Relationships between the Baltic Sea ice extent and ice parameters in the sheltered basins of the southern Baltic coast

Values of ice parameters (for example ice thickness, ice season duration, or the probability of ice occurrence) in the Baltic Sea are affected by the large latitudinal extent of the basin, as well as distinct climatic differences between its northern and southern regions. Ice occurrence probability increases toward the north. The probability varies from about 30% along the southern coast, e.g. in the Hel, Arkona or Westermarkelsdorf area, up to even 100% in the northern part of the Baltic Sea, i.e. in the Gulf of Bothnia (Janérus & Jansson 1982; Jevrejeva et al. 2004). The Baltic Sea ice cover (i.e. the basin area covered by ice) varies on a year-to-year basis, largely due to the type of atmospheric circulation prevailing during a given winter – usually the North Atlantic Oscillation (NAO). Since the 1980s, however, the ice extent is comparable to, smaller, or definitely smaller than in the period 1720–1995 (Omstedt & Chen 2001; Vihma & Haapala 2009; Haapala et al. 2015). Forecasts indicate that by the end of the 21st century, the ice extent in the Baltic Sea will be limited to the northernmost parts of the Gulf of Bothnia.

The ice conditions in the northern regions of the Baltic Sea are much more severe (longer ice season and thicker ice cover) than in the south. While the ice along the southern Baltic coast disappears in early March, it continues to develop in the north (Leppäranta & Myrberg 2009). Along the northern Baltic coast, the first ice occurs on average before 15 November, and disappears after 15 May. Due to longer and more severe winters, the ice season in the north lasts more than 180 days and the maximum ice thickness in the season is larger than 70 cm (Lepparänta et al. 1988). In Kemi (Finland) in the 20th century, the last ice melted on average on 21 May, ranging from 16 April to 27 June (Leppäranta & Myrberg 2009).

The ice conditions along the southern Baltic coast are considerably milder and more dynamic. Ice usually appears at individual research stations in January, while in the Hel region it is not observed until early February. It tends to disappear in late February through early March. The average ice season duration, i.e. the number of days from the first to the last day with ice, ranges from 10 to 30 days for most stations along the southern Baltic coast, and the mean maximum ice thickness in the winter season for the period from 1954/1955 to 1999/2000 ranges from 5 to 12 cm (Girjatowicz 2007). This variation in ice conditions is mainly due to the seabed bathymetry and the river ice influx.

The Baltic Sea ice conditions are characterized by seasonality and large annual variability in the ice extent. During very severe winters, the entire Baltic Sea is often covered by ice, as evidenced by winters of 1939/40, 1941/42, and 1946/47 (Seina & Palosuo 1993; Koslowski & Loewe 1994; Omstedt et al. 2014). In 1987, 96% of the Baltic Sea was covered by ice. Even during such severe winters, the sea ice in the Gotland Sea was very thin and fragile, i.e. easy to break (Leppäranta & Myrberg 2009). During very mild winters, however, only the northernmost areas of the Baltic Sea and the bays are covered with ice. This is exemplified by winters of 1924/25, 1988/89, and 2007/08, when ice developed only in the coastal zones of the Gulf of Bothnia and along the northern coasts of the Gulf of Finland and the Gulf of Riga (Betin 1957; Seina & Palosuo 1996; Styszynska 2010). In recent years, the 2014/2015 season witnessed a very mild winter. The ice appeared in the Gulf of Bothnia in mid-November, but December and January were exceptionally warm. The last ice melted in the first half of May, approximately two weeks earlier than usual. The ice season was also unusually short in 2015/16, when the Gulf of Finland was completely free of ice already in mid-April (Janecki et al. 2018).

In the coastal lagoons of the southern Baltic Sea, the ice conditions are milder than in the northern Baltic regions, but are more severe than in the southern unsheltered waters of the Baltic Sea (Preobraženskij 1960; Janérus & Jansson 1982; Girjatowicz 1990). Differences in ice conditions between sheltered and unsheltered basins of the southern Baltic are influenced by both climatic and morphometric (bathymetric) factors. On the one hand, unsheltered basins located further to the north experience more severe winters, but on the other hand, surface water cooling, conducive to more intense ice development, progresses faster in shallower basins with a higher exposure coefficient (i.e. surface to water depth ratio). Further, sheltered basins are less prone to water mass motion, e.g. due to wind-driven waves or currents. In the open sea, ice formation and ice durability are hindered because of frequent winter storms associated with the west to east migration of low pressure systems (Brayshaw 2005).

