Rivers Try Harder. Reversed “Differential Erosion” as Geological Control of Flood in the Large Fluvial Systems in Poland

: We study cross-sections on the Detailed Geological Map of Poland (SMGP) to ﬁnd a geologic and geomorphic pattern under river valleys in Poland. The pattern was found in 20 reaches of the largest Polish rivers (Odra, Warta, Vistula, Narew, and Bug) located in the European Lowland, in the landscape of old (Pleistocene, Saalian) glacial high plains extending between the Last Glacial Maximum (LGM) moraines on the North and the Upland on the South. The Upland was slightly folded and up-faulted during Alpine orogeny together with the thrust of Carpathian nappes and the uplift of Tatra Mts. and Sudetes. The found pattern is an alluvial river with broad Holocene ﬂoodplain and the channel developed atop the protrusion of bedrock (Jurassic, Cretaceous limestones, marlstones, sandstones) or non-alluvial, cohesive, overconsolidated sediments resistant to erosion (glacial tills, lacustrine or “ice-dammed lake” clays) of Cenozoic (Paleogene, Neogene, Quaternary— Elsterian). We regard the sub-alluvial protrusion as the limit of river incision and scour. It cannot be determined why the river ﬂows atop these protrusions, in opposition to “differential erosion”, a geomorphology principle. We assume it is evidence of geological ﬂood control. We propose an environmental and geomorphological framework for the hydrotechnical design of instream river training.


Introduction
Riverine floodplains cover ca. 1% of Earth's terrestrial surface, but are among the most distinctive landscape features on the globe, as well as the most biologically productive and diverse ecosystems on our planet [1]. Due to the great fertility of floodplains, people have reclaimed most of their surface. They had great cultural and economic importance for early civilizations, but floodplains became one of the most endangered landscapes on Earth [1]. Human pression on floodplains threatens them, and it resulted in the development of a false perspective-an overestimation of the role of human efforts to control the dynamics of the alluvial plains in the river valley, especially a proximal zone of these plains.
The proximal part of a floodplain [2] is a zone adjacent to the river channel, shaped by the deposition and erosion coming directly from the main channel. Sometimes such a zone raises above the flood basin level, and is called an alluvial ridge [3]. In fluvial sedimentology, the proximal zone of floodplain represents the following facies: channel-fill, channel-belt, natural levee, and crevasse-splay [4].
From the perspective of geomorphology, the floodplain is a landform developed by fluvial deposition (geological and geomorphological definitions compiled by [5]) for a long time, thus it can be regarded as a reflection of some tendencies in the river's behavior. From the point of view of geology, the reflection of the river's behavior (mentioned above) covers all territory of Poland by 1069 sheets (Figure 1) in a grid of the Polish Topographical Map according to the Coordinate system 1942 (EPSG: 3333, 3334 and 3335; the local implementation of Pulkovo 1942 coordinate system). All the sheets were elaborated by geologists and geomorphologists using the same, uniform instruction which is unique case in Europe and probably worldwide [35]. Each sheet of The Map contains a geological crosssection at a vertical scale of 1:2000. Our study focuses on these geological cross-sections. We review all the published sheets of The Map in order to find on the geological crosssections a repetitive pattern of geological setting under large, alluvial valleys. Such valleys exist in Middle and Northern Poland, but they dominate the landscape belt located between mountains and uplands in the south of Poland and The Lakeland (bright green and yellow) in the north of the country (colors according to Figure 1). Due to the dominance of large river valleys, this region was named The Belt of Great Valleys [36].
After a preliminary review of all published sheets (ca. 1% of 1069 sheets is unpublished yet) of The Map, the 20 of these sheets revealed the geological cross-section presenting the discussed pattern [11][12][13][14][15][16][17][18]. The pattern is known from previous studies which were based mainly on fieldwork [11][12][13][14][15][16][17][18]. These 20 cross-sections were split into 6 groups. Each group consists of the cross-sections through the same river (or the same reach as in the case of the Vistula River, the longest and largest one). In every group, we juxtapose the different cross-sections one on each other and it was done in order to locate the river channel at the same place and to follow the same orientation consistent with the general river flow direction in Poland, namely: SW on the left river bank and NE on the right river bank. We copy from the cross-sections (published below The Map) the following lines: (i) wetted perimeter of the river channel during a medium (lower than a bankfull) stage; (ii) surface topography of the floodplain, upper terraces and the valley edges (valley walls); (iii) the bottom of Holocene alluvial sediments which can be regarded as floodplain defined by [37]; (iv) bedrock in the extent of their protrusion under the river valley. As the bedrock, we regard not only the Mesozoic solid rocks (limestones, sandstones) but also glacial tills and lacustrine (or ice-dammed lake) cohesive sediments (mostly clays) deposited during the Cenozoic, but before the Holocene. These sediments are resistant to erosion (in comparison to the Holocene alluvial sand) due to their cohesive strength [38,39] and due to strength received from glacial pressure-preconsolidation [40][41][42]. The lines of the upper limit of the non-alluvial protrusion were marked by a color that refers to the age according to the chronostratigraphic chart [43].

