Storm Surges Versus Shore Erosion: 21 Years (2000–2020) of Observations on the Świna Gate Sandbar (Southern Baltic Coast)

The study area extends along a 16-km long stretch of the shoreline of the Usedom and Wolin Islands (southern Baltic Sea, Pomeranian Bay), forming a sandbar (Fig. 1A, E). The Pomeranian Bay in its southern part is shallow, with depths of up to 2–5 m close to the shore, and displays a system of 2–3 bars near the shore. The Świna Gate Sandbar consists of two sand spits between the moraine plateaus of the two islands (Fig. 1D). The sandbar is located between Polish coast kilometres 412 and 424 (shown in Fig. 1D); kilometre 0 is at the border between Poland and the Russian Federation’s Kaliningrad Region on the Vistula Spit, while kilometre 424 marks the border between Poland and Germany on the Świna Gate Sandbar; the coast kilometrage is used to locate the measurement points as well as for the location and management of threats along the coast. The two Świna Gate Sandbar spits emerged as a result of accumulation of the marine sand eroded from the Wolin and Usedom moraines (Łabuz 2009). The dunes and beaches are built by medium and fine sand. The annual sea level rise (SL) in Świnoujście exceeds 1.1–1.3 mm (Zeidler et al. 1995, Richter et al. 2012). The highest surge in the Pomeranian Bay recorded in recent years exceeded 1.86 m AMSL (Sztobryn et al. 2005). The concave shoreline of the Pomeranian Bay is exposed to NE or NW storms, for which reason coastal erosion is observed in the eastern or western part of the shore, depending on the main direction of waves advancing onto the shore and into the near-shore zone (Łabuz 2009). Long-term observations showed the shore in the eastern part of the sandbar, exposed to NW storm surges, is more prone to erosion, compared with the western part, affected by the NE winds. However, due to the accumulative nature of the sandbar in-between surges, the beach and dunes are usually rebuilt to their former state or to a new one (up to progradation of new dunes).

The hourly wind information (velocity, force, direction) for each surge was extracted from data of the Polish Maritime Bureau and readings at the Institute of Meteorology and Water Management (IMGW 2022) weather station at Świnoujście, a town located in the southern part of the Pomeranian Bay. The wind force associated with the low-pressure system is responsible for waves from a specific direction, which is important for a shore with an opposite exposure. Information on the occurrence of storm surges and their characteristics was collected, for the 21-year period of study, from the Polish Maritime Bureau and the Harbour Master’s Office in Świnoujście. In relation to the Amsterdam level, the zero level of the tide gauges in Poland and in Świnoujście is PN = NN-500 cm (Sztobryn et al. 2005). For this study, were selected surges producing an SL higher than 0.8 m above NN and AMSL. Such surges cause the state of alert for the town of Świnoujście. The temporal resolution of the data was 1 h. The main descriptors of the storm magnitude included the highest storm sea level (H_SL), the main wave direction (and possibly its change), and the duration (hours; T_SL) of a surge higher than 1 m (T_SL with H_SL > 1 m); the runup resulting in the SLr was also taken into account (see below). The storm surges were named after those used in media coverage in the Baltic countries, and were cross-checked against the lowpressure system names used in the Institute of Meteorology of the Free University of Berlin (FU 2000–2022, Wetterzentrale 2022). However, not every surge could be named as not every low-pressure system observed during the on-going storm was named. A total of 62 significant storm surges with SL higher than 0.8 AMSL (H_SL > 0.8 m) were observed. To examine the storm impact on the shore, the highest surges that appeared in each year were selected (Table 1).

Changes in the relief of coastal forms were determined based on the analysis of cross-shore profiles (transects) from the waterline to the stable dune. More than 80 profiles from two contrasting locations, exposed to different surge directions, were analysed (Fig. 1D). One location (at kilometre 422) is exposed to NE storm surges, the other one (kilometre 416) being exposed to NW surges. The transects extended from the fixed and stable parts of the dunes to the waterline. Measurements along the transects were before and after each selected storm surge, within 2000–2020. The dynamics of the shore relief could be followed this way as well (Fig. 2). The field measurements involved the use of geodesic tools, including a cross-shore linear levelling (Ni-20) and a 3D levelling using the GPS RTK base Hiper II and Hiper SR made by TOPCON. Some observations were also made with a TLS ground laser scanning device TLS-1500 made by TOPCON (Fig. 3). Field observations on the presence of a storm surge and its impact on the shore morphology involved the measurement of the highest surge range, called the run-up or swash (SLr); it is marked by erosional cuts, as well as the presence of debris on the beach or in the runnel behind the eroded dune (Figs 6 and 7).

Fig. 2

Fig. 3

The transect measurements were processed, using a range of software products (Microsoft Office Excel 97, Grapher and Surfer by Golden Software, Grab it!, Statistica and Quantum GIS), to determine changes in the beach and dune height and width, followed by calculation of the sediment volume displaced from every square metre of the shore. This way, changes in the sediment volume were determined for individual coast morphology features, i.e. the lower beach, the upper beach and mainly the foredune. The range of shore relief change-related variables used in subsequent analyses included the movements of the foredune base and ridge or edge, the foredune height and the dune base width, and the beach width and height, as well as the height and dynamics of embryo dunes on the beach. The plotted dynamics layer shows short-time changes in the shore relief surface (Fig. 4). The changes are subsequently used to calculate the sand volume per metre in a 1 m² cell surface. The variables associated with changes in the beach morphological forms (x) reflected the present, seasonal, annual and post-storm surge state and included:

– wind: maximum velocity (V_w), time (T_w) V_w > 10 m · s⁻¹ and azimuth (Az_w),

Fig. 4

Data for these variables were calculated based on transect data entered in Excel spreadsheets. Morphological changes were calculated for every square metre of each beach form (x) from the state determined before (B_xi) and after (A_xi) a measurement/storm surge (where i is the length of a form/profile in metres). The data were calculated mainly for the foredune (fr) and the beach (be). Further comparisons involved the following selected indicators: the foredune foot retreat (ΔP_fr) in metres; the foredune volume change (ΔQ_fr) in cubic metres; and dune width and beach sand volume changes (ΔQ_fr/1m and ΔQ_be, respectively) for each metre. Additionally, the beach height (before and after surge) and its change (ΔH_be) were analysed in relation to the SL and the run-up (SLr). The sand volume (Q_x) for the shore morphological form (x) was calculated using formulas developed earlier, where the change in sand volume is calculated per square metre surface (d) of each shore form (xi) as a sum of values derived from the profile (transect) width: Qx=ΣHxi⋅d {{\rm{Q}}_{\rm{x}}} = \Sigma {{\rm{H}}_{{\rm{xi}}}} \cdot d and added up to obtain the total volume change between the pre- and post-storm surge events (A and B, respectively): ΔΣQx=ΔQAxi−ΔQBxi. \Delta \Sigma {{\rm{Q}}_{\rm{x}}} = \Delta {\rm{Q}}{{\rm{A}}_{{\rm{xi}}}} - \Delta {\rm{Q}}{{\rm{B}}_{{\rm{xi}}}}.

Changes in the beach and foredune width (ΔW_x), the foredune foot (ΔP_fr) and the position of the dune top (ΔPK_fr) were calculated in a similar manner.

The final step involved exploration of relationships between the field measurements and storm surge parameters. This was done using the correlation analysis by calculating appropriate Pearson’s correlation coefficients (Table 2).

In 2000, there was one surge associated with a storm named Gilda, with the SL rising to H_SL = 1.01 m AMSL in the Świnoujście harbour. The surge produced no significant beach and dune erosion on the Świna Gate Sandbar, except for the lower part of the beach being flooded because of a higher upper beach. The water run-up on the shore exceeded 1.4 m AMSL.

The autumn-winter season of 2001/2002 witnessed five major surges of H_SL < 1 m AMSL. Each surge was produced by NW to NE or NE to NW winds and resulted in shore erosion along the entire sandbar. The two largest surges (H_SL of 1.44 m AMSL) occurred in January and February 2002 and washed away the young, 3–4 m high, foredunes formed on the upper beach between 1997 and 2001, following the sandbar erosion caused by a heavy storm in November 1995 (Łabuz 2009). The dune erosion was considerable because, in the first place, the beach was lowered and the foredune slopes were somewhat under-cut by the first storm in November 2001 (Janika); in early 2002, the dune erosion reached major dimensions. In November 2001, the dune receded by 2–3.5 m (Fig. 7, 2001, Fig. 8). The February 2002 storm (Wisia) (Fig. 10D), with H_SL = 1.44 m AMSL, caused the dune to recede by a further 6 m due to the presence of a low beach that had eroded in 2001. During the February 2002 storm, the SLr above 1 m was observed for 11 h. The water run-up reached 3.3 m AMSL. Along the spit section examined, three-fourths of the dune volume was washed away, the dune base receding by 8–9 m. Approximately 1.2–1.3 m³ · m⁻² of sediment was removed from the dune surface, the removal from both sandbar sections totalling 7–8 m³. The beach erosion exceeded 13 m³ · m⁻¹ (per metre of shore width) (Fig. 8). As a result of aeolian accumulation, the foredune was rebuilt along the entire sandbar in 2002 and 2003, including the sections analysed here (Łabuz 2009). Two consecutive storms in December 2003 with H_SL > 1 m AMSL resulted in another retreat or partial washing-out of the foredune ridge (Fig. 6, 2003, Fig. 7, 2003). The foredune foot retreat did not exceed 1 m. In places where the beach was lower than 2 m above the SL, the ridges were broken (Łabuz 2009). The main surge, when the SL exceeded 1 m, lasted only 3 h.

