 2013, 2018 and Wierzbicki, Ostrowski & Falkowski 2020. The geomorphological map also enabled us to choose a selection of the floodplain locations where we performed the soil sampling.

Objective (iii): Using a boat, we also explored sediments exposed in cut banks of the river channel. Forty sections were described using the lithofacies code (Zieliński 1998, Zieliński & Pisarska-Jamroży 2012), which denotes textural (lithology, grain size) and structural (sedimentary structures) features. Following the model (Fig. 1) and literature sources (Zieliński 1998), we attributed the particular interval of sections to certain facies (zones) of the fluvial environment (channel, proximal floodplain, distal floodplain).

The floodplain in the study area (Fig. 2) is asymmetric: the right side is narrower and reduced by erosion; the left side is wider and developed similar to the conceptual model (Fig. 1). A detailed description of the floodplain geomorphology (Fig. 2) has already been published (Wierzbicki, Ostrowski & Falkowski 2020). Crevasse splays and crevasse channels significantly predominate the riverscape of the left side floodplain between the km 600 and km 617 markers of the Vistula waterway (Fig. 2). We focused on the palaeoflood splay formed around the largest crevasse channels (Fig. 3, blue lines) and on the present-day flood crevasse splay (Fig. 3, yellow lines). The origin, age and shape of these two splays were analysed in detail in our previous study (Wierzbicki et al. 2013, 2018) but without the sedimentological approach, which we present in figures 4 and 5.

Figure 3.

Figure 4.

Figure 5.

The results of sieve analysis and the aerometric method (Fig. 4) show that the present-day flood splay (Fig. 4B) consists of grains that are a little coarser than the palaeoflood one (Fig. 4A), but are similar, in general. Most of the grains in both splays were deposited from transport in saltation and they represent a size of fine and fine-to-medium sand – the diameter varies from φ=4 (0.0625 mm) up to φ=0 (1 mm); however, 66% of the material taken from the palaeoflood splay (Fig. 4A) has a significant admixture of finer particles deposited from transport in suspension. These finer particles represent very fine sand, silt, and even clay – the diameter is smaller than φ=4 (0.0625 mm).

The grain size parameters calculated according to Folk & Ward (1957) locate the present-day splay deposits (Fig. 5) exactly on the margin of two fluvial facies: the channel and the proximal zone of the floodplain (overbank). The palaeoflood splay deposits also belong mostly to the overbank facies of the proximal floodplain but some of the samples reveal values typical for facies of the flood basin – the distal zone of overbank sedimentation on the floodplain.

The 40 cut banks with exposed alluvial deposits are presented in supplemental information (SI); each one is described using the lithofacies code and with the putative identification of the depositional environment (zone) – channel, proximal floodplain or distal floodplain. The sections are numbered (see SI) according to the detail on the maps (figures 2 and 3). For two of the sections – one that we found the most typical (No. 1) for the study area and another the most extraordinary (No. 15), we also present pictures (photos) taken in the field – figures 6A and 6B, respectively.

Figure 6.

We consider section No. 1 (Fig. 6A and SI section No. 1) to be the most representative sequence of deposits (for the whole study area), exposed in the cut banks of the channel. It consists of the lithofacies representing all three sedimentary zones (cf. Fig. 1) recorded in the study area – namely channel, proximal floodplain and distal floodplain. It is not only the sedimentary structures that are representative for the study area but also the sedimentary texture. Besides the fine sand that makes up most of the section, there is also the significant presence of horizontally laminated muds that form the middle part of the section. Such thick, muddy layers could be considered to have been deposited in the distal part of the floodplain but they could also be interpreted as the deposits infilling the channels that were abandoned as a result of avulsion.

Such muddy deposits were recorded in the following sections: Nos. 2, 5, 7 (but the layers are thinner there), 9, 15, 18, 20, 22, 24, 25, 26, 29, 31, 34, 35, 36, 37, 38, 39 (see SI); thus, in almost half of the mapped cut banks. These facies, interpreted as those of distal floodplain, constitute 30% of the total deposits exposed by us in cut banks.

We consider section No. 15 (Fig. 6B and SI No. 15) to be the most extraordinary sequence of alluvial sediments in cut banks because there we found a layer of gravel deposited inside the muds. The largest particles are very coarse gravel (pebbles) and even some cobbles. Another section with gravel admixture are Nos. 29 and 36 (SI) but these sections do not have any cobbles or coarse pebbles. This textural feature is characteristic of channel lithofacies, which make up 10% of all deposits mapped by us within the cut banks.

The sedimentary texture of the studied splays (figures 4 and 5) are consistent with the model of lithofacies distribution on a floodplain (Fig. 1). The palaeoflood splay (Fig. 3, sampling points 73 and 74) consists of slightly finer grains than the present-day flood splay (Fig. 3A). We suppose three possible reasons for such a difference: (i)

The palaeoflood splay was deposited before the artificial levees had been constructed; thus, this landform originates from the breaching of natural levees, which were significantly lower than the dikes (embankments) designed and built by humans. As a result, the hydraulic conditions of overbank flow were completely different in each case. In the case of the present-day splay from the 2010 flood, the estimated hydraulic parameters in the area of the levee reached enormous values (flow velocity > 5 m·s⁻¹, gross stream power = 981 W·m⁻¹, specific stream power = 9 810 W·m⁻²) (Wierzbicki et al. 2013) but they abruptly decreased, inducing deposition on a relatively small area < 0.5 km² with gradually decreasing grain size (Szmańda 2018, Fig. 39 therein). The palaeoflood splay was deposited (and possibly redeposited) on an area > 5 km² during many flood events (Wierzbicki et al. 2013). On the basis of a study that was very similar to our work but performed in the Upper Vistula River valley after the 1997 flood, Gębica & Sokołowski (2001) underlined the differences between splays formed during artificial and natural levee breaches. In contrast to the opinion of Gębica & Sokołowski (2001), we perceive the splays developed at the sides of breached dikes (artificial levees) as analogues of natural landforms.

We know exactly how the development of the present-day splay occurred. It was recorded in detail through very accurate and comprehensive data: hydrological observations in the adjacent river gauge equipped with a telemetry system, which records the water stage every 10 minutes; direct ADCP (acoustic Doppler current profiler) discharge measurements from a boat in the river channel upstream, downstream and around crevasse channel during the floodwave passage shortly after the levee breach; capturing the splay by laser scanning from a plane (ALS LIDAR); and echo sounding (bathymetry measurements) of the crevasse channel and scour. All of these data are presented, analysed and discussed in Wierzbicki et al. 2013, 2018, Wierzbicki, Ostrowski & Falkowski 2020. Regarding the variety of these data, our sedimentological analysis of the present-day splay is provided as a complement only. If these data were not available, the sedimentological data would be then the main basis for reconstructing the hydraulic conditions of fluvial deposition – see Weckwerth (2011) for the lower reach of the Vistula River during the Vistulian (Weischelian) glaciation period in the Pleistocene, or present-day fluvioglacial environment dynamics in the Arctic (Weckwerth, Greń & Sobota 2019) or GLOF reconstruction (Weckwerth et al. 2022). This potential for analysis of lithofacies is a key advantage in areas that underwent significant landscape change in the Pleistocene and Holocene, but it is not possible to detect these changes by traditional geomorphological analysis (Mycielska-Dowgiałło & Woronko 2004).

A relatively high percentage (30%) of overbank deposits, mapped by us in cut banks and representing distal floodplains, could lead to the conclusion that the river channel must have been far from the present-day location; however, overbank deposits of a braided river are usually more diversified (Zieliński 1998) than those of a meandering river – for example, the Warta River (Skolasińska et al. 2010; Skolasińska 2014) and the Middle Vistula River (Falkowska & Falkowski 2015, Falkowska et al. 2016).

Using sedimentological analysis as a tool to attempt the reconstruction of the Holocene evolution of the study reach, we cannot confirm the existence of a meandering river in the Lower Vistula River valley, as was proposed by Falkowski 1967 and 1975, for the Middle Vistula River, and further adopted in the study reach by Florek, Florek & Mycielska-Dowgiałło (1987), Chormański & Mycielska-Dowgiałło (1996), and Brinkmann, Magnuszewski & Zober (2000). This is because the mud thickness in the studied reach is too low in comparison with the upper reaches of the Vistula River – namely the Middle Vistula (upstream from the Narew outlet) and the Upper Vistula (upstream from the San outlet) – see, for example, Falkowska & Falkowski 2015, Falkowska et al. 2016, Sokołowski et al. 2014. It is also lower than in the Danube River (Lehotský et al. 2010a,b).

The sediments consisting of large-sized grains (i.e. pebbles, see Fig. 6B) might be deposited under specific conditions only – for example, in the case of an artificial levee breach when the stream power tends to be enormous (cf. Wierzbicki et al. 2013, 2018, Wierzbicki, Ostrowski & Falkowski 2020). We suppose that ice-jam floods could also induce such conditions (Wierzbicki & Mazgajski 2011; Ostrowski, Falkowski and Utratnażukowska 2021) but we do not exclude a human-altered scenario, in which the source of coarse pebbles and cobbles could be that a channel bottom, dredged by suction excavator, breached into the suballuvial layer of Pleistocene fluvioglacial or glacial deposits. Similar research performed upstream (a reach of the Middle Vistula River channel near Kozienice), and in the Pilica River channel, links the coarser alluvial deposits in the cut banks with bedrock protrusions at the bottom of the channel (Falkowski & Górka 2009). The bedrock protrusions are a common feature of the geological control of flood behaviour in all the largest Polish rivers – namely Vistula, Odra, Warta, Bug and Narew (Bihałowicz & Wierzbicki 2021). The coupling of these geological structures with crevasse channels and splays was confirmed in the study area, as well as in two another river reaches (Wierzbicki et al. 2018).

Present-day splay has similar sedimentary texture to the splay formed by palaeofloods; thus, splay developed due to an artificial levee breach could be regarded as an analogue of the natural landform of crevasse splay.

Most of the deposits exposed in the cut banks (60%) represent the lithofacies of proximal floodplain (levee, crevasse splay); thus, they are consistent with the model of a meandering river (Fig. 1).

The percentage of distal floodplain lithofacies (30%) in cut banks is three times higher than channel lithofacies (10%). This could be interpreted in two ways: as a result of channel avulsion, or as a characteristic of a braided river’s great diversity of overbank deposits compared with a meandering river.

Sedimentological analyses have proven their usefulness for the verification of geomorphological maps of floodplains. These analyses can be a tool for the reconstruction of flood dynamics, especially in case of a lack of data coming from direct observation and instrumental measurements.

Repetitive deposition of crevasse splays in the study reach is driven by an obstacle in the channel in the form of bedrock protrusion.