Results and Discussion
We found that the specific and repetitive pattern of geological setting is in 20 sheets (2%) (red squares in Figure 1). Moreover, 75% of them occur in The Belt of Great Valleys [36] (dark green on the map in Figure 1). Furthermore, 25% of cases are located northerly, in the Lakeland-yellow-green on the map in Figure 1. These results can be biased by the frequency of the geological cross-sections through the river valley with the deep boreholes drilled on the floodplain. One of the reasons is the fact that in the 1950s there was a concept of building 9 dams in the Lower Vistula River valley (The Lower Vistula Cascade, pol. Kaskada Dolnej Wisły) [47,48]. Another reason can be a larger number of bridges crossing the Lower Vistula River in comparison to the Middle Vistula, Bug, and Narew Rivers.  Geological cross-sections through the Bug river valley, according to [49,50]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and nonalluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].  [49,50]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and non-alluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].
In the Bug river, the protrusions were identified on two geological cross-sections located near Wyszków [49] and near Janów Podlaski [50]. The simplified cross-sections are shown in Figure 2. The protrusion near Janów Podlaski is located in the Podlasie Gap of the Bug River (in Polish: Podlaski Przełom Bugu) [51] and is visible as a single hump built on Upper Cretaceous chalk rocks. The next protrusion near Wyszków is a Pleistocene glacial till hump and it repeats its shape underneath in the Neogene (Miocene and Pliocene) lacustrine clays. The Cretaceous bedrock is just around 2 m below the river channel bottom in the Podlasie Gap of the Bug River. Non-alluvial sediments in the Bug river reach near Wyszków are at a depth of ca. 10 m (Pleistocene) and ca. 70 m (Miocene and Pliocene). Geological cross-sections through the Bug river valley, according to [49,50]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and nonalluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].

Figure 3.
Geological cross-sections through the Narew river valley, according to [52][53][54]. Each cross-section presents on: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and Figure 3. Geological cross-sections through the Narew river valley, according to [52][53][54]. Each cross-section presents on: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and non-alluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].
Water 2021, 13, x FOR PEER REVIEW 6 of 17 non-alluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].

Figure 4.
Geological cross-sections through the Odra river valley, according to [55][56][57][58]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and nonalluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].  . Geological cross-sections through the Odra river valley, according to [55][56][57][58]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and nonalluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].

Figure 5.
Geological cross-sections through the Warta river valley, according to [59][60][61]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley Figure 5. Geological cross-sections through the Warta river valley, according to [59][60][61]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and non-alluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].
Water 2021, 13, x FOR PEER REVIEW 7 of 17 profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and nonalluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].