The first ice in the lagoons appear already in December. In the Vistula Lagoon, it occurs at the beginning of December, and in the Puck Lagoon and in the Szczecin Lagoon – at the end of December. The last ice disappears from the lagoons in March: in the middle of March in the Vistula Lagoon, and at the beginning of March in the Puck Lagoon and in the Szczecin Lagoon. The average ice season duration clearly increases toward the east. In the Szczecin Lagoon, the ice season lasts for 73 days, in the Puck Lagoon – 77 days, and in the Vistula Lagoon – 104 days. The maximum ice thickness in the season is: 18, 22, and 31 cm, respectively (Girjatowicz 2007). The lagoons in questions considerably vary in terms of ice conditions. This is particularly evident when the Szczecin Lagoon is compared to the Vistula Lagoon, as the ice conditions in the latter are considerably more severe (see Łazarienko & Majewski 1975; Majewski 1980; Szefler 1993). In these basins, the dominant form of ice is permanent ice cover (Malinski 1971; Zorina & Malinski 1975; Szefler 1993).

In the present study, the inner part of Puck Bay (i.e. the western part of the Gulf of Gdansk) is referred to as the Puck Lagoon due to considerable shoaling of the Seagull Sandbar (depth range: from 0.9 to 1.6 m) and the Rewa Cape projecting toward the sandbar. These features together clearly distinguish this part of the gulf from the remainder, effectively making it a lagoon with respect to water circulation, biota, temperature and ice conditions. There is a dredged passage between the sandbar and the cape, which enables vessels to sail from Gdynia to Puck. The term “Puck Lagoon” was used, among others, by Nowacki (1976), Musielak (1983), and Majewski (1990). The Puck Lagoon is also termed “Little Puck Bay”, or “Inner Puck Bay” in the literature.

Ice conditions in the coastal lakes of the southern Baltic Sea are milder than in the northern Baltic regions, but are more severe than in the southern part of the Baltic (open sea). The first ice appears on average in early December and disappears in early March. The average ice season duration ranges from 85 to 100 days and the mean maximum ice thickness is 22–24 cm (Girjatowicz 2003; Skowron 2009). For the open southern Baltic, these parameters range from 10 to 30 days and from 5 to 12 cm, respectively. The coastal lakes are covered with permanent ice even during very mild winters (Girjatowicz 2003; Choinski et al. 2014). The ice conditions are slightly more severe in the eastern part of the coast, although these lakes show little variation in the ice conditions (Girjatowicz 2003; Skowron 2009; Ptak 2013). The eastward increase in the severity of ice conditions is even more pronounced for the other lakes of northern Poland, especially with regard to the date on which the last ice disappears, the duration of the ice season and the maximum ice thickness (Marszelewski & Skowron 2009).

The fact that the ice conditions in the coastal lakes are more severe than in the open southern Baltic results from the shallow depth of these lakes – the maximum depths of Jamno, Bukowo, Gardno and Łebsko lakes are only 3.9, 2.8, 2.6 and 6.3 m, respectively (Janczak 1997; Choinski 2006).

Although values of the ice parameters in the lakes and coastal lagoons differ from the corresponding values observed for the open Baltic, it can be assumed that there are statistically significant correlations between them. The objective of the present work is to analyze the relationships between the ice parameters (ice season duration, number of days with ice, and maximum ice thickness for a given winter) in the southern Baltic lagoons and the coastal lakes, and the maximum annual ice extent in the entire Baltic Sea expressed in km² (hereafter abbreviated as MIB – Maximum Ice Extent in the Baltic Sea). The MIB parameter, which refers to the ice-covered area of the Baltic Sea as a whole, is a good descriptor of the Baltic Sea ice conditions in a given winter. An additional objective is to emphasize the variation in the strength of connections between individual sheltered basins and to characterize the factors affecting this variation. It is likely that the similarity of climatic conditions, especially the similarity between the climatic trends observed in the entire Baltic Sea basin, will impact the strong relationships between the ice conditions in the lagoons and coastal lakes, and the Baltic Sea ice cover (MIB). Strong and statistically highly significant correlations can be used to estimate values of the selected ice parameters. In the case of discontinuous data series, due to lack of any given observation, it will be possible to estimate a variable for a given winter based on the observed relationships.

Archive data on MIB for the period 1961–2018, expressed in km², were obtained from the Finnish Institute of Marine Research (FIMR) database. Data for the period 1720–1995 were published by Seina & Palosuo (1996). The MIB was used in this study as a parameter characterizing the ice conditions in the Baltic Sea, given that the ice cover extent best characterizes the ice conditions in the Baltic Sea, even though the state of sea ice is basically described by the extent, thickness and drift directions (Haapala et al. 2015).