The storm in November 2004 (surge Pia), induced by NW–NNE winds, produced a surge that resulted in erosion of the dunes in the area examined and reduced their volume by up to onethird (Łabuz, Kowalewska-Kalkowska 2011). The SL reached 1.37 m AMSL and remained above 1 m for 15 h. The water run-up was 3.3 m AMSL. The dune base in the western part was subject to erosion by 6 m, the erosion in the eastern part – with NW waves – amounting up to 12 m (Figs 6 and 7, 2004). The eroded volume of dune and beach sediment exceeded 10 m³. The following winter (2004/2005) brought no major storms (with H_SL > 1 m AMSL). In November 2006, however, a storm (surge Britta) produced a surge with the highest SL thus far, with a 1.48 m AMSL (described by Kettle 2015). The main surge came from NW, resulting in a larger erosion in the eastern part of the spit. In the foredune base, erosion reached 7–8.5 m. The beach erosion was one of the highest recorded in the preceding 20 years, ranging from 14 to 15.8 m³. The water run-up, up to 3.6 m AMSL, washed away the foredune ridge almost completely (Figs 6–8). The water flooding through the storm gates produced erosion and accumulation of material in the interdune runnel (a washover fan).

Fig. 8

Subsequent major surges (associated with storms, the main one among which was named Kyrill) occurred in January 2007, causing further erosion and retreat of the dune base. Between November 2006 and January 2007, there was no chance for the beach or dune to rebuild. Throughout the whole of January 2007, the SL exceeded 0.5 m AMSL. There were three storm surges with SLs of 1.19–1.38 m AMSL. As a result of the highest one, with WNW–NW winds, the water run-up reached 3.2 m. The dune base retreated by 4 m and 9.5 m in the western (kilometre 422) and eastern (kilometre 416) parts of the sandbar, respectively. The beach sediment volume was reduced by 8.9–9.4 m³. Along the entire length of the sandbar, the foredune was completely eroded in both sections examined, whereas a somewhat less severe erosion was observed in the middle section (kilometre 420). The water flooded the eroded foredune ridge into the interdune space to create new washover fans.

The dune was rebuilt again in mid-2008, but retreated several metres into the interdune runnel. The process of dune ridge retreat and rebuilding higher up on the shore is typical of abrasive shores (e.g. Carter et al. 1990). In the meantime, small storm surges were recorded in March and October 2008; these, however, lowered the beach surface only. By autumn 2009, the sediment volume and dune height had increased. In October 2009, a prolonged and strong surge, elicited by a low-pressure system Wimar, eroded the dunes again. As a result of NNE–NE winds, the SL rose to 1.33 m AMSL and the run-up amounted to 3.2 m AMSL. The SL exceeding 1 m prevailed for 30 h. This was one of the longer storms, and produced a double surge (one surge after another in a short period of time). A shore section with a low beach profile in the eastern part of the area (kilometre 416) suffered major erosion. The dune in the section with the beach rising to 2.1 m AMSL retreated by 5.3 m, while that in the section where the beach was up to 2.8 m high retreated by as little as 1.5 m (Figs 6, 7). Beach erosion was less severe in the accumulated section (up to 2.3 m³) and heavier on the eroded section with a lower beach (9.2 m³). The marine and aeolian accumulation resulted in rebuilding of the beach and the foredune in spring 2010.

The surge in December 2010 (associated with an NNE storm named Xynthia) slightly undercut the lower part of the foredune slope and lowered the beach. The surge heightened the SL to H_SL = 1.02 m AMSL (Fig. 10B). The dunes were rebuilt in 2011 and were not eroded during the minor storm Joachim that took place in December 2011. Only the beaches at an SL of 1 m AMSL were lowered.

