Show Summary Details

Page of

Printed from Oxford Research Encyclopedias, Climate Science. Under the terms of the licence agreement, an individual user may print out a single article for personal use (for details see Privacy Policy and Legal Notice).

date: 18 April 2024

History and Future of Snow and Sea Ice in the Baltic Seafree

History and Future of Snow and Sea Ice in the Baltic Seafree

  • Matti LeppärantaMatti LeppärantaUniversity of Helsinki

Summary

The physics of the ice season in the Baltic Sea is presented for its research history and present state of understanding. Knowledge has been accumulated since the 1800s, first in connection of operational ice charting; deeper physics came into the picture in the 1960s along with sea ice structure and pressure ridges. Then the drift of ice and ice forecasting formed the leading line for 20 years, and over to the present century, ice climate modeling and satellite remote sensing have been the primary research topics. The physics of the Baltic Sea ice season is quite well understood, and toward future ice conditions realistic scenarios can be constructed from hypothetical regional climate scenarios.

The key factor in climate scenarios is the air temperature in the Baltic Sea region. The local freezing and breakup dates show sensitivity of 5–8 days’ change to climate warming by 1 °C, while this sensitivity of sea ice thickness is 5–10 cm. However, sea ice thickness and breakup date show sensitivity also to snow accumulation: More snow gives later breakup, but the thickness of ice may decrease due to better insulation or increase due to more snow-ice. The annual probability of freezing decreases with climate warming, and the sensitivity of maximum annual ice extent is 35,000–40,000 km2 (8.3%–9.5% of the Baltic Sea area) for 1 °C climate warming. Due to the large sensitivity to air temperature, the severity of the Baltic Sea ice season is closely related to the North Atlantic Oscillation.

Subjects

  • Modeling
  • Climate of the Baltic Sea Region

Introduction

The Baltic Sea belongs to the seasonal sea ice zone (SSIZ). The southern edge of this zone crosses subarctic marginal and semienclosed seas from the Baltic Sea to the Gulf of St. Lawrence and Hudson Bay in the Northwest Atlantic and the Sea of Okhotsk and Bering Sea in the North Pacific. Climate variations show up drastically in the ice conditions in the SSIZ. The length of ice season varies by months and the annual ice extent varies by several latitude degrees. Thermally grown ice is at most about 1 m thick.

On the northern coast of the Baltic Sea (66°N), the length of ice season is 5–7 months, whereas in the south (54°N), ice does not form every year (Jevrejeva et al., 2004; Leppäranta & Myrberg, 2009). Sea ice—or brackish ice—is one of the key physical characteristics of the Baltic Sea and has a significant role also in the North European climate system (Figure 1). The ice cover buffers the surface water temperature to the freezing point. The impact of ice melting is significant for the surface water salinity as it starts up the stable spring stratification with further consequences to the spring bloom (Kari et al., 2018), but the influence of the salt flux during freezing of the brackish sea water is too weak to force convection across the halocline.

Figure 1. The Baltic Sea, with the northern part of the Gulf of Bothnia (top), the Gulf of Finland (right), and the Gulf of Riga (center) covered in ice. This image was acquired on March, 19, 2010, by Envisat’s medium resolution imaging spectrometer instrument, working in full resolution mode to provide a spatial resolution of 300 m.

Source: © European Space Agency, CC BY-SA IGO 3.0.

Long-term data of the Baltic Sea ice season show very large variability. The local freezing and melting dates vary by 2–3 months and the local maximum annual ice thickness varies by half-meter. In very severe winters the whole Baltic Sea is covered by ice, but in very mild winters the coverage is only one-eighth of the total area. There is much historical information of the Baltic Sea ice season (Haapala et al., 2015; Jurva, 1952; Palosuo, 1953; Tarand, 1993).

The ice cover has a remarkable role in the human living conditions in the Baltic Sea region. Particularly this concerns sea traffic. Back in the history in the era of sailboats, winter shipping was cut off in the ice season, while presently 20–25 icebreakers take care for a feasible marine transportation system to all the main harbors of the Baltic Sea. On the other hand, between mainland and islands in the northern Baltic Sea there are ice roads that have had an important role in the early history. The northern part of the Baltic Sea has experienced ice-covered sea conditions annually, and therefore the Baltic Sea ice research history is largely concentrated in the north, although all Baltic Sea countries have had an operational ice service to support winter shipping. This article presents an overview on snow and sea ice in the Baltic Sea with past and projected future changes.

Rise of Sea Ice Research in the Baltic Sea

Sea Ice Monitoring for Winter Navigation

The Baltic Sea freezes over annually to one-eighth to all of its total area. Heat inflow from the North Atlantic is very weak, and the shallow mean depth of 54 m keeps the heat storage limited (Leppäranta & Myrberg, 2009; Omstedt, 1990; Palosuo, 1953). The temperatures of the freezing point and the maximum density are −0.4 to −0.2 °C and 2–3 °C, respectively, but the density maximum is weaker in brackish water than in fresh water. Therefore, a shallow fall thermocline breaks easily, and the permanent halocline at 40–80 m depth is the governing mixing depth during fall cooling. The halocline also limits the heat exchange between the surface layer and deep water.

Historically, the Baltic Sea ice cover prevented marine traffic in the sailboat era, and even thereafter severe restrictions have been caused to machine-powered ships. Only since 1972 have all of the main harbors of the Baltic Sea been operational all year. Ice charting and reporting systems were established in the 1800s in collaboration between navigation authorities and research institutions (Grönvall, 1988; Jurva, 1937; Palosuo, 1953; Strübing, 2009). They were based on fixed sites, such as manned lighthouses and pilot stations, on the coast and archipelago. In 1928 the first version of the Baltic Sea Ice Code for reporting was agreed, and in winter 1938 the First Baltic Sea Ice Week with an extensive observation program was organized (Granqvist, 1938).

In the early 20th century, observations of drift ice were quite limited. Icebreakers could report of the ice conditions in their operation area, and occasional aerial reconnaissance flights were performed that could provide a true insight into the morphology of drift ice fields. Sea ice research was focused on landfast ice where observation programs could be performed. Jurva (1937) presented a method to describe and predict the evolution of the ice season based on an idea of independent phases of the state of the landfast ice zone.

In the 1940s, the three winters of the last century occurred when the whole Baltic Sea froze over, two of them during the Second World War. Finnish Air Force captain (later professor) Erkki Palosuo acted as an ice reconnaissance pilot across the Central Baltic Sea between Finland and Germany and produced a large set of observations of ice conditions in the area. He used the data in his doctoral thesis (Palosuo, 1953) that is still a unique scientific analysis of severe winters in the Baltic Sea. After the war, aerial reconnaissance became a routine method in ice charting. An extensive Swedish–Finnish winter navigation research program was commenced in 1972 that included much work in the Baltic Sea ice science and engineering. For ice charting, satellite remote sensing methods stepped into the picture (Blomquist et al., 1976; Grönvall, 1988). Based on the Swedish and Finnish database of ice charts and reports, the first Baltic Sea ice atlas was published in 1982 (Swedish Meteorological and Hydrological Institute, and Finnish Institute of Marine Research [SMHI & FIMR], 1982).

Microwave techniques brought a new step into sea ice remote sensing due to their weather independency. Prior to the launch of the ERS-1 satellite, an extensive field program on remote sensing of sea ice by radar was performed (Askne et al., 1992; Leppäranta et al., 1992). Due to sea ice dynamics, ice conditions may change significantly for shipping on a daily time scale, and therefore intensive monitoring is required over drift ice fields (Berglund et al., 2007; Grönvall & Seinä, 2002; Strübing, 2009; Sztobryn, 2009). Satellite methods map very well the lateral features of sea ice fields, but vertical extent, that is, ice thickness, is the main problem in ice charting. Limited results in ice thickness mapping have been obtained with satellite synthetic aperture radars (see, e.g., Kozlov et al., 2020; Similä et al., 2005).

Snow Mapping for Management of Water Resources

Snow research and monitoring has been largely connected with hydrology in the Baltic Sea region, where flood prevention and regulation of water resources were the principal driving factors. Monitoring networks were set up in the late 1800s (e.g., Korhonen, 1915). Observations included snow thickness and density, which together provide the water equivalent of snow, the principal hydrological snow variable. Snow thickness is measured using snow stakes along snow courses, while density measurements are based on the mass/volume ratio of snow samples. Snow research covered snow stratigraphy, influence of topography on snow accumulation, and snowmelt. Sea ice and lake ice studies have included snow-ice since the 1960s (Palosuo, 1963, 1965).

Toward the end of the 1900s, more scientific geophysics of snow started to grow from the water resource monitoring. Also, investigations were commenced on the quality and geochemistry of snow as well as on snow ecology (Kuusisto, 1984). A snow fork was designed for the snow density and liquid water content measurement from the dielectric constant (Denoth et al., 1984), and methods were updated for snow crystal photography (Pihkala & Spring, 1985). Connected to snow monitoring, airborne and satellite remote sensing methods were developed (Kuittinen, 1988; Pulliainen et al., 1999).

Learning Properties of Baltic Sea Ice

Landfast Ice

The scale of ice thickness has long been known from practical experience since it is the fundamental quantity to estimate the bearing capacity of ice and ice forces. The thickness also largely determines the mobility of ice and the conditions for shipping (Jurva, 1937). Ice roads were important ways of traveling and transport in the past before the introduction of steamships and railroads. For ice-going ships, the thickness of ice is a key resistance factor (Enkvist, 1972; Kujala, 1996).

The Baltic Sea brackish water ice is a specific form of sea ice, and research of its physical properties was commenced in the 1950s (Palosuo, 1961, 1963). First, Palosuo (1961) showed that the crystal structure of Baltic Sea ice is similar with first-year sea ice in Arctic seas when the salinity of the parent water is more than 1–2 ppt (parts per thousand by mass). This result, confirmed later by, for example, Kawamura et al. (2001), means that the Baltic Sea ice crystals have jagged boundaries and there are brine pockets between crystal platelets. The bottom of ice has cellular structure that is effective in capturing sea water inside, and the salinity of new ice is around one-third of the salinity of the parent water (Leppäranta & Myrberg, 2009; Palosuo, 1963; Weeks & Ackley, 1986).

Salinity decreases during winter, especially during the melting period, reaching a residual level of about 0.1 ppt (Figure 2). The low salinity means that the Baltic Sea ice has much higher strength than polar first-year sea ice but still less than freshwater ice (Weeks & Ackley, 1986). During ice formation, rejection of salts from the ice lattice is a minor factor for vertical mixing, but melting of ice in spring produces a stable low-salinity surface layer as in the polar seas.

Figure 2. Evolution of salinity of sea ice along the coast of Finland in the Bay of Bothnia (Mässkär), southern Sea of Bothnia (Saggö), Archipelago Sea (Bodö), and Gulf of Finland (Porkkala).

Source: Based on Palosuo (1963).

The Baltic Sea ice cover is stratified with three types of layers: congelation ice, snow-ice, and frazil ice (Granskog et al., 2004; Leppäranta et al., 1992; Palosuo, 1963). Congelation ice is the most common type, while the snow-ice fraction can be up to half of the total thickness but can be lacking in dry winters (Leppäranta & Myrberg, 2009; Palosuo, 1963). Semiempirical ice thickness models account for the air temperature and snow accumulation for realistic results (Leppäranta, 1983a). Frazil ice forms often as the first ice layer, and one would expect to observe frazil layers also deeper down in high seas, generated in leads (Omstedt, 1985). There are media reports of frazil problems in nuclear power stations, frazil accumulation on fishing nets, and anchor ice formation.

Snow accumulation is a critical factor in the Baltic Sea ice season (Palosuo, 1963). If the sea surface freezes during snowfall, the first ice layer is frozen slush. Snow accumulation on ice slows down ice growth by insulation, but on the other hand, heavy snow accumulation causes flooding of ice and eventual snow-ice formation (Figure 3). Bare surface gives the largest ice thickness, while effective insulation can reduce the thickness down to one-half of the bare surface case. Due to the influence of snow cover on radar backscatter, detailed studies of the snow layer on sea ice were made in remote sensing programs (Askne et al., 1992; Leppäranta et al., 1992).

The presence of snow cover also reduces the transfer of sunlight that has impact on ice melting and primary production inside and below ice. Sea ice optical properties rose as a relevant question in the Baltic Sea in the 1990s in connection with ice climate modeling (Haapala & Leppäranta, 1996) and sea ice ecology (Arst et al., 2006; Ikävalko, 1997). Snow cover protects the ice from melting by its low light transmittance and thus delays the ice breakup.

Figure 3. Ice thickness in Oulu, Bay of Bothnia, calculated with climatological air temperature as a function of assumed constant daily snowfall. The solid line shows total ice thickness, and the dashed line shows the thickness of congelation ice formed of the sea water.

Several ice observation stations were founded in the 1800s in the Baltic Sea for local ice conditions. In consequence, after a few decades time series analyses were started, and they clearly showed the warming in the Baltic Sea region after the Little Ice Age. Risto Jurva (Finnish Institute of Marine Research) went through historical material and estimated the maximum annual ice extent in the Baltic Sea back to 1720. His original data were destroyed in the bombings of Helsinki in World War II, and the time series in its most original form is presented in graphical form in in Palosuo (1953). This time series has been later digitized by several authors. It accounts for area of the Baltic Sea including Kattegat, with a total area of 420,000 km2.

Ice time series papers showed significant trends for the freezing and breakup dates toward shorter ice season. Leppäranta and Seinä (1985) studied eight stations along the Finnish coast and archipelago with data starting in 1830–1910. They obtained the mean trend of 1.5 days over a decade for the freezing date (in four stations the trend was significant) and 1.2 days over a decade for the breakup date (in six stations the trend was significant). In the case of ice thickness, no systematic change was obtained that can be explained by the fact that ice growth depends on air temperature and snow accumulation. The maximum annual ice extent in 1830–1984 varied very much without any time structure, and the trend was 4,200 km2 (1% of the Baltic Sea area) over a decade. This trend was, however, highly influenced by the Little Ice Age Years in the early part of the time series.

Mobile Ice Fields Out From Landfast Ice

It has long been known among fishermen and seal hunters that farther away from the landfast ice zone there is mobile drift ice. The drift ice medium consists of ice floes with various morphological characteristics (World Meteorological Organization [WMO], 2014) as is shown in standard ice charts. These characteristics can be largely observed from ship deck or aircraft, and they have been used in ice reporting systems since the beginning of the 1900s (Grönvall, 1988; Jurva, 1937; Palosuo, 1953). Collecting information of drift ice was first based on fixed observation stations on the coast and in archipelago that were highly limited. More information was provided in ice reports by icebreakers; in the 1930s aerial ice reconnaissance was begun, and since the 1980s satellite remote sensing has provided good data, first with National Oceanic and Atmospheric Administration imagery and later on with synthetic aperture radars (Askne et al., 1992; Grönvall & Seinä, 2002).

Drift ice physics research was commenced by Soviet scientists in the Arctic Ocean (Zubov, 1945), while in the Baltic Sea that became a major issue in the 1970s (Leppäranta, 1981; Palosuo, 1975; Udin & Ullerstig, 1976). The background was the need to develop a short-term forecasting system for operational winter navigation in the Baltic Sea, when all the main harbors had been opened for all-year shipping. Then scientific and technical methods were developed to aid and guide the winter shipping in a research program funded by a joint Finnish–Swedish Winter Navigation Research Board.

Operational sea ice forecasting was commenced in the Baltic Sea in 1977 (Leppäranta, 1981). The time extent was 36 hours, and based on communication between the forecasting team and icebreakers, the most critical factors to forecast were ice drift direction with respect to coastline and ice pressure. Due to ice movement, ice situation may vary widely in a short time (Figure 4). Several field experiments on ice dynamics were performed in the Bay of Bothnia (Leppäranta, 1981; Leppäranta & Omstedt, 1990). It was shown that sea ice dynamics is similar in the Baltic Sea and the Arctic Ocean and scaled with the thickness and lateral extent of the ice cover.

Figure 4. Velocity at a drift ice station in the Bay of Bothnia. The solid lines show ice drift speed and direction, and the dashed lines show wind speed and direction.

Sea ice pressure ridges are the most difficult form of drift ice for shipping. They were investigated in detail for their structure, properties, and occurrence (Leppäranta & Hakala, 1992; Lewis et al., 1993; Palosuo, 1975). Large ridges are 15 m thick, and the record value is 31.5 m whereof 3.5 was above the water surface (Palosuo, 1975). In sea ice dynamics science, build-up of ridges represents the main sink of mechanical energy in sea ice pressure. Near the shore, ridge keels scour the bottom sediment and cause shore erosion by piling up and riding up on shore (Alestalo & Häikiö, 1975).

Sea ice dynamics is driven by winds and currents. Therefore, the atmospheric boundary layer above ice and the marine boundary layer beneath ice were widely studied for air–ice and water–ice momentum and heat exchange. The focus was in the atmosphere (Joffre, 1984; Launiainen, 1979), while very few field studies were made for the marine boundary layer under ice (Leppäranta & Omstedt, 1990).

The climate impact question gained large interest in the Baltic Sea in the 1990s. Modeling of the ice season expanded from local models and time series analyses to box models (Omstedt, 1990) and full two-dimensional numerical models (Haapala & Leppäranta, 1996). In an ice climate scenario study by Haapala and Leppäranta (1997), the prediction for a representative mean ice season in 2050 with an assumed temperature increase of 3.6 °C was the following: annual maximum ice extent 65,000 km2 (15% of the Baltic Sea area), and for the site Kemi in the north a freezing date earlier by 17 days, breakup date later by 10 days, and ice thickness lower by 23 cm (Figure 5). Thus, the sensitivity for a 1°C climate warming was 5 days for the freezing date, 3 days for the breakup date, 6 cm for the ice thickness, and 35,000 km2 (8%) for the ice extent. Interestingly, local first-order analytic modeling for Kemi forced by the air temperature gave similar results: a freezing date shift of 20 days, breakup date shift of 10 days, and thickness drop by 21 cm. However, climate prediction for the ice extent and quality needs a full two-dimensional dynamic-thermodynamic sea ice model since coupling of ice dynamics and thermodynamics plays an important role.

Figure 5. Modeled scenario of the mean sea ice thickness (cm, gray area) and sea surface temperature (°C, white area) in the Baltic Sea for the year 2050 (averaged over 2035–2065). The shaded area shows thickness and the white area shows temperature. The calculation is based on winter air temperature higher by 3.6 °C compared with the normal period 1961–1990.

Status of Knowledge of the Ice Season in the Baltic Sea

The present understanding of the physics of the Baltic Sea ice season was largely gained by the end of the 1990s (see reviews by Granskog et al., 2006; Leppäranta & Myrberg, 2009; Vihma & Haapala, 2009). The recent focus of research has been on modeling of climate impact on the Baltic Sea ice season and further development of satellite remote sensing methods for operational monitoring. New scientific findings have concerned the coastal sea ice zone where less work was done when the main research motivation was winter shipping. Coastal sites provide good possibilities for fundamental small-scale investigations, and coastal ice zone is connected to many practical issues, such as roads and wind farms.

Sea ice physics is examined over a wide range of scales (Leppäranta & Myrberg, 2009; Weeks, 2010; Weeks & Ackley, 1986). Microscale includes individual grains and ice impurities, extending from submillimeter scale to 0.1 m. In the small or local scale, 0.1–10 m, sea ice is a solid sheet, a polycrystalline continuum with substructure classified according to the formation mechanism as congelation ice, snow-ice, and frazil ice (Eicken & Lange, 1989). Ice floe scale extends from 10 m to 10 km and includes individual floes and ice forms such as rubble, pressure ridges, and landfast ice. When the scale exceeds the floe size, the scale is called mesoscale at 100 km and large scale at more than 500 km, and the sea ice medium is called drift ice or pack ice.

The Baltic Sea possesses only first-year sea ice. This ice is similar with first-year polar sea ice with two major differences. First, the salinity of first-year ice is order of 1 ppt in the Baltic Sea and 10 ppt in the polar seas. Second, the typical thickness of undeformed ice is 5–50 cm in the Baltic Sea and five times that much in the polar seas, and a corresponding scaling holds also for the length scales of the basins. First-year sea-ice with similar thickness and length scale as in the Baltic Sea exists in the seasonal sea ice zone, for example, in the Sea of Okhotsk and in the Gulf of St. Lawrence, but there the salinity of sea ice (and seawater) is oceanic. But Baltic Sea ice is unique as a low-salinity, or brackish, first-year sea ice.

Microscale

The structure of Baltic Sea ice is similar to that of first-year sea ice, apart from river mouths where the salinity is less than 1–2 ppt (Granskog et al., 2004; Kawamura et al., 2001; Palosuo, 1961; Weeks, 2010). Ice crystals have jagged boundaries, and the ice–water interface is cellular. The primary ice layer on top is very thin when formed in calm, clear weather but in windy conditions or in the presence of snowfall, a frazil top layer of several centimeters forms. Ice crystal c-axes are vertical or random in the top layer depending on the weather conditions. Congelation ice is underneath the primary ice. First, there is a 10–20 cm transition layer where ice crystals grow larger, and their c-axes turn horizontal. Farther below the crystals are columnar with diameter 0.5–5 cm and height 5–50 cm, and their size grows downward in the ice cover. Frazil crystals are of millimeter size or less, whereas snow-ice crystal size is connected to snow structure and is in the order of millimeters (Granskog et al., 2004; Kawamura et al., 2001).

Saline water or brine is found inside the ice in brine pockets as in polar seas, and the composition of the Baltic Sea brackish ice is assumed to follow the sea ice phase diagram of Assur (1958). The salt inclusions have an influence on the physical properties of sea ice (Leppäranta & Manninen, 1988; Leppäranta & Myrberg, 2009; Schwerdtfeger, 1963). Algae grows in brine pockets in spring when the ice is warm and sunlight penetrates in the ice, as takes place in polar seas in summer (Ikävalko & Thomsen, 1997; Kaartokallio, 2004). Microscale studies of Baltic Sea ice have been limited, but some work was done in connection of radar signature modeling in the 1990s (Askne et al., 1992; Leppäranta et al., 1992).

Small-Scale

The thickness of undeformed Baltic Sea is at maximum around 100 cm in the north (the record is 122 cm off Tornio, Finland), while in the south it is of the order of 10 cm (SMHI & FIMR, 1982). Ice formation is based on three different mechanisms, which produce congelation ice, frazil ice, and superimposed ice.

Congelation ice is the most common form, growing down from the ice–water interface (Weeks, 1998; Weeks & Ackley, 1986). Congelation ice is the governing ice type also in lakes in the Baltic Sea drainage basin and in the Arctic Ocean. In the Baltic Sea the bottom layer, with thickness on the order of 1 cm, consisting of separate crystal plates penetrating to the saline water, is called the skeleton layer that has no significant strength (Kawamura et al., 2001). In normal sea ice the skeleton layer is thicker whereas it is absent in freshwater ice (Weeks & Ackley, 1986).

Snow-ice is frozen slush that has formed on ice as a mixture of snow and seawater, meltwater of snow, or liquid precipitation (Leppäranta & Myrberg, 2009). In the Baltic Sea, the most common slush formation process is due to flooding of seawater when the snow weight has forced the ice down below the sea surface level, that is,

ρshs>(ρwρi)hi(1)

where ρw,ρi,ρs are the densities of water, ice, and snow, respectively, hs is snow thickness, and hi is ice thickness. Since (ρwρi)/ρs~1/3, the thickness of snow needs to be at least one-third of the thickness of ice for the flooding to occur. Sea ice is normally porous so that when pressed beneath the sea level, the sea water can penetrate through. By melt-freeze cycles all snow may change into snow-ice or surface ice, and liquid precipitation may give raise to snow-ice and surface ice production.

Occurrence of frazil ice is not well known in the Baltic Sea. It is not common, but according to occasional observations it does exist. Frazil ice forms in open water areas; crystals are fine (1 mm or less) and granular, drift free in turbulent flow, and may attach to the bottom of existing ice or join together into pancake ice at the surface. In shallow and well-mixed waters frazil ice may also attach on the sea bottom to form anchor ice. In rivers frazil ice is generated in rapid streams, which have open surface in cold periods, and in Antarctic seas frazil ice is the dominant ice type.

Routine thickness observations in the Baltic Sea provide the total ice thickness and often also the fraction of superimposed ice, that is, snow-ice and surface ice, where the surface ice proportion is usually very small (Seinä & Peltola, 1991). The thickness of superimposed ice is monitored with a stake frozen into the ice in the beginning of the ice season (Palosuo, 1965). The origin of superimposed ice does not come clear from the thickness measurements, but the oxygen isotope ratio can tell the proportion of precipitation in an ice floe (Granskog et al., 2004; Kawamura et al., 2001). Landfast sea ice in the Baltic Sea ice has usually three layers: primary ice, congelation ice, and snow-ice, where the primary ice may be frazil but without proper data it is normally taken into the congelation ice layer in data analysis. Drift ice has more complicated stratigraphy due to mechanical processes and possible frazil ice formation events during the winter.

The mean annual maximum ice thickness increases toward the north and east (Table 1). The variance is largest in the south where the sea is ice-free in mild winters. The mean ice thickness is 76 cm in the north and 20–40 cm in the Archipelago Sea. Kumlinge is located in the center of the Archipelago Sea while Utö is at its southern edge. The volume of snow-ice is 10%–50% of the total thickness, while the snow thickness is 25%–75% of the total ice thickness. Snow cover looks thicker than the Archimedes law supports (Eq. 1), but that can be explained by local variations in snow thickness and temporal variations in snow density.

Table 1. Statistics of the Annual Maximum Thickness of Ice, Superimposed Ice, and Snow During 1961–1990 for Landfast Ice Along the Coast of Finland

Station

Total ice (cm)

Superimposed ice (cm)

Snow (cm)

Mean ± SD

Min–Max

Mean ± SD

Min–Max

Mean ± SD

Min–Max

Kemi

76 ± 11

58–111

25 ± 12

4–43

34 ± 12

10–60

Valassaaret

51 ± 16

18–80

21 ± 14

0–40

38 ± 17

5–70

Kumlinge

37 ± 19

0–75

7 ± 5

0–20

22 ± 13

0–42

Utö

21 ± 22

0–69

8 ± 17

0–50

5 ± 8

0–30

Kotka

50 ± 18

16–90

18 ± 9

4–35

25 ±14

2–55

Note: Kemi is in the Gulf of Bothnia, Valassaaret in the Northern Quark, Kumlinge and Utö in the Archipelago Sea, and Kotka in the Gulf of Finland. SD = standard deviation.

Based on Seinä & Peltola, 1991.

Sea ice thickness in the Baltic Sea has been modeled by numerical one-dimensional models with good results (Launiainen & Cheng, 1998; Leppäranta, 1983a; Saloranta, 2000). The evolution of ice thickness is largely dependent on the air temperature and snow accumulation (Leppäranta, 1993; Leppäranta & Myrberg, 2009). A simple scaling law is

hi=a2S+b2b(2)

where S=0tmin{TfTa,0}dt is the sum of freezing-degree days and a and b are empirical parameters, a23cmd11 depending on snow accumulation and b~10cm depending on the intensity of air–ice interaction. This gives hi~10cm in southern lagoons and hi~75cm in north in normal winters. Analytic modeling of ice melting is straightforward, since ice loss is then due to the net gain of energy from the external fluxes and heat conduction is not significant (Leppäranta & Myrberg, 2009). Melting progresses as

hi(t)=hi(tM)1ρiLftMt[Q0+QI+Qw]dτ (3)

where Q0,QIandQw are the heat fluxes at the surface, in the ice interior, and at the ice bottom, Lf is the latent heat of melting, and tM is the beginning of the melting period. The heat flux in Eq. (3) can be simplified into a degree-day formula (Leppäranta & Wen, 2022)

Q0+QI+Qw=A0+A1max{ TaTf,0 }(4)

where A0 depends on latitude and time and A1~0.5cm1d1 is the degree-day coefficient.

In addition to the air temperature, ice growth is sensitive to the timing of snow accumulation, while ice melting is sensitive to the optical properties of ice. The space–time variability of the snow cover is a major factor during the growth of ice cover, and the albedo and light attenuation coefficient of melting ice raise parameterization problems for the solar flux. Therefore, numerical models are needed for good simulation of the local ice season cycle in the Baltic Sea.

Ice formation, growth, and melting are driven by atmospheric and solar heating. The surface layer temperature is dominated by the radiation balance and air–sea turbulent exchange, and the thermal memory of the Baltic Sea response is a few months. The mean air temperature is below the freezing point for 6 months on the northern coast but less than 1 month in the south, which explains the large spatial variability in ice conditions. The winter weather in the Baltic Sea is largely determined by the North Atlantic Oscillation that results in a large interannual variability in the severity of the ice season (Jevrejeva, 2002; Omstedt et al., 2004; Palosuo, 1953; Tinz, 1996). A key point in statistical analyses is to account for the probability of ice occurrence, that is, standard statistics are conditional but fractiles such as median and quartiles are more feasible for ice climate studies (Leppäranta, 2014).

Ice Floe Scale

The size of ice floes ranges from 20 m to several kilometers in the Baltic Sea. Ice pieces smaller than 20 m are ice blocks (WMO, 2014). Drift ice landscape consists of leads and ice floes with ridges, hummocks, and other morphological characteristics (Figure 6). Sea ice types have been defined originating from practical shipping activities in ice-covered waters (WMO, 2014). They are based on how the ice looks to an observer on a ship or in an aircraft. The formation mechanism, aging, and deformation influence the appearance, which therefore contains information of the ice thickness, seldom known from direct measurements.

Figure 6. Drift ice landscape in the northern Baltic Sea in April. Drifting scattered floes are seen in front, and farther back there is a compact ice field.

Source: Photograph by the author.

Ice floes have been mapped by remote sensing methods for their shapes and size distribution (see, e.g., Leppäranta, 1983b). Floes are rectangular or pentagons in winter and rounded in spring. The significance of their size and shape shows up in the diffusion of pollutants, for example, oil spills, on surface (Wang et al., 2008), and in sea ice dynamics they show up in characteristic length scales (Leppäranta, 2011).

The most studied floe scale question is sea ice ridging in the Baltic Sea ( Leppäranta & Hakala, 1992; Leppäranta et al., 1995; Palosuo, 1975). Ridges are the most difficult sea ice obstacles for icebreakers (Kujala, 1996) and cause the highest loads on offshore structures (Gravesen & Kärnä, 2009), and their formation appears as the principal sink of mechanical energy in sea ice dynamics (Leppäranta, 2011). Ridges appear as long linear formations with a triangular or trapezoidal cross-section.

Ridges form under pressure when the thickness of ice is more than about 10 cm. If the ice is thinner, rafted ice or slush belts form. Freely floating ridges follow the Archimedes law for the ratio of their sail and keel volumes. They contain a frozen layer near the sea surface level, and higher up and deeper down the ice blocks are loose or weakly frozen together (Figure 7). Typical large ridges are 5–15 m in total vertical extent (sail height plus keel depth), and their spatial frequency is typically 1–5 km–1 (Lewis et al., 1993).

Near the shoreline, drifting ridges may hit the sea bottom, ground, and possibly continue movement scouring the bottom sediment. Resulting tracks have been recognized in sonar data. Scouring makes a risk to cables and pipelines buried in the sediment. At open coastlines, strong winds may drive ice ride-up and pile-up on shore that causes coastal erosion and gives load on structures near the shoreline (Alestalo & Häikiö, 1975; Girjatowicz, 2004; Leppäranta, 2013).

Figure 7. Seasonal evolution of the cross-section of a sea ice ridge in the Bay of Bothnia in winter 1991 based on three cross-sectional profilings (Leppäranta et al., 1995). The top and bottom curves are drawn, and the black and white columns represent drill holes with black for ice and white for pores. Horizontal axis shows distance from a reference point and vertical axis shows elevation with reference to sea surface.

Mesoscale

Drift ice forms since the forcing, the wind stress times the fetch, can be more than the strength of the ice cover. Ice breaks into floes, and this granular medium undergoes near-continuous movement and deformation. If the ice is thick enough, depending on the size of the basin, the plastic yield strength may stop the movement of ice, as has taken place in the Bay of Bothnia and Gulf of Finland when the whole ice pack has been at least a half-meter thick (Leppäranta & Myrberg, 2009). The annual ice season in the Baltic Sea is a basin-wide process, where ice drift drives spatial interactions (Leppäranta, 2011). Thermodynamics and dynamics are coupled in that ice growth strengthens the ice cover while ice dynamics creates open water where new ice may grow rapidly. In warm periods, ice conditions change only due to sea ice drift and deformation, while very thick ice in a basin may stay stationary.

Drift ice medium is described with the material state variables, which contain the properties needed to model the internal stress. The state information is contained in the ice thickness distribution Π(h,t), defined for a given spot, for example, a grid cell, as the normalized area of ice with thickness less than or equal to h. The drift ice state is a multi-dimensional functional of the thickness distribution, J=J(Π). The two-level ice state is J={ A,h˜ }, where A=Π(0) is ice concentration or compactness, and h˜ is the mean ice thickness (Doronin, 1970). For higher resolution of the ice state, multilevel forms are needed (Coon, 1980; Thorndike et al., 1975). An example of the thickness distribution in the Baltic Sea is given as a histogram in Figure 8. Its upper tail covers ridged ice.

Figure 8. An example of ice thickness distribution in the Baltic Sea based on airborne electromagnetic mapping in three ice patches with different degree of ridging. The vertical axis shows the number of cases, and each thickness observation represents an average over 100 m distance.

An ice conservation law is needed for the thickness distribution. In general, the ice thickness field in a given spot changes by advection, mechanical deformation, and thermodynamics (Thorndike et al., 1975). The conservation law is written for the thickness distribution as

Πt+u·Π=ΨΠ·u+Φ(h)Πh(5)

where Ψ represents the mechanical deformation of the ice field, and Φ(h) is the growth rate of ice with thickness h. The mechanical redistributor Ψ tells how ice thickness field changes under compression that involves parameterization problems. A general approach has been to compress thin ice to a multiple of 5–15 of its original thickness. The growth rate Φ(h;t) shifts the distribution in the thickness space.

The size and shape of floes are horizontal properties and therefore observable by aerial photography and satellite images. In mid-winter the floes are more difficult to distinguish since they freeze together and often have a snow cover. In spring floes break into smaller pieces and become rounded from the wintertime polygonal shape. The size and shape of ice floes have a secondary role in sea ice dynamics. When an ice field is open and floes are small, diffusion is significant. Collisions between ice floes transfer momentum and consume mechanical energy but their significance is much less than friction in a compact ice cover. In close ice internal friction becomes large, floes group together, and diffusion weakens.

Drift ice is a granular, two-dimensional, nonlinear medium (see, e.g., Leppäranta, 2011). The rheology of drift ice is in general form σ=σ(J,ε,ε˙), where ε is strain and ε˙ is strain-rate. The simplest rheology is stress-free ice (σ0), the motion called free drift, applicable for low ice compactness (A<0.8). When A~0.8, collisions and shear friction between ice floes become significant but stresses are still low, and the ice behaves in a nonlinear viscous manner (Shen et al., 1986). Compact drift ice (A>0.9) has a plastic rheology (Coon, 1980; Hibler, 1979). Hunke and Dukowicz (1997) developed further a computationally efficient elastic-viscous-plastic model, and recently a Maxwell elasto-brittle rheology has been introduced for sea ice modeling (Dansereau et al., 2016).

The compressive plastic ice stress scales as P~3·104Nm1 for A=1 and h~1m (Coon, 1980; Hunke & Dukowicz, 1997; Leppäranta et al., 1998). In general, this stress is sensitive to ice compactness and depends on the mode of deformation, being very low for tension. For the thickness h and compactness A, the compressive strength can be approximated by

P=Pn*hnexp[ C(1A) ](6)

where Pn and n are the strength parameters of compact ice, and C is the strength reduction for lead opening. In the original version, Hibler (1979) had n=1, but in general ½n2. The inverse of the parameter C is the e-folding value of strength for change in compactness, C1 (normally C=20) due to the observed high sensitivity. Note that the mesoscale drift ice strength is two orders of magnitude lower than the local scale strength (see Sanderson, 1988).

The two-dimensional equation of motion of drift ice is derived by integrating across the thickness of the ice and accounting for the Coriolis acceleration (see, e.g., Coon, 1980; Doronin, 1970; Leppäranta, 2011). The result is

ρih˜(ut+u·u+fk×u)=·σ+τa+τwρgh˜ξ(7)

where f=2Ωsinϕ is the Coriolis parameter, Ω=7.292·105s1 is the angular velocity of the Earth, ϕ is latitude, τaandτw are the tangential air–ice and water–ice stresses on the ice top and bottom surfaces, respectively, and ξ is the water level elevation. The air and water stresses are given by the bulk formulae (see, e.g., Andreas, 1998; McPhee, 2008):

τa=ρaCaUa[cosθaUa+sinθak×Ua](8a)
τw=ρwCw| Uwu |[ cosθw(Uwu)+sinθwk×(Uwu)](8b)

where Ca and Cw are the air and water drag coefficients, θaandθw are the boundary-layer angles in air and water, and Ua and Uw are the wind and current velocities. Representative values of the geostrophic drag parameters are Ca=0.6·103 and θa25° or, for the surface wind (10 m altitude), Ca10=1.7·103 and θa100° (Joffre, 1984), Cw=3.5·103 and θw25° (Leppäranta & Omstedt, 1990). In the Baltic Sea, wind is usually the driving force, balanced by the ice–water drag and the internal friction of the ice (Table 2). Internal friction may be strong due to the limited size of the Baltic Sea basins. Coriolis force is smaller than the three major forces and is always perpendicular and to the right of the ice motion.

Table 2. Scaling of the Equation of Motion of Drift Ice in the Baltic Sea

Term

Scale

Value mPa

Comments

Local acceleration

ρiHUT1

1

50 for rapid changes (T=103s)

Advective acceleration

ρiH(UUw)2L1

0.2

Usually much less than water stress

Coriolis term

ρiHfU

10

50 for H=1 and U=50cms1

Internal friction

PHL1

0–200

0 for open ice field, limited by forcing

Air stress

ρaCaUa2

175

Dominant forcing

Water stress

ρwCw(UUw)2

140

Range 75–200 ← directions of U,Uw

Pressure gradient

ρiHgξ

5

Slopes limited by water dynamics

Note: The representative fundamental scales are the following: ice thickness H=½m, ice velocity U=20cms1, ice strength P=25kPa, wind velocity Ua=15ms1, water velocity Uw=5cms1, surface slope ξ=106, time T=1 day, and horizontal length L=100km.

There are three principal time scales: local acceleration TI=H(CwU)~minutes, the Coriolis period f112hours, and deformation TD=L/U~days. These time scales are well separated, TIf1TD. The first-order free drift solution provides a limiting upper scale for the ice velocity (Leppäranta, 2011), obtained from the steady state equation as:

u=aUa+Uw,aρaCaρwCw~1.8%.(9)

That is, wind-driven free ice drift (u=aUa) is parallel to the geostrophic wind or directed 25° to the right from the surface wind. The coefficient a is 3% for the surface wind in the Baltic Sea. This solution works in open drift ice, for example, for the broken state of ice cover in the ice decay period. In compact ice, forcing needs to be strong enough to break the yield strength σY and start the ice drift.

The lateral boundary of an ice field can be taken as the shoreline where the ice velocity is zero. At the land boundary, the ice is allowed to stay or move into the basin. Open water is taken as “ice with thickness equal to zero”. In spring, sea ice becomes “rotten” and its strength decreases due to internal melting, and the ice cover may be easily broken. In general, the time scale of ice growth and melting is much longer than the dynamics time scale, and often for short-term simulations thermodynamics can be ignored.

Ice Climate

The evolution of a sea ice cover is a three-dimensional process. In large basins, solid sea ice lids are statically unstable and break into fields of ice floes, undergoing transport as well as opening, closing, and ridging, which altogether create the exciting sea ice landscape as it appears to the human eye. Ice grows and melts in the vertical direction, while the dynamics transports and deforms ice fields. Growth and melting influence ice strength, and transport and deformation influence the ice–air and ice–sea heat exchange. Thus, thermodynamics and dynamics of sea ice constitute a coupled problem.

Ice Phenology, Thickness, and Extent in the Baltic Sea

In fall, the surface layer temperature in the Baltic Sea decreases at the rate of 3–4 °C per month, lagging behind the air temperature (Haapala & Leppäranta, 1996, 1997; Omstedt, 1990). The ice season begins on average in the middle of November on the northern coast of the Bay of Bothnia shortly after air temperature has downcrossed the freezing point. In the 20th century the earliest, average, and latest freezing dates were October 6, November 10, and December 23, respectively, at Kemi in the north, with the range of 2.5 months (Jevrejeva et al., 2004). In the Szczecin lagoon in the south, the corresponding dates were November 10, December 25, and February 25, showing the range of 3.5 months (the probability of annual ice occurrence is 0.9).

The freezing front progresses southward and westward from north and east (SMHI & FIMR, 1982), and in the central basins the freezing date is much later than at the coast due to the large heat content in the water body. The Bay of Bothnia freezes over on average in mid-January, and in normal winters 1 month later ice covers the Sea of Bothnia, the Gulf of Finland, and the Gulf of Riga. Farther south, ice occurs only in shallow coastal areas in normal winters, such as the Curonian Lagoon and the Szczecin Lagoon. The annual ice extent is at largest within mid-February to mid-March, when on average the ice-covered area is 45% of the total area of the Baltic Sea (Leppäranta & Myrberg, 2009). The Gotland Sea takes a long time to freeze because the halocline is deep and climate milder than in the north. In the 1800s, the Baltic Sea froze over ten times but in the following century only three times. The most recent case dates to 1947, and in 1987 the coverage was 96% of the sea area. Even then the ice in the Gotland Sea is rather thin and breaks easily (Palosuo, 1953).

In the landfast ice zone, the mean annual maximum ice thickness is 75 cm in the north, 35 cm in the Archipelago Sea, and 20 cm in the southern shallow lagoons (Seinä & Peltola, 1991; SMHI & FIMR, 1982). The variance is largest in the south where the sea is ice-free in mild winters. Ice roads are set up in the landfast ice zone, for example, from the mainland to the island of Hailuoto in the north. Landfast ice may be broken by storms and forced to ride and pile onshore that may cause coastal erosion and bottom scouring.

Climate is moist in the Baltic Sea region, and winter precipitation is usually connected to intense cyclones and accounts for 20–40 mm per month. In the north, most of this water comes down in solid phase. The mean annual maximum thickness of snow onshore is 40–50 cm in the north and 10 cm in the south (Finnish Meteorological Institute web site; Szwed et al., 2019), which correspond to the snow water equivalent of about 10–12 cm and 2.5 cm, respectively. In the north, the snow water equivalent on land is close to average net precipitation, whereas in the south only a small fraction of the net precipitation shows up in the snow cover.

Apart from new ice and melting stage, Baltic Sea ice is covered by snow. In the north, the mean thickness of snow on ice is 26 cm (Leppäranta & Myrberg, 2009) while the mean thickness of snow-ice is 25 cm (Seinä & Peltola, 1991). Snow-ice is a mixture of snow and seawater, and assuming the proportions 50%–50% (snow is compressed in flooding), the snow and snow-ice correspond to about 40 cm snow accumulation that corresponds to the accumulation on land. Farther south this reasoning does not work, but ice cores can be analyzed for their oxygen isotopes O18 and O16 to estimate the fraction of precipitation in the ice cover (Granskog et al., 2004).

Major consequences of snow accumulation on ice are increased thermal insulation and reflectance of solar radiation. Thermal conductivity and optical thickness are one order of magnitude smaller in snow than in ice, albedo is 0.3–0.5 for bare ice and 0.5–0.9 for snow-covered ice, and an additional consequence is snow-ice formation. Atmospheric deposition is accumulated on the ice cover to be released during the short melting period. The most common form of ice in the Baltic Sea is congelation ice, which grows down from the ice–water interface (Granskog et al., 2004; Palosuo, 1963). Snow-ice forms on top of the congelation ice in a slush layer, and the fractional volume of snow-ice is 10%–50% (Leppäranta & Myrberg, 2009; Palosuo, 1963).

There are long time series of ice phenology in the Baltic Sea that have provided highly important information of ice climatology (Haapala et al., 2015; Jevrejeva et al., 2004; Lépy, 2012; Pärn et al., 2022). The longest ones start in the 1800s, ice extent goes even to the year 1720 but there the old data is indirect (Leppäranta & Seinä, 1985). Recently also satellite-based time series have been used for ice phenology in the Baltic Sea (see, e.g., Idzelytè et al., 2019).

At the annual maximum, ice extent is 12%–100% of the area of the Baltic Sea and Kattegat (50,000–420,000 km2). The evolution of the ice extent follows largely air temperature, wind, and the bottom topography. The sea depth gives the necessary heat loss before freezing, and wind transports and deforms the ice cover. Assuming wind effects are statistically similar, the ice extent follows the hypsographic curve until the cooling has reached the halocline (Karetnikov et al., 2017; Myrberg et al., 2006).

Melting of ice begins at the time the heat balance turns positive in late winter. In the Baltic Sea springtime, the radiation balance is the dominant factor in ice decay and at times the turbulent air-ice heat fluxes are also large (Leppäranta & Myrberg, 2009). By liquid water formation and recrystallization of snow grains, the albedo starts to decrease, and as the incoming radiation increases day by day, the melting rate increases. Due to sea ice dynamics, open water spots add on the absorption of solar radiation that also provides positive feedback in ice decay.

The onset of melting takes place in the south in early March and melting progresses in the central basins with open water formation. Somewhat later, melting starts from the shoreline due to the shallow sea depth and the neighborhood of warm land. In the 20th century, the earliest, mean, and latest dates of ice break-up in the south were December 25, March 10, and April 18, respectively, showing the range of 3.8 months (Jevrejeva et al., 2004). In the north, the corresponding dates were April 16, May 21, and June 27, respectively, the range being 2.4 months. In general, close to the climatological margin dispersion in ice climatology is at its largest since random factors have a strong role there.

The length of the melting season is 40–50 days in the north, and the average melting rate is about 2 cm per day in equivalent ice thickness. Ice melts at the top and bottom surfaces and in the interior, and as the porosity of ice has achieved 50% or so, the ice breaks and the pieces disappear fast. The outer fast ice boundary with large and grounded ridges is the last place where ice is seen at the end of the ice season. Remnants of old ridges are seen in June in some years in the Bay of Bothnia. In 1867, which is the latest very cold, famine year in Finland, drifting ice was reported on July 17 in the Gulf of Bothnia.

Modeling the Seasonal Cycle of Baltic Sea Ice

Elements, for example, grid cells, of drift ice contain many ice floes and have the length scale of 10 km (Leppäranta, 2011). Ice thickness is presented as a histogram for each element, with zero representing open water, and the largest values are from sea ice ridges. To simulate the evolution of the ice season in a marine basin, the following elements are needed: (a) momentum equation, (b) ice conservation law, (c) ice state, and (d) ice rheology.

Seasonal models have been used in the Baltic Sea for ice climate investigations (Haapala & Leppäranta, 1996, 1997; Haapala et al., 2015). In such a limited region, the treatment of the boundary conditions is important, but initial conditions are irrelevant since they can be removed starting the simulation in the previous summer. The deformation length scale of compact drift ice is around 100 km, and therefore internal friction has a major role in fully ice-covered Baltic Sea basins. In seasonal modeling, the objective is basic research, ice climatology, or coupled Baltic Sea–ice–atmosphere climate modeling.

First, local thermodynamic models were applied for the times of freezing and ice break-up and for the evolution of ice thickness. However, realistic ice dynamics are needed in seasonal modeling for the ice transport and, particularly, for opening and closing of leads. A large amount of heat is transmitted through leads from the sea to the atmosphere. Ice melting has a major influence on the hydrographic structure of the sea, and the motion of ice has an important role in driving the water circulation and transporting ice to melt in a basin different from the birth basin.

Seasonal models are forced by synoptic weather conditions for calibration with ice charts for the validation. The models are capable to simulate the whole ice season from one summer to the next one. Calibrated models have been then used for ice season scenarios based on the existing atmospheric climate scenarios. Calibration of the Baltic Sea ice climate model for a normal winter is shown in Figure 9. The initial time was May 1. The figure shows comparison between the model and observed ice condition in March when the ice extent was at its largest. There were small discrepancies in that the surface temperature in the south was too high in the model, and farther north the model showed an open water region, but the ice chart showed thin ice. Elsewhere the mean thickness came out quite well. Similar results were obtained for comparisons in mild and severe winters.

Figure 9. Calibration of a Baltic Sea ice climate model: (A) ice chart showing reality, (B) ice thickness (cm) and open water surface temperature from the model (°C) from the model, and (C) mean thickness of deformed ice (cm). The model initial time was May 1, 1983, and the comparison here is for March 22, 1984. SST = sea surface temperature.

Conclusions

The article has presented an overview of the Baltic Sea ice climate. This ice forms from the brackish water and is similar with first-year polar sea ice in the crystal structure and as well in the large-scale morphology. The ice climate problem has been examined by time series analyses, first-order analytic and local modeling, and full two- or three-dimensional ice–ocean numerical models.

In the climate perspective, the main question is sensitivity of the Baltic Sea ice season to air temperature (Haapala & Leppäranta, 1997; Haapala et al., 2015; Jevrejeva et al., 2004; Omstedt et al., 2004). It is clear that a higher temperature means a shorter ice season. In the data and model analysis (see the “Small-Scale” section of “Status of Knowledge of the Ice Season in the Baltic Sea”), it can be said that the local sensitivity of the freezing and breakup dates of landfast ice is 5–8 days per 1 °C change of air temperature. Assuming no significant change in the hydrography of the water masses, the change in drift ice period would be the same. The delay of the freezing date is clear since it depends on the mixed-layer thickness and rate of atmospheric cooling, but ice decay and breakup depend also on the maximum ice and snow thickness of the season. However, in the past 100–150 years the statistics show about the same length of the shifts in the freezing and breakup dates, suggesting that the air temperature is strongly connected to the zero upcrossing of the heat balance in late winter.

Since the Baltic Sea is located at the climatological margin of global seasonal sea ice, the annual probability of ice occurrence is of great interest (Jevrejeva et al., 2004; SMHI & FIMR, 1982). Along with climate variations, the probability contours move in the Baltic Sea chart. Presently, 50% contour goes near the mouths of the Gulf of Bothnia, Gulf of Finland, and Gulf of Riga. In the time series analysis by Jevrejeva et al. (2004), it was shown that in the southern Baltic Sea, the probability has decreased significantly, by 2%–3% over a decade. However, in the freezing and breakup dates there were no significant changes, and thus climate warming showed up only in the decreasing probability of freezing.

The thickness of sea ice is more complicated than ice phenology, since it depends on the snow accumulation (Leppäranta, 1983a; Palosuo, 1963). The locally largest thicknesses have been obtained when the snow accumulation is weak, and in the case of strong insulative snow layer, the thickness of ice is about half of the thickness under bare ice conditions. In the case of heavy snow accumulation, snow-ice growth adds on the ice thickness but asymptotically the level is less than for bare ice conditions. Thus, when the snow layer has only insulative effect, change in snow accumulation causes opposite effect to ice thickness, whereas in the case of snow-ice existence, increase in snow accumulation increases ice thickness and vice versa. A possible additional factor in the future thermodynamic thickness of drift ice is the role of frazil ice generation in open water spots and leads.

The extent of ice in the Baltic Sea is closely related to the regional air temperature (Haapala & Leppäranta, 1997; Haapala et al., 2015; Omstedt et al., 2004; Palosuo, 1953). Palosuo (1953) compared the ice extent A with the mean December–March air temperature in Stockholm TSt with a regression line A80,00054,000TSt, where the area is in km2 and temperature in °C. Using a two-dimensional numerical model, the ice extent decreased from 190,000 in 1900–1994 to 65,000 km Haapala and Leppäranta (1997) in a 3.6 °C warming scenario; that is, the sensitivity was 35,000 km2 for 1 °C warming. This sensitivity was also obtained by Omstedt and Nyberg (1996). The long-term trend of the ice extent in the Baltic Sea has been about 4,000 km2 per decade (Haapala et al., 2015) and, considering the climate warming by about 0.1 °C per decade (Rutgersson et al., 2015), results in the sensitivity of 40,000 km2 per °C. The difference of Palosuo (1953) from the later works is that he had very cold winters from the Little Ice Age that brought in nonlinearity in the sensitivity to a very large extent. Thus, it seems that warming from the present climate leads to decrease of ice extent by 35,000–40,000 km2 per 1 °C air temperature change.

Acknowledgments

This article was written after I had retired from professorship in geophysics at the University of Helsinki, working on sea ice since 1974. I am most grateful for very good collaboration in sea ice field work and science to numerous colleagues in the Finnish Institute of Marine Research, sea ice researchers in the Baltic Sea countries, and students in the University of Helsinki.

Further Reading

  • Haapala, J., & Leppäranta, M. (1996). Simulating Baltic Sea ice season with a coupled ice-ocean model. Tellus, 48A(5), 622–643.
  • Haapala, J. J., Ronkainen, I., Schmelzer, N., & Sztobryn, M. (2015). Recent change—sea ice. In BACC II Author Team (Eds.), Second assessment of climate change for the Baltic Sea basin (pp. 145–153). Springer.
  • Leppäranta, M. (2011). The drift of sea ice (2nd ed.). Springer-Praxis.
  • Leppäranta, M., & Myrberg, K. (2009). Physical oceanography of the Baltic Sea. Springer-Praxis.
  • Omstedt, A. (1990). Response of Baltic Sea ice to seasonal, interannual forcing and climate change. Tellus, 42A, 286–301.
  • Saloranta, T. (2000). Modeling the evolution of snow, snow ice and ice in the Baltic Sea. Tellus, 52A, 93–108.
  • Swedish Administration of Shipping and Navigation, and Finnish Board of Navigation. (1982). An ice atlas for the Baltic Sea, Kattegat, Skagerrak and Lake Vänern. Swedish Administration of Shipping and Navigation, and Finnish Board of Navigation.
  • Weeks, W. F. (1998). Growth conditions and the structure and properties of sea ice. In M. Leppäranta (Ed.), Physics of ice-covered seas (Vol. 1, pp. 25–104). Helsinki University Press.
  • World Meteorological Organization. (2014). WMO sea-ice nomenclature, terminology, codes and illustrated glossary. WMO/OMM/BMO 259.

References

  • Alestalo, J., & Häikiö, J. (1975). Ice features and ice-thrust shore forms at Luodonselkä, Gulf of Bothnia, in winter 1972/73. Fennia, 144, 1–24.
  • Andreas, E. L. (1998). The atmospheric boundary layer over polar marine surfaces. In M. Leppäranta (Ed.), The physics of ice-covered seas (Vol. 2, pp. 715–773). Helsinki University Press.
  • Arst, H., Erm, A., Leppäranta, M., & Reinart, A. (2006). Radiative characteristics of ice-covered freshwater and brackish water bodies. Proceedings of the Estonian Academy of Science & Geology, 55(1), 3–23.
  • Askne, J., Leppäranta, M., & Thompson, T. (1992). Bothnian experiment in preparation for ERS-1, 1988 (BEPERS-88)—an overview. International Journal of Remote Sensing, 13, 2377–2398.
  • Assur, A. (1958). Composition of sea ice and its tensile strength. In W. Thurston (Ed.), Arctic Sea ice (pp. 106–138). Publ. No. 598. U.S. National Academy of Science/National Research Council.
  • Berglund, R., Kotovirta, V., & Seinä, A. (2007). A system for icebreaker navigation and assistance planning using spaceborne SAR information in the Baltic Sea. Canadian Journal of Remote Sensing, 33, 378–387.
  • Blomquist, Å., Pilo, C., & Thompson, T. (1976). SEA ICE-75, summary report. Winter Navigation Research Board, research report 16:9. Swedish Administration of Shipping and Navigation.
  • Coon, M. D. (1980). A review of AIDJEX modeling. In R. S. Pritchard (Ed.), Proceedings of ICSI/AIDJEX Symposium on sea ice processes and models (pp. 12–23). University of Washington.
  • Dansereau, V., Weiss, J., Saramito, P., & Lattes, P. (2016). A Maxwell elasto-brittle rheology for sea ice modelling. The Cryosphere, 10, 1339–1359.
  • Denoth, A., Folgar, A., Weiland, P., Mätzler, C., Aebischer, H., Tiuri, M., & Sihvola, A. (1984). A comparative study of instruments for measuring the liquid water content of snow. Journal of Applied Physics, 56, 2154–2160.
  • Doronin, Y. P. (1970). On a method of calculating the compactness and drift of ice floes. Trudy Arkticheskii i Antarkticheskii Nauchno-issledovatel’skii Institut, 291, 5–17. [English transl. in AIDJEX Bulletin, 3, 22–39.]
  • Eicken, H., & Lange, M. (1989). Development and properties of sea ice in the coastal regime of the southern Weddell Sea. Journal of Geophysical Research, 94, 8193–8206.
  • Enkvist, E. (1972). On the resistance encountered by ships operating in the continuous mode of icebreaking. Svenska Tekniska Vetenskapsakademien i Finland, 24, 1–181.
  • Girjatowicz, J. P. (2004). Ice thrusts and piles on the shores of the southern Baltic Sea coast (Poland) lagoons. Baltic Coastal Zone, 8, 5–22.
  • Granqvist, G. (1938). Den baltiska isveckan 12–18. Terra, 50(4), 385–391.
  • Granskog, M., Kaartokallio, H., Kuosa, H., Thomas, D. N., & Vainio, J. (2006). Sea ice in the Baltic Sea—A review. Estuarine, Coastal and Shelf Science, 70(1–2), 145–160.
  • Granskog, M., Leppäranta, M., Ehn, J., Kawamura, T., & Shirasawa, K. (2004). Sea ice structure and properties in Santala Bay, Baltic Sea. Journal of Geophysical Research, 109, C02020.
  • Gravesen, H., & Kärnä, T. (2009). Ice loads for offshore wind turbines in the southern Baltic Sea [conference session]. Proceedings of the 20th International Conference on Port and Ocean Engineering Under Arctic Conditions, Luleå, Sweden.
  • Grönvall, H. (1988). Finnish ice service. Finnish Marine Research, 256, 95–110.
  • Grönvall, H., & Seinä, A. (2002). Satellite data use in Finnish winter navigation. Elsevier Oceanography Series, 66, 429–436.
  • Haapala, J., & Leppäranta, M. (1996). Simulating Baltic Sea ice season with a coupled ice-ocean model. Tellus, 48A(5), 622–643.
  • Haapala, J., & Leppäranta, M. (1997). The Baltic Sea ice season in changing climate. Boreal Environment Research, 2, 93–108.
  • Haapala, J. J., Ronkainen, I., Schmelzer, N., & Sztobryn, M. (2015). Recent change—Sea ice. In BACC II Author Team (Eds.), Second assessment of climate change for the Baltic Sea basin (pp. 145–153). Springer.
  • Hibler, W. D., III. (1979). A dynamic thermodynamic sea ice model. Journal of Physical Oceanography, 9, 815–846.
  • Hunke, E., & Dukowicz, J. K. (1997). An elastic-viscous-plastic model for sea ice dynamics. Journal of Physical Oceanography, 27, 1849–1867.
  • Idzelytè, R., Kozlov, I. E., & Umgiesser, G. (2019). Remote sensing of ice phenology and dynamics of Europe’s largest coastal lagoon (the Curonian lagoon). Remote Sensing, 11, 2059.
  • Ikävalko, J. (1997). Studies of nanoflagellate communities in the sea ice of the Baltic and the Greenland Sea [Ph.D. dissertation, University of Helsinki].
  • Ikävalko, J., & Thomsen, H. A. (1997). The Baltic Sea ice biota (March 1994): A study of the protistan community. European Journal of Protistology, 33, 229–243.
  • Jevrejeva, S. (2002). Association between the ice conditions in the Baltic Sea and the North Atlantic Oscillation. Nordic Hydrology, 33, 319–330.
  • Jevrejeva, S., Drabkin, V. V., Kostjukov, J., Lebedev, A. A., Leppäranta, M., Mironov, Y. U., Schmelzer, N., & Sztobryn, M. (2004). Baltic Sea ice seasons in the twentieth century. Climate Research, 25, 217–227.
  • Joffre, S. M. (1984). The atmospheric boundary layer over the Bothnian Bay: A review of work on momentum transfer and wind structure (Report No. 40). Winter Navigation Research Board.
  • Jurva, R. (1937). Über die Eisverhältnisse des Baltischen Meeres an den Küsten Finnlands. Merentutkimuslaitoksen Julkaisu/Havsforskningsinstitutets Skrift, 114. Finnish Institute of Marine Research.
  • Jurva, R. (1952). On the variations and changes of freezing in the Baltic during the last 120 years. Fennia, 75, 17–24.
  • Kaartokallio, H. (2004). Food web components and physical and chemical properties of Baltic Sea ice. Marine Ecology Progress Series, 273, 49–63.
  • Karetnikov, S., Leppäranta, M., & Montonen, A. (2017). Time series over 100 years of the ice season in Lake Ladoga. Journal of Great Lakes Research, 43(6), 979–988.
  • Kari, E., Merkouriadi, I., Walve, J., Leppäranta, M., & Kratzer, S. (2018). Development of under-ice stratification in Himmerfjärden Bay, north-western Baltic proper, and their effect on the phytoplankton spring bloom. Journal of Marine Systems, 186, 85–95.
  • Kawamura, T., Shirasawa, K., Ishikawa, N., Lindfors, A., Rasmus, K., Ehn, J., Leppäranta, M., Martma, T., & Vaikmäe, R. (2001). A time series of the sea ice structure in the Baltic Sea. Annals of Glaciology, 33, 1–4.
  • Korhonen, V. V. (1915). Die Ausdehnung und Höhe der Schneedecke [Ph.D. dissertation, University of Helsinki].
  • Kozlov, I. E., Krek, E. V., Kostianoy, A. G., & Dailidienė, I. (2020). Remote sensing of ice conditions in the southeastern Baltic Sea and in the Curonian lagoon and validation of SAR-based ice thickness products. Remote Sensing, 12, 3754.
  • Kuittinen, R. (1988). Determination of areal snow water equivalent using satellite images and gamma ray spectrometry. Acta Polytechnica Scandinavica, Civil Engineering and Building Construction Series, 91, 139.
  • Kujala, P. (1996). Semi-empirical evaluation of long term ice loads on a ship hull. Marine Structures, 9(9), 849–871.
  • Kuusisto, E. (1984). Snow accumulation and snowmelt in Finland. Publications of the Water Research Institute, 55, 149.
  • Launiainen, J. (1979). Studies of energy exchange between the air and the sea surface on the coastal area of the Gulf of Finland [Ph.D. thesis]. Finnish Marine Research, 246, 110.
  • Launiainen, J., & Cheng, B. (1998). Modelling of ice thermodynamics in natural water bodies. Cold Regions Science and Technology, 27(3), 153–178.
  • Leppäranta, M. (1981). An ice drift model for the Baltic Sea. Tellus, 33, 583–596.
  • Leppäranta, M. (1983a). A growth model for black ice, snow ice and snow thickness in subarctic basins. Nordic Hydrology, 14(2), 59–70.
  • Leppäranta, M. (1983b). Size and shape of ice floes in the Baltic Sea in spring. Geophysica, 19(2), 127–136.
  • Leppäranta, M. (1993). A review of analytical models of sea-ice growth. Atmosphere-Ocean, 31(1), 123–138.
  • Leppäranta, M. (2011). The drift of sea ice (2nd ed.). Springer-Praxis.
  • Leppäranta, M. (2013). Land-ice interaction in the Baltic Sea. Estonian Journal of Earth Sciences, 62(1), 2–14.
  • Leppäranta, M. (2014). Interpretation of statistics of lake ice time series for climate variability. Hydrology Research, 45(4–5), 673–684.
  • Leppäranta, M., & Hakala, R. (1992). Structure and strength of first-year sea ice ridges in the Baltic Sea. Cold Regions Science and Technology, 20, 295–311.
  • Leppäranta, M., Kuittinen, R., & Askne, J. (1992). BEPERS pilot study: An experiment with X-band synthetic aperture radar over Baltic Sea ice. Journal of Glaciology, 38(128), 23–35.
  • Leppäranta, M., Lensu, M., Kosloff, P., & Veitch, B. (1995). The life story of a first-year sea ice ridge. Cold Regions Science and Technology, 23, 279–290.
  • Leppäranta, M., & Manninen, T. (1988). Brine and gas content of sea ice with attention to low salinities and high temperatures. Internal Report, 1988(2), 15.
  • Leppäranta, M., & Myrberg, K. (2009). Physical oceanography of the Baltic Sea. Springer-Praxis.
  • Leppäranta, M., & Omstedt, A. (1990). Dynamic coupling of sea ice and water for an ice field with free boundaries. Tellus, 42A, 482–495.
  • Leppäranta, M., & Seinä, A. (1985). Freezing, maximum annual ice thickness and breakup of ice on the Finnish coast during 1830–1984. Geophysica, 21(2), 87–104.
  • Leppäranta, M., Sun, Y., & Haapala, J. (1998). Comparisons of sea-ice velocity fields from ERS-1 SAR and a dynamic model. Journal of Glaciology, 44(147), 248–262.
  • Leppäranta, M., & Wen, L. (2022). Ice phenology in Eurasian lakes over spatial location and altitude. Water, 14(7), 1037.
  • Lépy, E. (2012). Baltic Sea ice and environmental and societal implications from the comparative analysis of the Bay of Bothnia and the Gulf of Riga. Fennia—International Journal of Geography, 190(2), 90–101.
  • Lewis, J. E., Leppäranta, M., & Granberg, H. B. (1993). Statistical features of the sea ice surface topography in the Baltic Sea. Tellus, 45A, 127–142.
  • McPhee, M. (2008). Air-ice-ocean interaction: Turbulent boundary layer exchange processes. Springer.
  • Multala, J., Oksama, M., Hautaniemi, H., Leppäranta, M., Haapala, J., Herlevi, A., Riska, K., & Lensu, M. (1996). An airborne electromagnetic system on a fixed wing aircraft for sea ice thickness mapping. Cold Regions Science and Technology, 24(4), 355–373.
  • Myrberg, K., Leppäranta, M., & Kuosa, H. (2006). Itämeren fysiikka, tila ja tulevaisuus [The physics, state and future of the Baltic Sea]. Yliopistopaino.
  • Omstedt, A. (1985). An investigation of the crystal structure of sea ice in the Bothnian Bay (p. 19). SMHI Reports in Hydrology and Oceanography, 40.
  • Omstedt, A. (1990). Response of Baltic Sea ice to seasonal, interannual forcing and climate change. Tellus, 42A, 286–301.
  • Omstedt, A., & Nyberg, L. (1996). Modelling the Baltic Sea as thirteen sub-basins with vertical resolution. Tellus, 48A, 644–662.
  • Omstedt, A., Pettersen, C., Rodhe, J., & Winsor, P. (2004). Baltic Sea climate: 200 yr of data on air temperature, sea level variation, ice cover, and atmospheric circulation. Climate Research, 25(3), 205–216.
  • Palosuo, E. (1953). A treatise on severe ice conditions in the Central Baltic. Merentutkimuslaitoksen Julkaisu/Havsforskningsinstitutets Skrift, 156. Finnish Institute of Marine Research.
  • Palosuo, E. (1961). Crystal structure of brackish and fresh-water ice. Publication 54 (pp. 9–14). IASH Commission of Snow and Ice.
  • Palosuo, E. (1963). The Gulf of Bothnia in winter, II: Freezing and ice forms. Merentutkimuslaitoksen Julkaisu/Havsforskningsinstitutets Skrift, 208. Finnish Institute of Marine Research.
  • Palosuo, E. (1965). Frozen slush on lake ice. Geophysica, 9(2), 36–45.
  • Palosuo, E. (1975). Formation and structure of ice ridges in the Baltic (Winter Navigation Research Board, Rep. No. 12). Board of Navigation.
  • Pärn, O., Friedland, R., Rjazin, J., & Stips, A. (2022). Regime shift in sea-ice characteristics and impact on the spring bloom in the Baltic Sea. Oceanologia, 64(2), 312–326.
  • Pihkala, P., & Spring, E. (1985). A practical method for photographing snow samples. Report Series in Geophysics, No. 20. University of Helsinki.
  • Pulliainen, J., Grandell, J., & Hallikainen, M. (1999). HUT snow emission model and its applicability to snow water equivalent retrieval. IEEE Transactions on Geoscience on Remote Sensing, 37(3), 1378–1390.
  • Rutgersson, A., Jaagus, J., Schenk, F., Stendel, M., Bärring, L., Briede, A., Claremar, B., Hanssen-Bauer, I., Holopainen, J., Moperg, A., Nordli, Ø., Rimkus, E., & Wibig, J. (2015). Recent change—Atmosphere. In The BACC II Author Team (Eds.), Second assessment of climate change for the Baltic Sea basin, pp. 69–97. Regional Climate Studies. Springer.
  • Saloranta, T. (2000). Modeling the evolution of snow, snow ice and ice in the Baltic Sea. Tellus, 52A, 93–108.
  • Sanderson, T. J. O. (1988). Ice mechanics: Risks to offshore structures. Graham & Trotman.
  • Schwerdtfeger, P. (1963). The thermal properties of sea ice. Journal of Glaciology, 4, 789–807.
  • Seinä, A., & Peltola, J. (1991). Duration of the ice season and statistics of fast ice thickness along the Finnish coast 1961–1990. Finnish Marine Research, 258, 48.
  • Shen, H. H., Hibler, W. D., III, & Leppäranta, M. (1986). On applying granular flow theory to a deforming broken ice field. Acta Mechanica, 63, 143–160.
  • Similä, M., Karvonen, J., Hallikainen, M., & Haas, C. (2005). On SAR-based statistical ice thickness estimation in the Baltic Sea. Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2005 (pp. 4030–4033).
  • Strübing, K. (2009). Charting the Baltic Sea ice—A historical review. Report Series in Geophysics, 61, 11–27. University of Helsinki.
  • Swedish Administration of Shipping and Navigation, and Finnish Board of Navigation. (1982). An ice atlas for the Baltic Sea, Kattegat, Skagerrak and Lake Vänern. Swedish Administration of Shipping and Navigation, and Finnish Board of Navigation.
  • Sztobryn, M. (2009). Sea ice services in the Baltic Sea. TransNav, 3(4), 415–420.
  • Szwed, M., Dobler, A., Mezghani, A., & Saloranta, T. M. (2019). Change of maximum snow cover depth in Poland—trends and projections. IDŐJÁRÁS, Quarterly Journal of the Hungarian Meteorological Service, 123(4), 487–500.
  • Tarand, A. (1993). The Tallinn time series of break-up as climate indicator. Report Series in Geophysics, 27, 91–93. University of Helsinki.
  • Thorndike, A. S., Rothrock, D. A., Maykut, G. A., & Colony, R. (1975). The thickness distribution of sea ice. Journal of Geophysical Research, 80, 4501–4513.
  • Tinz, B. (1996). On the relation between annual maximum ice extent of the ice cover in the Baltic Sea and sea level pressure as well as temperature field. Geophysica, 32(3), 319–341.
  • Udin, I., & Ullerstig, A. (1976). A numerical model for forecasting the ice motion in the Bay and Sea of Bothnia (Research Rep. No. 18). Winter Navigation Research Board.
  • Vihma, T., & Haapala, J. (2009). Geophysics of sea ice in the Baltic Sea: A review. Progress in Oceanography, 80, 129–148.
  • Wang, K., Leppäranta, M., Gästgifvars, M., Vainio, J., & Wang, C. (2008). The drift and spreading of Runner-4 oil spill and the ice conditions in the Gulf of Finland, winter 2006. Estonian Journal of Earth Sciences, 57(3), 181–191.
  • Weeks, W. F. (1998). Growth conditions and the structure and properties of sea ice. In M. Leppäranta (Ed.), Physics of ice-covered seas (Vol. 1, pp. 25–104). Helsinki University Press.
  • Weeks, W. F. (2010). On sea ice. University of Alaska Press.
  • Weeks, W. F., & Ackley, S. F. (1986). The growth, structure, and properties of sea ice. In N. Untersteiner (Ed.), The geophysics of sea ice (pp 9–164). NATO ASI Series. Springer.
  • World Meteorological Organization. (2014). WMO sea-ice nomenclature, terminology, codes and illustrated glossary. WMO/OMM/BMO 259.
  • Zubov, N. N. (1945). L’dy Arktiki [Arctic ice]. Izdatel’stvo Glavsermorputi.