Data on the ice parameters for the four largest southern Baltic coastal lakes: Jamno, Bukowo, Gardno and Łebsko, and for Szczecin, Puck, and Vistula lagoons come from the Maritime Branch of the Institute of Meteorology and Water Management – National Research Institute (IMGW-PIB) in Gdynia (Poland). The time series for the lakes covers the period from 1960 to 2018, and for the lagoons – from 1946 to 2018. Ice observations of individual basins were performed in the following fixed points: Uniescie (Lake Jamno), Bukowo Morskie (Lake Bukowo), Gardna Wielka (Lake Gardno), Izbica (Lake Łebsko), Trzebiez (Szczecin Lagoon), Puck (Puck Lagoon), and Tolkmicko (Vistula Lagoon; Fig. 1). The availability of continuous time series of data on the ice occurrence during the ice season (i.e. ice occurrence on nearly annual basis) was taken into account when selecting the coastal lakes and lagoons. The basins selected for the analysis are characterized by similar morphological and climatic conditions of ice formation. The main morphometric and bathymetric data for these basins are presented in Table 1.

Figure 1

The analysis started with an estimation of the probability of ice occurrence in the coastal lagoons and coastal lakes based on the principles used in the works by Jevrejeva et al. (2004), Leppäranta (2014) and Karetnikova et al. (2017). The sea ice occurrence probability for the studied basins and its standard deviation were calculated using the following equations (Leppäranta 2014):

where:

p– probability of ice occurrence,

N– the total number of seasons (years),

n– the season number,

I(n) – binary variable,

I(n) = 0 – season n is ice free,

I(n) = 1 – season n is with ice,

SD – standard deviation of the estimator.

Three ice parameters were selected for the detailed analysis: the number of days with ice (N), i.e. days during which any form of ice was observed; the ice season duration (S), i.e. the period from the day when the first ice was observed to the day when the last ice disappeared, but including possible ice-free periods between days with ice (expressed in days); and the maximum ice thickness (H) in a given winter. Mean values of the ice parameters, their standard deviation and ranges (extreme values of the ice parameters as well as the season and the basin of their occurrence) in the groups of coastal lagoons and coastal lakes are tabulated in Table 2. Mean and maximum values of these ice parameters for both the selected coastal lagoons and lakes and are tabulated in Table 3.

Correlations (especially those where outliers were observed in regression plots) were interpreted based on mean monthly and mean winter (December through February) air temperature values from Szczecin (for the western Baltic coast), Ustka (central coast), and Gdynia (eastern coast) for winters from 1946/47 to 2017/2018. On the basis of these data, the deviation from the mean air temperature for the whole study period was calculated for each winter.

Relationships between the selected ice parameters for the studied basins and the MIB were analyzed using the correlation and regression method. Linear regression equations (y = ax + b) were determined. The ice parameters for the lakes and lagoons (N, S and H) were assumed as dependent variables (y), while MIB values were treated as independent variables (x). Ice parameter relationships were also calculated between individual lagoons (within the group of three coastal lagoons), individual lakes (within the group of four coastal lakes), and between average values for the lagoon group and average values for the lake group.

The strength of these correlation relationships was examined using the correlation coefficient (r) and the determination coefficient (r₂ 100%), and their statistical significance was evaluated using the Fisher–Snedecor test (Łomnicki, 2014; Time Series Analysis 2010). The correlation coefficients estimated in this study were high, and all proved to be statistically significant at p< 0.001. For linear relationships presented in the Cartesian coordinate system, a 99% confidence interval was applied.

In the sheltered basins analyzed in the present work, the ice occurrence probability is very high (ice occurs during nearly every season), considerably higher than along the open coast – the ice occurrence probability in Ustka and Kołobrzeg (located in the close vicinity of the studied basins) was 0.58 (Jevrejeva et al. 2004). The ice occurrence probability estimated for the Szczecin Lagoon is comparable to the value estimated by Jevrejeva et al. 2004, calculated on the basis of data from the period 1900–2000.

To date, relationships between ice conditions in the sheltered basins of the southern Baltic coast and the open Baltic Sea have not been studied. Previous studies focused on ice parameter correlations between study sites located within individual basins, but also between individual basins along the southern Baltic coast.

The correlation coefficients established in this study (between MIB and S, N and H, respectively) are weaker than those determined by previous researchers for the ice parameters (S, N and H) between the study sites located within individual basins, or between basins of the southern Baltic coast.

The strongest relationships are reported between sites located in the Szczecin Lagoon between Trzebiez and Podgrodzie stations. Correlation coefficients for these sites were usually above 0.90 (Girjatowicz et al. 1995). Strong correlations at these sites are due to their location within the same basin and in close proximity to each other. For this reason, ice conditions at these sites are shaped by similar climatic and hydrological factors.