Results and Discussion
We found that the specific and repetitive pattern of geological setting is in 20 sheets (2%) (red squares in Figure 1). Moreover, 75% of them occur in The Belt of Great Valleys [36] (dark green on the map in Figure 1). Furthermore, 25% of cases are located northerly, in the Lakeland-yellow-green on the map in Figure 1.
These results can be biased by the frequency of the geological cross-sections through the river valley with the deep boreholes drilled on the floodplain. One of the reasons is the fact that in the 1950s there was a concept of building 9 dams in the Lower Vistula River valley (The Lower Vistula Cascade, pol. Kaskada Dolnej Wisły) [47,48]. Another reason can be a larger number of bridges crossing the Lower Vistula River in comparison to the Middle Vistula, Bug, and Narew Rivers. The specific pattern revealed in our study looks like the river channel sits atop the bedrock or non-alluvial protrusion, and it repeats in all the large rivers of Poland- Figures  2-7.
In the Bug river, the protrusions were identified on two geological cross-sections located near Wyszków [49] and near Janów Podlaski [50]. The simplified cross-sections are shown in Figure 2. The protrusion near Janów Podlaski is located in the Podlasie Gap of the Bug River (in Polish: Podlaski Przełom Bugu) [51] and is visible as a single hump built  [67][68][69]. Each cross-section presents: water in the river channel during medium stage (blue body); schematic river channel with flow direction (blue line) the valley profile graph with the floodplain surface and the valley's walls (black), alluvial cover of the floodplain (grey) and non-alluvial protrusion underneath. The color of the non-alluvial protrusion indicates its age according to the stratigraphic chart [43].
The non-alluvial protrusions under the Narew river channel Figure 3 were revealed at three locations: Łomża [52], Ostrołęka [53], and Plutycze [54]. The protrusion near Łomża forms a hump located deeply, ca. 100 m below the channel bottom. It consists of Paleogene (Eocene and Oligocene) lacustrine sediments. The protrusions near Ostrołeka and Plutycze are built of Pleistocene glacial tills, located 10-20 m below the river channel bottom.
The analysis of the geological cross-sections through the Odra river valley Figure 4 revealed four protrusions near Czerwieńsk [55], Sulechów [56],Ścinawa [57], and Oława [58]. At all four sites, the river flows atop the Neogene (Miocene) humps located less than 50 m below the bottom of the river channel. The protrusion near Oława is located just a few meters below the river channel bottom and it is built of Miocene lacustrine deposits and Pleistocene (Elsterian) glacial tills. The protrusion nearŚcinawa and Sulechów are made of Miocene lacustrine sediments.
The Warta river channel flows atop the Mesozoic protrusions, as revealed by three geological cross-sections in Figure 5. The first protrusion, near Koło [59], is built of Upper Cretaceous marlstones, limestones, and sandstones and it can be found around 10 m below the bottom of the river channel. The next one has similar lithology and stratigraphy (Maastrichtian) to that near Koło, but it is located near Warta (city) [60]. It is placed just 5 m below the river channel bottom. The third one is built of limestones, but older than those mentioned above-Upper Jurassic (Upper Oxford). It is located near Osjaków [61]. The protrusion is also shallowly located, just 5 m below the bottom of the river channel.
The analysis of the geological cross-sections through the Vistula river revealed 8 reaches with a specific pattern of geological setting. In order to increase the readability of our results, the river was divided into the Lower Vistula (5 reaches- Figure 6) and the Middle Vistula (3 reaches- Figure 7).
The protrusions on the Lower Vistula River are located near Kwidzyn [62], Ciechocinek [63], Płock [64], Słubice [65], and Wyszogród [66] as shown in Figure 6. The protrusion near Kwidzyn is located very deeply (70 m below the river channel bottom) and it is made of lacustrine deposits (glauconitic sands with phosphorites and clays) of the Paleogene (Oligocene) age. The protrusion near Ciechocinek is the oldest in the Lower Vistula river. It is made of Lower Cretaceous limestones and shales and Upper Jurassic (Portland) marlstones, mudstones and limestones located 20 m below the bottom of the river channel. The next protrusions are located adjacent to Last Glacial Maximum (LGM) margin near Płock and Słubice. They consist of Negoene (Pliocene) lacustrine clays. Their surface can be found just 10-20 m below the bottom of the river channel. The protrusion near Wyszogród comprises the river channel bottom and it is made of Neogene (Miocene) lacustrine clays.
The Middle Vistula river has three protrusions according to The Map: near Modlin [67], in the southern part of Warsaw [68], and near Kazimierz Dolny [69] visible in Figure 7. The protrusion near Modlin is made of ice-dammed lake clays dated on the Lower Pleistocene (Elsterian) and it is located 20 m below the river channel bottom. Protrusion near the southern margin of Warsaw is built of Neogene (Pliocene) lacustrine clays and it can be found 10 m below the bottom of the river channel. The Kazimierz Dolny protrusion is the oldest in the middle reach of the Vistula river-Upper Cretaceous (Maastricht). The hump is made of chalk (limestones) and it is located around 15 m below the bottom of the river channel.
A fact of the specific location (Figure 1) of the river valley reaches with the repetitive, specific pattern of geological setting (Figures 2-7) can be a result of several advances of Fennoscandian Ice Sheet in Polish Lowland [18]. A continental glaciation is a very significant and long-lasting (a timescale of Quaternary) disturbance for the fluvial systems, because glacial till completely covers the initial (preglacial) surface. Usually, after ice-sheet retreat, an incision occurs, and a new river valley has to be formed again. Bedrock protrusion can serve in such a case as the limit of an incision. Many large river valleys in Poland are landforms inherited from the ice-marginal valleys [70], which were impacted by fluvial and glaciofluvial deposition during the last (Vistulian; from German also: Weichselian) glaciation. This deposition formed the upper terraces that reworked the fluvial setting in the river valley and enabled the later floodplain formation in Holocene [30]. The fluvial system can recover therefore to the initial state but in different postglacial frames. However, the river channel tends to anchor on the non-alluvial protrusion again.
Hauer and Pulg [71,72] found a similar pattern in the Norwegian rivers and they linked its origin with long-lasting river incision into fluvial deposits that has exposed glacial tills buried underneath. However, Hauer and Pulg [71,72] described such non-fluvial protrusions in the longitudinal profile of mountainous streams in Norway. The results of our study reveal the existence of non-fluvial protrusion in the transverse profile of the river valley and locate the protrusion exactly below the river channel.
Our findings can prove useful in understanding the geological control of Holocene flood dynamics in large, alluvial valleys ( Figure 8). We distinguish two different types of flood: (I) events induced by ice-jam when relatively small discharge in the channel can involve overbank flow on the floodplain (Figure 8-left side); and (II) rainfall or snow-melt events with very high discharge in the river (Figure 8-right side).
(I) Jamming of the river channel by ice leads to scouring. When scour reaches the bedrock protrusion, it cannot be eroded due to high resistance and an overbank flow initiates on the floodplain surface. The overbank flow cuts in the floodplain surface an erosional through, a landform that is parallel to the main channel and serves as the temporary sidearm. At the final phase of the flood, a filling process occurs at the bottom of the channel. Such a type of flood tends to occur very often, and it has a local extent. The pattern of geological setting, which we reveal in Figures 2-7 and idealized on the scheme in Figure 8, can be adopted as an environmental framework by hydrotechnical engineers who aim to develop instream river training [73,74]. Such a design practice faces a problem of too high stream power [39,[75][76][77][78][79][80][81][82][83][84][85] in the channel which leads to erosion. There are few concepts of energy dissipation of the stream. One of them involves putting the concrete baffles or micro groynes into the channel [73,74,86]. (II) Snowmelt or rainfall flood acts, in opposite to ice-jams, in the non-confined river channel. Overbank flow on the floodplain occurs only on the condition of very high discharge and it can initiate after breaching of natural levees at the banks of the river channel. Scour reaches the bedrock protrusion as in the case of ice-jam, but overbank erosion on the floodplain is limited to the crevasse channel cut in the natural levee. The erosional landforms can be masked by the latter rebuild of levees.
The pattern of geological setting, which we reveal in Figures 2-7 and idealized on the scheme in Figure 8, can be adopted as an environmental framework by hydrotechnical engineers who aim to develop instream river training [73,74]. Such a design practice faces a problem of too high stream power [39,[75][76][77][78][79][80][81][82][83][84][85] in the channel which leads to erosion. There are few concepts of energy dissipation of the stream. One of them involves putting the concrete baffles or micro groynes into the channel [73,74,86].
The present state of the art in science is not enough to explain or even understand the origin of the specific geological pattern of river channel flowing atop bedrock protrusion we present in this study. If this pattern exists more widely in fluvial systems on Earth, some basic principles of fluvial geomorphology should be revised. The "differential erosion" concept says that a river uses less resistant (to erosion) bedrock to form its channel and whole fluvial valley. The results of our study prove something different, that rivers try harder. Rivers tend to flow on the most resistant (to erosion) rock available under the alluvial belt.
The study area underwent glaciotectonic modelling [87][88][89] and a postglacial rebound, glacial isostatic adjustment [90], which undoubtedly modified non-alluvial protrusions, or even formed these structures from the relatively flat preglacial surface ( Figure 9). The present state of the art in science is not enough to explain or even understand the origin of the specific geological pattern of river channel flowing atop bedrock protrusion we present in this study. If this pattern exists more widely in fluvial systems on Earth, some basic principles of fluvial geomorphology should be revised. The "differential erosion" concept says that a river uses less resistant (to erosion) bedrock to form its channel and whole fluvial valley. The results of our study prove something different, that rivers try harder. Rivers tend to flow on the most resistant (to erosion) rock available under the alluvial belt.
The study area underwent glaciotectonic modelling [87][88][89] and a postglacial rebound, glacial isostatic adjustment [90], which undoubtedly modified non-alluvial protrusions, or even formed these structures from the relatively flat preglacial surface ( Figure  9). Figure 9. The geological pattern of the river channel flowing atop the non-alluvial protrusion on the cross-section through the Lower Vistula River at Troszyn-Wykowo (near Płock) [91], modified by [92] with the locator map of Poland from Figure 1. The red rectangle indicates the location of the cross-section exactly at the margin of the LGM paleo-ice stream (Vistula Ice Stream-compare [89,93]).

Conclusions
A basic source of geological information in Poland (SMGP; The Map) enables us to determine the specific pattern of geological setting under the large, alluvial river valleys. The pattern can be identified in the in 2% sheets of The Map, however, it exists under all the large rivers of Poland: Odra, Warta, Middle Vistula, Narew, Bug, and Lower Vistula.  [91], modified by [92] with the locator map of Poland from Figure 1. The red rectangle indicates the location of the cross-section exactly at the margin of the LGM paleo-ice stream (Vistula Ice Stream-compare [89,93]).

Conclusions
A basic source of geological information in Poland (SMGP; The Map) enables us to determine the specific pattern of geological setting under the large, alluvial river valleys. The pattern can be identified in the in 2% sheets of The Map, however, it exists under all the large rivers of Poland: Odra, Warta, Middle Vistula, Narew, Bug, and Lower Vistula.
In a wide and alluvial valley, a river tends to flow above the protrusion of non-alluvial sediments which are the most resistant to erosion and the oldest ones among different deposits available under the alluvial belt.
The geographical distribution of the pattern in Poland indicates its dominance in the broad valleys which are cut by the rivers flowing through the old-glacial (Pleistocene but beyond the LGM) high plains-a type of morainic plateau without postglacial lakes. This location used to be named The Belt of Great Valleys.
The geological pattern controls on the Holocene incision of large rivers, what we regard as a reasonable driver of fluvial resistance [16]. The pattern also controls on floodplain recovery after a disturbance because a river channel stays at the same location in the alluvial belt during thousands of flood events in the Holocene. The control of incision resistance combined with the ability of floodplain recovery create the resilience of the studied fluvial systems, as proposed by [94].
A reason for horizontal stability of the channel exactly atop the non-alluvial protrusion during a very long period is unknown. Further research on this topic is necessary.
The time of lateral stability of the river channel is the Holocene in the LGM extent, but beyond the LGM extent, it can last longer, in some cases at least since the younger interglacial periods of the Pleistocene, or even since the preglacial period of Quaternary.
Our conception of reversed "differential erosion" on Polish Lowland should be verified through a perspective without tectonic quiescence, similar to other revisions of the old-fashioned topics in geomorphology, eg., revisions of planation surfaces in the Outer Carpathians [27] and in Northern Etiopia [95].

Acknowledgments:
The authors would like to thank Tomasz Falkowski for the inspiration to search for the geological pattern under the large rivers in Poland. The inspiration came to the mind of the 2nd author after having a look at the Bug River geological cross-section drawn by Tomasz Falkowski and Piotr Ostrowski. The 1st author would like to thank Hanna and Stefan Mikulik for inspiration to study Water Engineering and Management. The second author thanks also Michał Wierzbicki for indicating the instream river training design. The authors acknowledge the valuable comments and suggestions received from two anonymous reviewers. The paper was submitted to the Special Issue "Fluvial Geomorphology and River Management" edited by Salvatore Ivo Giano.

Conflicts of Interest:
The authors declare no conflict of interest.