Two very strong surges in January 2012 (associated with the storms Andrea I and Andrea II) resulted in beach and dune erosion along the whole Polish coast (Łabuz 2014). The storm Andrea I occurred in early January and was followed 6 days later by Andrea II (or Elfriede, according to the Institute of Meteorology of the Free University of Berlin). The maximum SL exceeded 1.4 m AMSL (Fig. 10E). The SL of H_SL > 1 m persisted for as long as 39 h. These were the 21st century’s two longest storm surge events at the southern coast of the Baltic Sea. The maximum water run-up reached 3.4 m AMSL. The water eroded the dunes where the beaches were lower than 2.5 m AMSL. This explains the large difference between the profiles analysed: there was virtually no dune erosion (as little as 0.3 m) in the accumulative section (kilometre 422), with a beach 3.2 m high, whereas there was a dune erosion of up to 6 m in the section (kilometre 416) with a beach height of 2.8 m AMSL, in the eastern part of the spit (Fig. 6, 2012). The Andrea storm surge was advancing from WSW to N, and thus erosion was more severe in the section exposed to waves from that direction (kilometre 416). The differences were also reflected in the magnitude of beach erosion, whereby up to 3.5 m³ and 12.6 m³ of sediment were lost in the western and eastern parts of the spit, respectively. The extent of erosion depended on wave direction and pre-surge beach height. The dunes were gradually rebuilt throughout 2012 and 2013, and embryo dunes 2.5–3.2 m AMSL high were formed in the upper beach.

In December 2013, a strong hurricane named Xavier, with very strong westerly winds, produced a fairly small surge with H_SL = 1.04 m AMSL, which flooded the beaches along the entire sandbar. The ensuing erosion was slight in the foredunes, but the embryo dunes in the upper beach were completely destroyed and the beach itself was considerably eroded. As the waves were advancing from W–WNW, the dune with a NW exposure in the eastern part of the sandbar (kilometre 416) was eroded, its base retreating by 4.2 m. Due to very strong (up to 21 m · s⁻¹) winds, the surge affected mainly the beaches that had been subject to deflation and then abrasion (erosion of approximately 15–18 m³ · m⁻¹). Large amounts of the sediment from the beach in the western part of the sandbar were blown onto the foredune ridge and into the interdune runnel in the eastern part, enlarging them by the sediment load equivalent to the total annual accumulation.

The dunes were rebuilt in 2014. New embryo forms emerged and the upper beach began to rise again. A minor surge in December 2014 (associated with a storm named Alexandra) of H_SL = 0.85 m AMSL caused no dune erosion. The water run-up was 1.8 m AMSL, i.e. 1–1.4 m lower than the foredune base.

Surges associated with two storms in early 2015, Felix in January and Ole in February, eroded the dunes again. The maximum SL was 1.09 m AMSL. The NW waves eroded the eastern stretch of the shore (kilometre 416) and the dune retreated by 2 m.

By October 2016, the dunes and embryonic forms on the beach were rebuilt. The storm Angus in October 2016 and recurrent storms in November and December resulted in surges from different directions, with the SL rising to a maximum of 1.13 m AMSL (Fig. 6, 2016). As a result, the beaches were lowered and the embryo dunes eroded down to 3 m AMSL. As the beaches were lowered, further dune erosion ensued the January 2017 storm event (Axel).

The January 2017 storm surge associated with the Axel turned out to be one of the heaviest surges at the southern Baltic coast since November 1995. With NW to NE waves, it lasted 3 days. The maximum SL at Świnoujście was 1.42 m AMSL and the water run-up amounted to 3.6 m AMSL. The beaches, even those as high as 3.2–3.4 m, failed to protect the dunes from erosion (Figs 6 and 7, 2017). The prolonged surge (over 36 h with H_SL > 0.8 m and 28 h with H_SL > 1 m AMSL) resulted in a significant erosion of the dunes in the shore sections analysed. It was the largest erosion in the 21st century, comparable to that caused by the high storm surges of 1983 and 1995 (Sztobryn et al. 2005). The erosion resulted in the dune base retreat by 6–8 m. The high ridge shrank by 12 m³ and 7 m³ in the eastern and western parts of the sandbar, respectively. The beach profile was lowered to 1 m AMSL. The beach erosion amounted to 12–16 m³ · m⁻¹ of the shore width. The autumn storms in 2017, with an SL of 1 m AMSL, caused no dune erosion despite the low profile of the beach. From spring 2018 onwards, the beach grew in height, and sand was accumulated at the foredune base along the shore. Similarly, no dune erosion followed the storm surge of October 2018; it was only the beach that was eroded to the height of 1.7 m AMSL.

In January 2019, a storm (named Zeetje) produced a surge with an SL of 1.33 m AMSL. The Zeetje developed in a manner similar to that observed in the previous storm (Axel), from W to NNE. The SL above 1 m was persisting for 15 h. The Zeetje-associated surge resulted in another erosion of the dune slopes that were in the process of rebuilding after the January 2017 storm. The dune base was eroded down to 1–4 m (Figs 6 and 7, 2019). The beaches were significantly eroded as well, with 8.5–15 m³ sand being removed per metre of the shore width. The foredune slope in both sections (and along the entire sandbar) was being rebuilt throughout 2019 and 2020, the dune rebuilding process continuing without disturbance by numerous minor storm surges from January to late March 2020, with SLs usually up to 0.8 m AMSL (four with H_SL ≥ 0.8 m). As larger surges were absent, embryo dunes could form on the beaches from May to September 2020. These were, however, eroded by the storm surge in October 2020 (Fig. 7, 2020). With NE wind and waves, the SL reached 1.1 m AMSL (H_SL > 1 m persisted for 11 h). In February 2021, the beaches in the shore section analysed were devoid of embryo dunes. In numerous places, the foredunes were still undercut by the October 2020 surge. They began rebuilding in April 2021.

Worth mentioning are 23 minor storm surges that occurred in autumn-winter 2021/2022 along the southern Baltic coast, including 9 with H_SL ≥ 0.8 m at Świnoujście. It was the highest number of surges recorded so far, comparable to the 18 that took place in 2019/2020. The two most severe storms in late January 2022, Marie and Nadine, produced surges with an SLr as low as 0.99 m AMSL (including short surge Ida with H_SL > 1 m for 3 h in mid-January). The surges mentioned resulted in beach lowering only, but this effect is not covered in the present study.

A storm surge in the part of the Baltic Sea coast examined here is regarded as major when the water level is higher than 0.8 m AMSL (after Zeidler et al. 1995, Sztobryn et al. 2005, Wolski et al. 2014). In the period covered by this study (2000–2020), there were 62 such surges, including 11 surges with H_SL > 1.2 m AMSL and 5 surges with H_SL > 1.4 m AMSL. The detailed analysis in this study concerned only the maximum surges in each year (Table 1, Fig. 5). The wind velocity (V_w, m · s⁻¹) and direction (A_zw, 0⁰–350⁰) as well as the sea level (H_SL) during the storm surges analysed and the resulting water run-up (H_SLr) were subjected to a correlation analysis. The maximum wind velocity was fairly weakly correlated with the highest SL (r = 0.29) and the duration of a level higher than 1 m (r = 0.31). However, a higher wind velocity produced stronger waves that resulted in a higher water run-up (r = 0.49). During larger storm surges and at wind velocity above 15 m · s⁻¹, the waves can reach 3–4 m.

A significant correlation (r = 0.92) was observed between the maximum SL during a storm and the water run-up (Fig. 9). During surges with H_SL = 0.8 m AMSL, the ascent of water onto the beach reached 1.6–1.8 m AMSL. With the SL up to 1 m, the usual wind speed of 15 m · s⁻¹ generates, according to the Beaufort scale, waves up to 3 m high. The run-up of a subsiding wave at an SL of 1 m AMSL reaches 2–2.5 m AMSL. In the period analysed, such storm surges occurred at the Świna Gate sandbar at least once a year. The water run-up at H_SL = 1.2 m AMSL was up to 2.8 m AMSL. The largest surges observed, with H_SL>1.3 m AMSL, resulted in a run-up (SLr) of up to 3.8 m AMSL. These effects were recorded eight times, on average every 2 years (2002, 2004, 2006, 2007, 2009, 2012, 2017 and 2019). The shore-eroding surges analysed showed an average H_SL of 1.2 m AMSL, the maximum H_SL reaching 1.48 m AMSL (during the storm Britta in 2006). However, the SL at the gauging site (the Świnoujście harbour) is not necessarily identical with that at the sandbar section examined. Therefore, generally, the run-up is more important than the SL for assessing the extent of storm-surge-caused dune erosion. During each maximum surge, the run-up was up to 3.2–3.8 m AMSL, as could be inferred from the location of organic matter layer in the sediment as well as from a direct observation of the surge itself.

Fig. 9

The maximum SL during a storm and duration of H_SL > 1 m AMSL were strongly and significantly correlated (r = 0.87). During smaller and shorter surges, the duration was in the range of 2–5 h. With H_SL = 1.2 m, the duration of H_SL > 1 m AMSL averaged 10 h. During the period analysed, there were six storm surges with H_SL > 1 m AMSL lasting more than 15 h (with a maximum of 27–32 h). Typically, such surges were produced by storms lasting 2–3 days and resulted in the heaviest beach and dune erosion (2004, 2006, 2007, 2009, 2012 and 2017). Some surges were long-lasting (2004, 2006 and 2017), some were double (e.g. 2009 and 2012) and others were alternating (2007 and 2016). The period analysed also witnessed surges with lower SLs, which eroded the beach only (2001, 2011, 2015 and 2020). The longer the duration of a high-level surge (or the higher the number of their occurrences in a given period), the more extensive was the beach (ΔQ_be) and dune (ΔQ_fr) erosion (2006/2007, 2009, 2011/2012, 2016/2017). In November 2006 and January 2007, there were four surges with H_SL > 1 m AMSL. The high SLs persisted from November 2006 through the end of January 2007. These surges, owing to their occurring in a sequence, resulted in a more extensive erosion than that produced by a single large storm. As a result, the dunes retreated further inland and were rebuilt at a higher altitude (Fig. 2).

A single larger surge was capable of removing more than 10–12 m³ of sediment from the beach. The erosion was weaker, 6–10 m³, during shorter surges, even at a high SL (e.g. in January 2002, October 2020). One or several surges with a lower SL, H_SL < 1 m AMSL, could erode away 5–8 m³ of the beach sediment. The recurrence of lower surges (January–February 2020 or November 2001) posed a weaker threat for the shore. There was no strong relationship (r = 0.24) between the number of surges and the beach erosion observed. The extent of erosion depends on a combination of factors: the SL and run-up, storm duration, wind effect and wave propagation angle with respect to the shore. The duration of storm surges with H_SL > 1 m AMSL was found to be strongly associated with the extent of shore erosion (r = 0.66 for the decrease in the beach volume, r = 0.71 for the decrease in the dune volume, ΔQ_be; the correlation coefficients are averages for the two shore sections examined). The longer the duration of a surge with H_SL > 1 m AMSL, the more extensive was the beach and dune erosion. Moreover, the higher the SL, the more extensive was the erosion of the beach and dune (Table 3).

The rate of erosion could be also related to the time during which the SL was rising to its maximum. A more rapid rise in the SL results in a more extensive erosion, as observed during the storm Axel in January 2017. The SLr to 1.4 m AMSL observed then was the fastest in the 21st century. The analysis of surge duration and the SL as well as their recurrences allowed the distinguishing of several types with respect to the shore erosion (Fig. 10):

– a very low surge (H_SL > 0.5 m AMSL) eroding the lower beach only,

Fig. 10

As already mentioned, the study involved surveying two opposite types of beach, one with a NW and the other with a NE exposure. Based on previous measurements, it was assumed that a shoreline perpendicular to the incoming wind and wave would be more eroded. This was usually the case. However, due to the changing wave direction resulting from the change in wind direction, the correlation was not very strong. No significant correlation (r = −0.22 to 0.37) was found between the wind-wave direction and the extent of erosion. A greater correlation (r = 0.23–0.59) was observed with respect to the erosion extent, as against the wind and waves varying between NW and NE. The dune was eroded by waves hitting the shore at an angle, after the beach had been lowered. Erosion on the two opposite shore sections was mostly observed when the wave direction was perpendicular to the shore.

Comparison of the beach morphology before and after a storm surge showed the beach to have been levelled off after the storm (Fig. 11A). The beach on both the accumulative and erosive sections was levelled by an average of 0.6–1.8 m. All the concave (beach lagoons) and convex (coastal banks, embryo dunes in the upper beach) forms were destroyed. Figure 12 illustrates the relationship between the change in beach parameters versus the SL and run-up. The correlations proved to be strong and significant (r = 0.66 for H_SL vs. ΔQ_be; r = 0.70 for H_SLr vs. ΔQ_be; the correlation coefficients shown are averages) (Tables 2 and 3). During storm surges with a lower SL, only the lower part of the beach was eroded, the erosion not exceeding 4 m³ (Tables 3 and 4). The erosion affected the entire beach as well as the dune base where the beach was lower due to, for example, previous storm surges. During surges with H_SL < 1 m AMSL, the beach was lowered down to 1–1.3 m AMSL. The amount of the beach sediment removed then (ΔQ_be) reached 5–10 m³ (Fig. 11C). During the heaviest storms with H_SL > 1.3 m AMSL, the water run-up was 3.6 m, the entire beach up to the foredune base being eroded. Such single surges resulted in a considerable removal of the beach sediment, 10–15 m³ (ΔQ_be). The beach sand removal (ΔQ_be) was strongly and significantly correlated with the maximum SL (H_SL) and the run-up (H_SLr) (an average r = 0.66 for H_SL vs. ΔQ_be; r = 0.70 for H_SLr vs. ΔQ_be 0.70) (Fig. 12). A weaker erosion is observed on the accumulative part with a wider beach (r = 0.55 for H_SL vs. ΔQ_be), the erosive part showing a stronger sediment removal (r = 0.79 for H_SL vs. ΔQ_be). Although the beach width was observed to generally increase after surges, the beach width changes (ΔW_be) has a weak relationship (r = 0.01 to 0.10) with storm and surge parameters. The beach width was almost identical (r = 0.85) before (W_be B) and after (W_be A) the surge (Fig. 11B, C). After a few storms, the beach width was narrower by up to 9 m, whereas other storms resulted in the beach being wider by up to 8 m. Such small changes are typical of the Świna Gate Sandbar where the beach is usually wider and higher than in other sections of the Polish coast. When the SL decreases gradually, the water removes more sediment from the dune to form a new beach after the storm. The beach formed that way is wider, often with a deep lagoon and an extensive coastal bank, previously an underwater bar formed by near-shore currents.

Fig. 11

Fig. 12

The beach slope profile and the beach width after a storm are shaped by the rate at which the waves subside (the run-up). In addition, the post-storm beach width depends on the SL at which the beach width and slope were measured. It was 0.3 m to −0.2 m either above or below the mean, which influenced the measurement. For this reason, the post-storm beach width, especially in typical accumulative sections, is not a reliable indicator of shore erosion. It is usually related to the post-storm SL, and the height of the beach in its upper part is more important.

When a storm was subsiding rapidly, i.e. there was a rapid decline in the wave height and SL, the beach was usually low, flat and slightly inclined. At a strong wave action, the water would wash away larger amounts of sediment from under the dune. The rapid sediment removal from the beach resulted in a significant decline of the beach height in its upper part. Then, after the storm, the beach height would not exceed 1 m AMSL, and the beach would be narrow and steeply inclined towards the sea. Also, the foredunes showed the formation of the highest abrasive slopes.

After the heaviest storms, the beach was widened seaward because of a large amount of sediment from the eroded dunes that was deposited to form a flat surface covered with ripplemarks, resembling a tidal plain. At a low SL, this post-storm plain was 10–30 m wide. Its formation on the shore was a result of long-term beach submergence during heavy storm surges and the post-storm recovery (as observed in 2002, 2007, 2012 and 2017). During a storm with a very high SL, the beach became shallow near the shore where ripplemarks and an underwater shoal were formed. After the storm, the shoal turned into a coastal bank. The sediment eroded from the dunes was accumulated at their foot and filled in any irregularity of the beach surface. Convex forms (embryo dunes) were eroded and depressions were filled with the sediment to form an even beach surface, inclined seaward. The beach width and height depended on the rate of the sea-level decline. A poststorm plain was formed where a large amount of sediment from the eroded dunes was deposited in the breaking wave zone during the rapid wave retreat and SL drop. This resulted in sediment accumulation, which translated into an increase of the beach width and the formation of a post-storm plain covered by wave ripplemarks. When the SL was rising again, the plain would be flooded, with a part of the beach being eroded and rebuilt. Table 3 shows the average and maximum values of beach erosion during surges with selected SLs in the accumulative and erosive shore section, while changes in beach parameters due to the storm surge-related erosion are detailed in Table 4.

Figure 13 illustrates changes in dune parameters: dune retreat (Fig. 13A), sand volume removed (Fig. 13B) and erosion per square metre (Fig. 13C), over 21 years in a selected accumulative (kilometre 422) and erosive (kilometre 416) shore profile. Correlations between the magnitude of dune erosion, the retreat of dune foot (ΔP_fr) or crest (ΔK_fr) and the change in the volume of dune-building sediment (ΔQ_fr) on the one hand, and the SL (H_SL), storm surge duration, including duration of surges with H_SL > 1 m (T_SL), and the run-up (H_SLr) on the other, were explored (Tables 2 and 3; Fig. 14). The foredune retreat increased with the SLr (Fig. 14A). The largest removal of dune sediment occurred in both shore sections after the strongest surges in 2002, 2004, 2006 and 2007. In 2002 and 2004, a heavier erosion was related to the still insignificant beach height. The higher the beach prior to the surge (H_be B), the less likely dune erosion (ΔQ_fr, ΔP_fr) was to occur. This the reason why dune erosion in recent years, even during strong surges (2012, 2017, 2019), was weaker on accumulative shores. November 2006 and January 2007 saw some of the strongest storm surges with a maximum sea-level rise of almost 1.5 m AMSL. They resulted in a complete erosion of the foredune along the sandbar. For several days in January 2007, the sea water flowed through the foredune ridge that had been eroded and lowered earlier, in November 2006. Through the wide storm gates, the water approached the older dune ridge, eroding the stabilised base of the interdune runnel. Washover fans were formed where the sediment was deposited in the runnel. Dune erosion was observed again on the accumulative shore during a long surge with a high water level (associated with the storm Axel in January 2017).

Fig. 13

Fig. 14

When an autumn–winter season witnessed only one surge event, the dunes were quickly reconstructed. This was also the case after a storm with a higher SL. When there were several surges during the storm season, dune erosion was larger (2001/2002, 2006/2007, 2016/2017). It took approximately 2 years for the dunes to be rebuilt. After a succession of strong surges occurring during a short period of time, the foredune was destroyed almost completely and was rebuilt deeper in the runnel, at a higher elevation (2002/2003 and 2007/2008). This exemplifies a foredune ridge retreat as a result of intense erosion, which was described by Carter et al. (1990). The intensely eroded or completely destroyed dune was rebuilt in the interdune runnel, which may happen only on the accumulative section of the shore.

The dune sediment removal (ΔQ_fr) was primarily correlated with the dune heights (H_fr). The higher the eroded dune, the greater its negative sediment budget. The retreat of the dune base (ΔP_fr) and a change in dune volume per square metre (ΔQ_fr · m⁻²) proved to be more important indicators of dune erosion; they were correlated with the beach height before the storm (r = 0.53 for H_beB vs. ΔQ_fr · m⁻²). It may be assumed that dunes on the southern Baltic shores featuring beaches higher than 3 m AMSL would be protected from erosion during typical storm surges. Such a correlation was evident when changes in the dune volume along the sandbar, caused by the same event, were compared. The beach height is a more important obstacle to erosion on the accumulative shore (r = 0.48 for H_beB vs. ΔP_fr,) than in the erosive part (r = 0.62 for H_beB vs. ΔP_fr), where, during an average surge, the beach is mostly flooded. Where the beaches are still at least 3 m high as a result of aeolian accumulation (kilometre 420–421), the dunes are not eroded during a surge (except the heaviest events with a run-up H_SLr > 3.8 m AMSL). Embryo dunes are formed most often in such places. Their growth in height is accelerated by vegetation (growing out from root remnants left by the previous surge). Hills 3–3.5 m AMSL high are formed and protect the foredune from abrasion. During a major storm surge in January 2012 (associated with the storm Andrea), the dune in the accumulative shore section was protected from erosion by the beach with a height equal to the maximum water run-up (H_SLr = 3.5 m AMSL). Where the upper beach was lower than 2.5 m AMSL, the foredune ridge was eroded by most surges with H_SL > 1.2 m AMSL. On lower beaches, the surge water reached the dune slope faster, and thus it took longer for erosion to occur.

The dune retreat rate was also influenced by vegetation which stabilises the ground (this was observed during a surge, but not measured). At densely vegetated sites, plant roots inhibited erosion and thus the ridge retreat, whereas at other places, the ridge was washed away faster and to a greater extent. Sparsely vegetated dunes were prone to the formation of storm gates facilitating water advance into the interdune runnel. This effect was also observed, but not measured.

After the surge, loose sediment continues to be gravitationally removed from the disturbed dune slope. The so-called tension cracks are then formed, leading to the detachment of whole sediment lumps from the slope. Landslides on the slope decrease the sediment volume of the dune and increase the sediment volume of the beach. Table 3 presents average and maximum values of dune erosion during surges with various SLs in the accumulative and erosive sections of the shore. Table 4 shows changes in dune parameters after storm-surge-caused erosion.

Each storm surge in the area of study is triggered by strong (about 15–20 m · s⁻¹) winds. Before the water starts flooding the beach, the sediment is blown away from it. Storms with winds advancing at an angle or perpendicularly to the shore blow the sediment off the beach onto the dune ridge and behind it. After each such surge, aeolian sediment accumulation can raise the dune and the runnel by up to 10 cm and 5–6 cm, respectively. This is important from the standpoint of dune formation dynamics. As a result, although the dune base is undercut by the incoming waves, the sediment budget on the dune crest can be positive. Accumulation on the foredune ridge was observed after the storms of 2003, 2004, 2009 and 2012 (mainly on the accumulative shore represented by the profile on kilometre 422). The largest accumulation occurring in a short period of time was observed during the storm Xavier of December 2013, with a wind velocity of up to 18–24 m · s⁻¹. Such a westerly wind blowing along the shore produced sediment aeolian accumulation of 20–25 cm on the foredune in the eastern part of the area (kilometre 416). Usually, the accumulation is not higher than 10 cm on the dune top. Aeolian accumulation during (or just before) surges in the western part was larger because the source beach, from which the sand was blown away just before flooding, was wider. The accumulated fresh sand layer was more than 0.1–0.15 cm thick on the dune and behind it, which amounts to an average 0.4 m³ of sand (maximum of 1.2 m³). Accumulation occurs only before the surge and during low surges. In the latter case, aeolian deposition is observed on the windward slope. The stronger the wind, the higher the likelihood of accumulation on the lee slope. The correlation between the wind velocity (or direction) and the aeolian accumulation during surges proved to be weak (r = 0.14–0.36). Accumulation between grass shoots was accidental only and hardly visible on the profiles. In addition, some of the accumulated sediment was eroded by storms, and therefore such data are neither presented nor illustrated.