The strength of the correlation between the ice parameters was already studied for Jamno and Łebsko lakes (Girjatowicz 2003). The correlations reported by Girjatowicz (2003) were also stronger than in the present work. The highest correlation coefficients between these lakes exceeded 0.90 and were observed for the number of days with ice (r = 0.96), the number of days with fast ice (r = 0.97) and for the maximum ice thickness (r = 0.90). The lowest correlation coefficients between Jamno and Łebsko lakes were determined for the number of days with floating ice (0.60) and the first ice appearance date (r = 0.71). All correlations were statistically significant at p< 0.001. The correlations between the MIB and the ice parameters for the southern Baltic sheltered basins are weaker than between the individual sheltered basins of the southern Baltic. This is obviously due to the varying degree of climatic and morphometric (bathymetric) similarities. Nonetheless, the differences in the strength of the correlations are lower than would result from the latitudinal differences. The weaker relationships determined for the number of days with floating ice were associated with higher fluctuations in the presence of floating ice, largely depending on anemometric conditions, i.e. wind direction regulating the ice inflow and outflow. Similarly, weaker relationships concerning the first ice appearance date may be associated with high variability in this parameter, especially along the southern Baltic coast (cf. Sztobryn et al. 2012). In the southern Baltic basins, the first ice may occur throughout the winter, i.e. from November to March (Girjatowicz 2007).

Strong ice parameter correlations exceeding r = 0.90 are also observed between such basins as the Szczecin Lagoon or the Vistula Lagoon, and between some river mouths, straits and harbors (Girjatowicz 2011). The strongest correlations were determined for the number of days with ice between Swibno (Vistula river mouth) and Tolkmicko (Vistula Lagoon, r = 0.94), the ice season duration between Swinoujscie and Ueckermünde harbors (Szczecin Lagoon, r = 0.94), and the maximum ice thickness between Puck (Puck Lagoon) and Tolkmicko (r = 0.92). The weakest relationships, mostly characterized by correlation coefficients lower than 0.90, were found for the ice season duration (Girjatowicz 2011). The lowest correlation coefficients were found between the Swinoujscie harbor and Trzebiez (Szczecin Lagoon, r = 0.81), Ueckermünde and Trzebiez (Szczecin Lagoon, r = 0.83), and between Baltiysk (Strait of Baltiysk) and Krasnoflotskoye (Vistula Lagoon; r = 0.84). These weaker relationships are influenced by relatively high variability of ice conditions, associated with coastal currents at sites located at the river estuaries, i.e. in Swinoujscie, Trzebiez and Ueckermünde, and within straits – in Baltiysk.

The low strength of the relationships concerning the ice season duration is associated with the observation that the ice season may in fact comprise several periods with the presence of ice separated by ice-fee periods. The variability of these periods is not only due to fluctuations in alternating cooler (T< 0°C) and warmer periods (T > 0°C), but may also be caused by dynamic factors such as ice inflow and outflow. The number of such ice periods during the ice season can be up to six in the Vistula Lagoon (Łazarienko & Majewski 1975), and even up to eight in the Szczecin Lagoon (Girjatowicz 2011).

On the local scale in the Baltic Sea basin, the severity of winter is connected with the prevailing type of atmospheric pressure systems. The prevalence of lows during winter favors mild ice conditions in a given basin, while the prevalence of highs is conducive to severe ice conditions (Omstedt et al. 2004). This explains why in the sheltered basins of the southern Baltic, during some winters (represented as outliers in the regression plots), ice conditions are milder or severer than the Baltic average. For instance, during the unusually severe winter of 1995/96, air masses from the Arctic moved from the Norwegian Sea to the Szczecin Lagoon and the Pomeranian Bay (SW Baltic Sea) as early as mid-November, and the high pressure belt covered Western and Central Europe including the southern Baltic Sea, which caused the air temperature in Szczecin to drop to −7°C. The anticyclone in December contributed to a decrease in the minimum temperature in Szczecin to −13.3°C, while on the coast the temperature was about 1°C higher. In mid-March, the high pressure system over Scandinavia and Central Europe caused the air temperature over the Szczecin Lagoon to drop to −11°C. Anticyclonic systems and the inflow of Arctic air masses over the southern Baltic contributed to an unusually long ice season and more intensive development of ice during that winter (Schmelzer et al. 1996).

Based on the analysis of relationships between the maximum annual sea ice extent in the Baltic Sea (MIB) and the ice parameters recorded for the sheltered basins of the southern Baltic Sea (first ice appearance date, last ice disappearance date, number of days with ice, ice season duration and maximum ice thickness in season), we conclude the following: