Consolidation of first-year sea ice ridges

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. C6, 3062, /2000JC000526, 2002 Consolidation of first-year sea ice ridges Knut V. Høyland 1 The University Courses on Svalbard, Longyearbyen, Norway Received 6 July 2000; revised 9 March 2001; accepted 11 June 2001; published 28 June [1] Measurements of spatial and temporal temperature development, geometry morphology, and physical properties in three first-year sea ice ridges at Spitsbergen and in the Gulf of Bothnia have been performed. The corresponding thickness and the physical properties of the surrounding level ice were also measured. The thickness of the consolidated layer was examined through drilling and temperature measurements: the temperatures gave a ratio of the thickness of the consolidated layer to the level ice thickness from 1.39 to 1.61, whereas the drillings indicated a ratio of The measured consolidated layer appeared to be 28% thicker when based on drillings in comparison to temperature. Thus the result depended on the method of investigation; the drillings included a partly consolidated layer. However, the measured growth of the consolidated layer did not depend on the method of investigation. The scatter of the physical properties in the consolidated layer was higher than that of the level ice. The consistency of the unconsolidated rubble differed markedly at the two sites. It was soft and slushy at Spitsbergen and harder in the Gulf of Bothnia. Three possible explanations for these differences are discussed: surrounding currents, different keel shapes, and difference in salinity. INDEX TERMS: 4540 Oceanography: Physical: Ice mechanics and air/sea/ice exchange processes; KEYWORDS: ice ridges, consolidation, thermodynamics 1. Introduction [2] Sea ice ridges are formed by compression or shear in the ice cover. The ice cover is broken, and a pile of broken ice, water, snow, and air is created. The volume over the water-line is called the sail, and the pores in this part are filled with air or snow. Most of the volume is below the waterline, the keel. Ridges are important ice features from engineering as well as from a geophysical point of view. From a geophysical point of view the ridges are important as they account for a large part of the ice volume and are essential parts of an ice cover. The ridging process can be considered an energy sink and creates an uneven surface and bottom of the ice cover. It increases the concentration of ice in some parts and creates open leads in other places. This changes the drag from winds and currents and makes ridging and ridges important to large-scale constitutive models for sea ice. Ridges are zones of material inhomogeneity in the ice cover, and they represent zones of strength or weakness in the ice cover depending on their degree of consolidation. The consolidation process thus also becomes of interest for mechanical sea ice models. For the engineer, ridges may represent the design load for ships, coastal, and offshore structures in many arctic and subarctic marine areas. However, it is not clear what load a first-year ridge can exert on a given structure or how the ridge deforms. It depends on the age and composition of the ridge as well as 1 Formerly at Department of Structural Engineering, University of Science and Technology, Trondheim, Norway. Copyright 2002 by the American Geophysical Union /02/2000JC the structure. Different ridge failure modes observed at the Molikpaq platform are presented by Timco et al. [1999]. For a large number of the ridges the first failure occurred in the level ice behind the ridge. The ridge itself failed in different modes: shear, bending, or spine failure. They report that the measured loads were lower than those given by different load algorithms. The loads are usually calculated by assuming that the load contributions from the sail, the consolidated layer, and the keel can be found independently and then added together [see e.g., Blanchet, 1998]. Increased knowledge about the internal structure of first-year ridges is necessary to improve these load estimates. The worst-case scenario for the consolidated layer is crushing. Thus its size and crushing strength are of special importance. The rubble is usually assumed to behave as an elastic plastic material with Coulomb-Mohr yield criterion until failure and as a viscous fluid in continuous deformation. [3] Ice ridges are, in general, long and curvilinear; that is, they are three-dimensional features. However, the characterization of ice ridges is often done by studying one or several cross sections. Geometrical ratios such as keel depth to sail height, keel depth to keel width, and keel area to sail area are often given (see Timco and Burden [1997] for a summary of ridge geometries). Typical keel depth/sail height ratios are 4.5/3.3 for first-year/multiyear ridges. The porosity h is another important characteristic; it is usually defined as the ratio of the volume of any non-sea ice material to the total volume. As the season proceeds, the porosity decreases because the water in the pores freezes up. This creates a frozen zone that grows downward in the ridge; that is, the ridge is consolidating. The consolidation process in first-year ridges is the main focus of this paper, 15-1

2 15-2 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Figure 1. The different fluxes and heat transfer mechanisms in an ice ridge. and the aim is to present field measurements and discuss these in relation to basic thermodynamic processes and to compare with results published in the literature. Models to calculate ridge loads on structures need input of ice strength, and the strength of ice is dependent on the temperature. Thus vital input into these models is the thickness as well as the temperatures and the physical properties of the consolidated layer. 2. Thermodynamics of Ridges 2.1. Thermal Processes and Energy Transport [4] The consolidation of a ridge is a complex process involving both conductive and convective heat transfer. Mechanical erosion may also be of importance. Four different energy transfer fluxes can be identified: heat transfer up into the surroundings, q sur ; flux up through the consolidated layer, the sail, and the snow cover, q c ; internal redistribution of energy, q re ; and heat transfer from the ocean, q ocean. From Figure 1 we see that different mechanisms are responsible in different parts of a floating ridge. In a grounded ridge, a stamucha, it is also necessary to take into consideration the temperatures in, and the heat flux from, the sea bottom. The surrounding heat transfer q sur is a combination of sensible, latent, longwave, and shortwave radiation; q c is basically a conductive term; and q re is a mixture of conduction and convection, whereas the oceanic flux is a convective term. [5] One can distinguish between three different time phases: an initial phase, a main phase, and a deterioration phase. The deterioration phase will not be discussed in the following. The internal redistribution and the flux from the ocean are crucial in the initial phase. Cold ice blocks are submerged into warm water, and energy is exchanged. The ice surfaces exposed to the surrounding water gain energy from the ocean, whereas the fluxes through the internal surfaces redistribute the energy internally in the keel. Because of the permeability of the keel, convective heat flux may also be added through the internal surfaces. The temperatures in the keel approach the freezing point of the ridge material (T f ). However, energy is still being transferred between the ice and the water/slush in the pores, but it is now consumed by phase change and is therefore not reflected as a change in temperature. The duration of the initial phase is a function of the oceanographical conditions, the permeability of the keel, the initial temperature, and the thickness of the ice blocks. At the end of the initial phase the keel can be divided into two parts: a cold upper frozen part (the consolidated zone) and a lower unconsolidated part in which the temperatures are close to T f. In the main phase the energy transport up through the consolidated layer and out through the air-exposed surface is the dominating process. This phase continues throughout the cold season and is governed by the meteorological conditions. The heat transfer q sur during the growth season (excluding liquid precipitation) can be divided into a turbulent part consisting of the sensible and the latent flux, and a radiation balance term consisting of longwave and shortwave radiation. This gives the following terms: q sur ¼ q sens þ q lat þ q long þ q sol ; where q sens is the sensible heat flux, q lat is the latent heat flux, q long is the longwave heat radiation, and q sol is the shortwave or solar radiation. [6] The shortwave radiation (q sol ) is a function of latitude, day, time, and cloudiness [Ashton, 1986], whereas the three other terms (q sens, q lat, and q long ) can be estimated from the air temperature, the surface temperature, the wind speed, the relative humidity, and the cloudiness [see, e.g., Løset, 1992; Maykut, 1986]. The surface temperature is also crucial for the conductive flux up through the consolidated layer, the sail, and farther up through the snow. The temperature gradient through these layers, as well as the layers size and thermal properties, determines q c. The snow cover has an important insulating effect that has to be taken into account in any model created to estimate this flux. It is a multilayer complex structure in constant transformation, and precise measurements are necessary to characterize it properly [Granberg, 1998]. A nonsymmetrical snow drift around a ridge creates a nonsymmetrical consolidated layer [Høyland and Løset, 1999a]. I assume that most of the energy that is transported upward (q c ) and into the cold air originates from latent heat that is released as the pores freeze up Consolidated Zone [7] The formation and growth of the consolidated zone is the most striking seasonal development of the internal structure of a first-year ice ridge. However, it is not clear how the consolidated layer should be defined, nor does a standardized method for examining the thickness of this layer seem to exist. The waterline is often used as an upper boundary as the blocks are loosely bound in the sail. The lower boundary is more difficult because there exists a partly consolidated layer beneath the fully refrozen one. This partly consolidated layer is a porous high-temperature zone with less mechanical strength than the fully refrozen one, but the strength is probably higher than that of the unconsolidated rubble. The determination of the lower boundary can be done either in a thermal or in a mechanical sense. If the thermal one is chosen, then the temperature becomes the prime indicator of consolidation. The temperatures in the consolidated layer are said to be below the freezing point. Measurements can be made by coring or by ð1þ

3 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES 15-3 installing thermistor strings. The thermal definition has the advantage that it is precise and that it is clearly related to measurements. It is also relatively easy to do continuous measurements. The disadvantage is that it excludes the partly consolidated layer and that it is more time and money consuming than drilling. The thermal definition seems to estimate a thinner consolidated layer than what is found by drilling. This is probably because a thermal definition excludes the partly consolidated zone. A mechanical definition will be related to a pronounced drop in some mechanical strength at a certain depth level. The best way to examine this is to take samples from different depths and to perform, for example, uniaxial compression tests. However, this is even more time and money consuming, and so, mechanical drilling is often used. The major advantage of drilling is that it is quick to perform. It is, however, less precise, as it to some extent depends on the driller. Drilling is also a destructive method, and so, it becomes more difficult to examine the temporal development of the consolidated layer. There are also other methods, such as the Arctic and Antarctic Research Institute s (AARI) hot point needle, and other thermal drilling systems. For a discussion of thermal versus mechanical drilling to examine ridges see Mironov et al. [1999]. Uniaxial compression tests, without temperature corrections, seem to be in correspondence with temperature measurements [Høyland et al., 2000]. [8] The internal structure of a ridge is not directly observable; hence it is of great interest to relate it to more easily observable quantities such as meteorological and oceanographical conditions and undeformed level ice conditions. Numerical two-dimensional finite element simulations of conductive heat transfer within the ridge and with the meteorological conditions giving the upper boundary conditions (q sur ) with the use of ABAQUS have been completed. ABAQUS is a commercial finite element program that is well developed to simulate heat transfer [Hibbitt, Karlsson and Sorensen, Inc., 1996]. The simulations predict the measured growth of the consolidated layer [Høyland, 2000]. Another easy way to estimate the thickness of the consolidated layer is to compare it to the level ice thickness. Adjacent level ice is subjected to the same physical environmental factors as the ridge, but it grows at a diffierent rate. A simple analytical model based on Stefan s law has been developed by Leppäranta and Hakala [1992], and this model relates the thickness of the consolidated layer to the porosity of the ridge and the thickness of the level ice. The ratio of the size of the consolidated layer over the level ice thickness has been studied by several researchers [e.g., Kankaanpää, 1997; Leppäranta et al., 1995; Croasdale et al., 1990; Timco and Goodrich, 1988; Frederking and Wright, 1982], and the reported results are from 1.2 to 2.1. It is, however, not always clear whether a thermal or a mechanical definition is used. Measurements in three first-year ridges reported in this paper gave a ratio of when a thermal definition was used and when the drilling results were used. However, the growth of the consolidated layer was not dependent on the method of investigations and ranged from 0.5 to 1.0 cm d Deterioration of the Keel [9] A less dramatic but still important process is the deterioration of the lower part of the keel. As we have seen, energy is redistributed internally (q re ) and gained from the ocean (q ocean ). Therefore new ice crystals may be formed in the pores at the same time as the blocks become softer. If there is enough time, I assume that the blocks and the pores will approach a slushy consistency. The importance of this process is not yet elucidated; more experiments and theoretical work are needed. The salinity of the seawater and thereby also the porosity of the level ice, the permeability, the shape of the keel, and the surrounding oceanographical conditions may be of importance for the process. When the ice blocks are submerged, water penetrates into the vacated brine channels or any other porosity. This increases the effective area through which energy can be transferred and thereby the speed of the deterioration process. The speed and temperature of the surrounding currents, together with the shape, the roughness, and the permeability of the keel, give the oceanic flux, and this may be of high importance. The surface divided by the volume of the keel is also important as it determines the energy/volume that is added to the keel. The oceanic flux for level ice has been examined, and Maykut and Perovich [1987] suggest 2 W m 2 as an annual average for the Arctic basin. These estimates are done for level ice, but the flux into a ridge keel may be higher because of the permeability and the exposed area/volume of the keel Porosity [10] Two levels of porosity can be identified: macroporosity h and total porosity h t. The macroporosity h is the ratio of volume of any non-sea ice material to the total volume. This is what is usually meant by the porosity of a ridge. This porosity is commonly measured by drilling holes along a cross section of a ridge and registering any drop of the drill as a pore. It is easily established early in the season, or as long as the blocks can be distinguished from the pores. Later on in the season it can become more problematic as the consistency of the blocks and the pores approach slush. Most of the reported values are between 30 and 35% [see, e.g., Kankaanpää, 1997; Coon et al., 1995; Leppäranta and Hakkala, 1992; Veitch et al., 1991a; Frederking and Wright, 1982]. A total porosity can be defined by including the porosity of the level ice from which the ridge is formed. In this case, there are two possibilities: either it can be defined as the ratio of the volume of any nonpure ice material (including air pockets) to the total volume or as the ratio of the volume of liquid (excluding the air) to the total volume. Measurements of this kind of porosity can be done with a thermal drill (e.g., AARI s hot point needle). Beketsky et al. [1997] report an initial hummock porosity of 44 47%, and this can represent the total porosity. Simple theoretical calculations of a total porosity can also be done by the use of the equations derived by Cox and Weeks [1983]. The porosity of young sea ice may be 10 12%. A ridge with a macroporosity of 35% made from this level ice would then have a total porosity of 42 43%. Finite element simulations of the heat transfer process in ridges support the idea of a total porosity [Høyland, 2000]. The macroporosity has been used to calculate the ridge isostacy, and the keel has been found to be too big [Kankaanpää, 1997; Leppäranta and Hakkala, 1992]. A higher porosity, as suggested with the total porosity, may help to explain the measured isostacy. [11] Seasonal development of the porosity h of the unconsolidated parts of the keel has been done. Leppäranta et al. [1995] found no change, whereas Høyland et al. [2000]

4 15-4 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES found indications of a decreasing porosity. The porosity does not seem to be evenly distributed vertically, as a midkeel minimum has been measured [Lensu, 1993]. [12] The porosity of the consolidated layer is interesting in a mechanical model. Measurements of consolidated zone porosity h done by drilling 2 00 holes are reported by Høyland and Løset [1999a]. The porosity of the consolidated and the partly consolidated layer of the ridge were found to be 3.4%, but most of these pores were located in the lower part. Little evidence was found of pores in the colder parts. This could either be because there were none, because they were smaller than could be detected with the present equipment (a few centimeters), or because they had frozen. It is obvious that small-scale porosity, such as that of level ice, also exists in the consolidated layer. There seems to be a bigger scatter of the physical properties [Høyland et al., 2001]. [13] The measured sail porosity is often smaller than the keel porosity [Timco and Burden, 1997; Leppäranta and Hakala, 1992]. The reason for this is not clear, but two reasons have been suggested; the volume of the sail may be too small compared to the block size to get proper measurements, and the blocks in the sail may be smaller than the pieces in the keel, thereby giving a more compact packing [Leppäranta et al., 1995]. It is also possible that the deterioration of the keel actually increases its porosity. The measured sail porosities ranged from 21 to 24%, which is less than the measured keel porosity (33 38%). 3. Experimental Work [14] Measurements have been done on three different first-year ice ridges; the locations are shown in Figure 2. The first ridge was instrumented in 1998 and was situated off Svartodden in the Van Mijen fjord on Spitsbergen; the second was off Camp Morton in the same fjord; and the third was off the Marjaniemi weather station on the island Hailuoto west of Oulu in Finland. The general meteorological and oceanographical conditions differ significantly and should therefore give a good basis for a comparative analysis. In the Van Mijen fjord the ice usually forms in December. The melting begins from mid-april (1999) to mid-may (1998), and the ice stays until the end of June. The seawater salinity is about 34 parts per thousand (ppt), and there are tidal currents in the Van Mijen fjord [Kangas, 2000]. The winter of 1998 was colder than that of In the northernmost part of the Bothnian Bay the ice formation starts in late October or early November, and the ice winter lasts 5 7 months [Kankaanpää, 1997]. The ice thickness reaches between 0.5 and 1.2 m [Haapala and Leppäranta, 1997], and the water salinity is around 3 ppt [Haapala and Alenius, 1994]. There are practically no tides or currents in the northern part of the Gulf of Bothnia. The meteorological record shows low pressure and strong wind around 20 January 1999, and simulations of level ice growth predict a level ice thickness equal to the block thickness of the ridge (20 cm) around this date. This means that if the ridge was made from level ice, it formed around 20 January [15] Measurements of temporal and spatial temperature development, geometry, block size and structure, salinity, and porosity were done on all three ridges, as well as on level ice. The crystal structure of the consolidated layer was examined in the two 1999 cases, and uniaxial compression tests of the consolidated layer from the ridge outside Marjaniemi were done at the University of Oulu. The measurements are presented by Heinonen et al. [2000], Kyhring [1999], Høyland and Løset [1999a], and Langeland [1998]. [16] Five to six thermistor strings were installed in a cross section of all the ridges (see Figure 3), and one was installed in the level ice (only in 1999). The strings in the ridge were 2 m apart and had 16 temperature sensors each. The level ice string was deployed close to the ridge and had 10 sensors. The vertical distance between the sensors was 0.1 m in the short strings and 0.21 m in the long ones. One data logger was attached to each string and logged the temperatures periodically as shown in Table 1. The strings, loggers, drilling equipment, and procedures are described by Løset et al. [1998]. The surface was leveled, and several (3 5) cross sections were drilled (2 00 auger) one or two times to examine geometry and porosity. Cores were taken up to measure salinity and to make thin sections to examine the crystal structure. A 1 m wide channel was cut through the sail to examine the block structure. Oceanographic measurements of currents and temperatures were provided by Kangas at The University Courses on Svalbard (UNIS) [Kangas, 2000]. The meteorological stations in Svea and at Marjaniemi provided weather data. In addition, a weather station was deployed 100 m away from the Svartodden ridge. 4. Results 4.1. Geometry, Morphology, and Porosity [17] The two ridges in the Van Mijen fjord were singular ridges, whereas the one at Marjaniemi was part of a large ridged area with 3 6 layers of rafted ice. The Svartodden ridge had a sail height h s of 1 m, the keel depth h k was 4.4 m, the keel width w k was 10 m, and the ridge was about 15 m long. The thickness of the blocks in the sail was between 0.2 and 0.25 m, and the length was between 0.5 and 1.45 m. The Camp Morton ridge was long and curvilinear, had an h s of 1 m and an h k of 5 m, and was 15 m wide. The thickness of the blocks (h b ) in the sail was m, and the length of the blocks ranged from 0.2 to 1.9 m. Visual inspection of the channels that were cut through the sail indicated that both of the Van Mijen ridges were formed in a similar way. The internal structure of the ridges at Spitsbergen indicates that the ice from one side was lifted and the other ice sheet was pushed down. The dates of formation for the two ridges are not known, but it is reasonable to assume that they were formed in December or January. The ridge at Marjaniemi did not have the idealized triangular shape. The sail consisted of m of rafted ice and was relatively flat. The keel depth was between 4 and 5 m, and the thickness of the blocks was 0.2 m. Figure 4 shows the same cross section at the beginning and at the end of the measurements for the Marjaniemi ridge. The shaded area is the keel, the white is the sail, the dotted white is the snow, and the black dots are pores. The solid line gives the consolidated layer, being defined to reach down to the first pore. [18] Porosity measurements were done by drilling 2 00 holes and registrating any drop of the drill as a pore. Cross sections with 1 m spacing were drilled, and the space between each hole along one cross section was 1 m. The keel consistency of the two ridges at Spitsbergen made it difficult to get good data for the rubble porosity. It was

5 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES 15-5 Figure 2. Map of the sites. difficult early in the season and impossible later. At Svartodden six different cross sections were drilled, four of them twice, 93 holes altogether. Three cross sections were drilled at Camp Morton, two of them twice, 43 holes altogether. At Marjaniemi five cross sections were drilled, three of them twice, 81 holes altogether. The porosity results are given in Table 2. The Marjaniemi ridge showed some decrease in the porosity during the season (1% in average). The measured porosity in the consolidated and the partly consolidated layer of 3.4% was basically situated in the partly consolidated zone. Figure 3. Thermistor strings in the ridge off Camp Morton (north is to the right) Physical Properties and Ice Texture [19] The measured salinities and densities of the consolidated layer and the level ice are shown in Tables 2 and 3.

6 15-6 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Table 1. General Environmental Conditions on the Three Sites a Svartodden 1998 (Spitsbergen) Camp Morton 1999 (Spitsbergen) Marjaniemi 1999 (Finland) Latitude, N Water salinity, ppt Surrounding currents tidal currents tidal currents little Instrumental period and FDD, C days 2 March to 12 May March to 5 May Feb. to 25 March 134 Growth Period and FDD, C days 2 March to 12 May March to 28 April Feb. to 19 March 126 Characterization cold winter warm winter... Type of ridge short linear single ridge long curvilinear single ridge large rafted ridge field Exposed area/keel volume, m 2 /m Assumed age at instrumentation, months Maximum snow depth on level ice, m Snow drift around the ridges, m a The meteorological data are from the weather stations in Svea (the Van Mijen Fjord) and at Marjaniemi. The average salinity of the ice on Spitsbergen ranged from 2.88 (late in the season) to 5.64 ppt (in the middle of the season) and was, as expected, higher than the salinity in the Finnish ice ( ppt). The average values of the ridge salinities were about the same as the corresponding level ice values, but there was a bigger variation in the ridge measurements. Some measurements were also done of the salinity in the sail, and the salinity of the sail was less than a) b) that of the consolidated layer. Density measurements were only done at Marjaniemi, and they show the same results with similar average values and higher standard deviations for the ridge. [20] Thin sections were made from the Camp Morton and the Marjaniemi ridges. One core was taken from the level ice in the Van Mijen fjord (10 March 1999), and two were taken from the ridge in the Van Mijen fjord (23 April 1999 and 12 May 1999). The ice from the ridge at Marjaniemi was sampled 23 March The level ice in the Van Mijen fjord had the regular structure, with a granular layer on top and columnar ice underneath. The grain diameter was between 0.5 and 3 mm, and the length of the columnar grains were up to 20 mm. The ice from the ridge showed a more irregular pattern, the grain diameter was mm, and the length was up to 20 mm. Most of the elongated grains were vertical or close to vertical. Boundaries between either new and old ice or two adfrozen blocks could be seen. The ice from the Finnish ridge showed the same pattern, with partly granular and partly columnar ice, but the length of the columnar grains was up to 70 mm. [21] Uniaxial compression tests were done on the consolidated layer from the Marjaniemi ridge. The results are reported and discussed by Høyland et al. [2000]; only the results related to the consolidation process are presented here (Figure 5; note that the unit for strength is MPa/10) Meteorological Conditions, Snow Cover, and Level Ice Thickness [22] The heat flux out from the ice is governed by the meteorological conditions and the snow cover. One simple way of quantifying the heat that is conducted up through the ice sheet is the cumulative freezing degree days FDD: FDD ¼ Z t t 0 ðt f T air Þdt; T f T air 0; ð2þ Figure 4. Cross sections of the Marjaniemi ridge. The water line is the zero, the top line marks the top of the snow, the second line from the top marks the top of the sail, the bottom line marks the bottom of the keel, and the black dots are the pores: (a) 26 February 1999 and (b) 25 March where T f is the freezing temperature of the water, T air is the air temperature, and t t 0 is the time period of interest. [23] Table 1 shows FDDs and snow conditions. The meteorological data are from the weather stations in Svea (the Van Mijen fjord) and at Marjaniemi. The FDDs for

7 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES 15-7 Table 2. Porosity, Salinity, and Density of the Consolidated and Partly Consolidated Layer Svartodden Camp Morton Marjaniemi Sail 24% (93 holes) 21% (6 holes) a Consolicated + partly consolidated 3.4% (93 holes) b Rubble 33% (4 holes) 35% (uncertain) 38% (81 holes) Svartodden Camp Morton Marjaniemi 25 March 28 April 23 April 12 May 23 March Salinity Average, ppt / Standard Deviation, ppt / Maximum, ppt / n 5 16/ Marjaniemi c Drained Undrained Density Average, kg L Standard Deviation, kg L Maximum, kg L n 7 7 a The drilling was done in a nearby sail, <200 m from the thermistor strings. b Most of these pores were found close to the rubble, mostly in the partly consolidated layer. c Date: 23 and 24 March. Spitsbergen were considerably higher than for Marjaniemi. However, some care should be taken when comparing FDDs for different latitudes because the influence of the solar radiation is not incorporated in the FDD expression. The solar radiation is more important on Spitsbergen because part of the growth season takes place after the vernal equinox. The snow drift around the Van Mijen ridges was not symmetric and corresponded to the dominating wind direction. The amount of snow increased in April and May, and so both sides of the sail were covered with snow. There was generally less snow on the ice outside Marjaniemi. [24] Level ice is the undeformed surrounding ice sheet. At the sites, level ice was defined as an area of undeformed ice close to the ridges ( m). Simulations of level ice growth and meteorological data [Høyland, 2000] show that the level ice outside Marjaniemi was remains of the sheet from which the ridge field was made. Kangas [2000] found some variation in the ice thickness throughout the Van Mijen fjord in 1998, and the measurements in 1999 also showed some variations (<0.1 m) in the vicinity of the ridge. This means that rafting of the ice had occurred earlier on in the season. No deformation of the ice cover was observed during the field work. The thickness of the level Table 3. Level Ice Conditions Svartodden Camp Morton Marjaniemi Manual Manual Temperature Manual Temperature h i, h i,f 1.16 a h i b Svartodden Camp Morton Marjaniemi 18 April 25 April c 8 May 14 May 23 March Salinity Average, ppt / Standard Deviation, ppt / Maximum, ppt 7 12/ n 7 9/ Marjaniemi d Drained Undrained Density Average, kg L Standard Deviation, kg L Maximum, kg L n 4 4 a Two centimeter increase in freeboard. b Five centimeters of imposed ice. c The samples from the top of the ice had a salinity of 12 ppt; this was probably because of flooding. The first set of numbers includes these measurements, and the second does not. d Date: 23 and 24 March.

8 15-8 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Figure 5. The physical and mechanical properties of the Marjaniemi ridge. The dotted line with diamonds is the salinity (ppt), the solid line with squares represents the compression tests (MPa/10), the solid line with triangles is the drained density (kg L 1 ), and the dotted line with triangles is the undrained density (kg L 1 ). ice was measured manually with a meter stick and with thermistor strings (not in the Svartodden case). Table 3 sums up the measurements. The freeboard was about zero for the two Van Mijen ridges, and superimposed ice was formed. Off Marjaniemi the freeboard was positive (3 5 cm) the entire time Temperatures and Fluxes in the Ridges [25] The temperature development in the three ridges was measured as described in section 3. Figure 6 shows the average temperature profiles for two strings in the Svartodden ridge: string F was situated close to the level ice and was in the thinner part of the ridge, whereas string B was situated in the thicker part of the ridge. The temperature profiles for the other strings and the other ridges are similar. The average flux through the consolidated layer can be estimated by using these temperature profiles and Fourier s law. Figure 7 shows the weekly average flux up through the consolidated layer for each string in the three ridges. However, this is not the flux q c described in section 2.1, as we do not have good measurements of the temperature profiles through the sail and the snow cover. The amount of snow increased in weeks 11 and 12, and this resulted in a better insulation and thus also a lower flux up through the Svartodden ridge Consolidation, Keel Consistence, and Oceanographic Conditions [26] Table 4 displays some key features from the measurements of consolidation such as the growth of the consolidated layer (h c ), its relation to the growth of the level ice (h c /h i ), the growth of the consolidated layer per day (h c /day), the average final thickness of the consolidated layer (h c ), its relation to the final thickness of the level ice (R avg = h c /h i ), the ratio of the maximum thickness of the consolidated layer to the final thickness of the level ice (R max = h c,max /h i ), the average maximum spatial variation of h c, and the duration of the growth. Each ridge has two values for every quantity, one derived from the temperature measurements and one from the drillings. The thickness of the consolidated layer estimated by the temperatures was defined so that the consolidated layer reached down to the lowest node in which T < T f. The consolidated layer estimated by drillings was done by feeling the change in resistance as the drill bit entered the unconsolidated rubble. This was quite easy at Spitsbergen because of the soft rubble consistency. At Marjaniemi this was not so easy so the consolidated layer was defined to reach down to the first pore, or when the driller could feel that the ice became softer, whichever occurred first. [27] The spatial variability of the thickness of the consolidated layer is given as the average maximum difference between the thickest and the thinnest point of the consolidated layer within one cross-section for the different ridges in Table 4. The solid line in Figure 4 gives the lower limit of the consolidated layer defined by drillings. The variation was generally less for the temperature-derived values than for the results obtained from drillings. [28] Figure 8 shows the seasonal development of R avg for our three ridges as well as for ridges from the literature [Blanchet, 1998; Leppäranta et al., 1995]. The ratio increased throughout the season for all of the ridges. [29] The consistency of the keels was quite different at the two latitudes, soft and slushy in the Van Mijen fjord and harder in the Gulf of Bothnia. It was difficult early in the season, and impossible later on, to get good porosity measurements of the rubble in the Spitsbergen ridges. More slush came up from the boreholes in these two ridges. For the Marjaniemi ridge it was relatively easy to examine rubble porosity also late in the season. [30] Kangas [2000] did measurements of water temperatures. He found that the temperatures in the water masses were homogeneous around the freezing point in the fjord in late March 1998 and that an intrusion of warm coastal water (up to T = 1.6 C) entered the fjord by late April 1999.

9 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Physical Properties and Ice Texture [33] The salinity and density values presented in section 4.2 are comparable to what others have found (Table 5). Latent heat is released and transported upward as a pore freezes. It is reasonable to assume that the initial salinity of such a water pocket is the same as that of the surrounding seawater. Measurements of the salinity in a partly frozen pore in an ice tank ridge show a higher salinity than the neighboring ice but less than the water [Jensen et al., 2000]. Examinations of the salinity in neighboring ridges of different ages show substantial differences in the salinity: 4 7 ppt in the old ridge and 9 29 ppt in the young ridge [Høyland and Løset, 1999b]. Several measurements of the salinity in ridges have shown little difference between the salinity of level ice and that of the consolidated layer [see, e.g., Kankaanpää, 1997; Leppäranta et al., 1995; Frederking and Wright, 1982]. All these results indicate that brine is drained quite efficiently in the consolidated layer so that the brine content approaches what can be found in the sura) b) Figure 6. Weekly average temperature profiles for two of the strings in the Svartodden ridge: (a) string B (in the thicker part of the ridge cross section) and (b) string F (close to the level ice, in the thinner part of the cross section). Temperatures were measured in the ice cores from the Camp Morton ridge 12 May The partly consolidated layer had a temperature of 1.6 C, while the temperatures in the central parts were about 1.9 C. 5. Discussion 5.1. Geometry, Morphology, and Porosity [31] The two Van Mijen ridges were small or medium arctic ridges. The block thickness indicates that they were formed during one event. It is also probable that they were created from level ice, but this is not certain. The keel depth to sail height ratios correspond well with Timco and Burden [1997]. As described in section 4.1, it looked like one ice sheet had been pressed up and the other bent down. This was also found by Tuhkuri et al. [1999]. The Marjaniemi ridge was part of a large ridge area and did not have the typical triangular shape. This kind of ridged ice feature is not rare outside Marjaniemi, although the ice was less broken and there were more singular ridges in the previous and the following years. [32] The measurements of porosity correspond well with results from the literature (Table 5). None of the other authors have reported values of the porosity in the consolidated layer. As could be expected, the sail porosity is less than the keel porosity.

10 15-10 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES a) b) c) Figure 7. Weekly average fluxes up through the consolidated layer (based on the temperature profiles and Fouriers law): (a) Svartodden ridge (strings A and F are on the fringes), (b) Camp Morton (string A failed), and (c) Marjaniemi ridge. rounding level ice. The scatter of the physical properties of the consolidated layer was almost exclusively higher than for the level ice. The number of samples was limited, but it is still an indication that the consolidated zone is a more inhomogeneous material than the level ice. If this is valid also for the mechanical properties, it is of great importance in attempts to find the mechanical strength and forces applied by ridges on structures in a ridge-structure interaction Meteorological Conditions, Ice Growth, and the Snow Cover [34] The meteorological conditions around the three ridges differed significantly. The cold winter of 1998 resulted in ice

11 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Table 4. Key Features of Consolidation for the Three Ridges a Svartodden 1998 Camp Morton 1999 Marjaniemi 1999 Temperature b Drillings Temperature Drillings (93 Holes) (43 Holes) c (81 Holes) h c, m h c /h i h c /day, cm d h c, m R avg = h c, /h i R max = h c, max /h i Average maximum variation of h c, m Duration of growth, weeks Rubble consistence soft soft soft soft harder harder a Minimum values are used for the temperature-derived h c. b At Svartodden we did not get data from the level ice string, so the values of the level ice thickness are from drilling. This gives a higher value than a temperature string does, so the values for R avg and R max may be a little low. The central string failed at Svartodden and was not replaced until week 16. c At Camp Morton we did not perform so much drilling as in the other two cases, so the values from the drilling results are less reliable in this case. growth in the Van Mijen fjord until mid-may, whereas in 1999 it stopped at the end of April. The growth per day during the measuring period was also lower for the Camp Morton case. At Marjaniemi, less snow and a higher growth rate were measured. The nonsymmetrical snow drift early in the season gave a corresponding nonsymmetrical consolidated zone for the two Van Mijen ridges. These observations stress the importance of the insulating effect of the snow cover. There is some discrepancy between the level ice thickness measured manually and by temperatures (see Table 3). This may be explained by the resolution of the thermistor strings or by an uneven ice cover (the thickness measurements were not done at the exact same spot each time). In the Van Mijen fjord the discrepancy was probably due to superimposed ice. The freeboard was little in both cases, and the ice was flooded occasionally when holes were made. In the spring of 1999 the formation of 5 cm of superimposed ice was measured. The salinity measurements performed 25 April 1998 also indicate that superimposed ice was formed here. As the freeboard was all the time positive outside Marjaniemi, there is reason to believe that little superimposed ice was formed Deterioration of the Keel and Oceanic Flux [35] Table 4 shows that a marked difference in the consistence of the keel was found between the ridges at Spitsbergen and the one in the Gulf of Bothnia. Three possible factors can explain this discrepancy: (1) salinity, (2) velocity and temperature of surrounding currents, and (3) exposed area to volume ratio. These three factors indicate a more effective internal redistribution of heat and a higher oceanic flux on Spitsbergen. The relative contribution of the two fluxes to the deterioration of the keel is not clear. The internal redistribution of energy is a complex process including internal flow due to variations in density, salinity, and temperature, as well as melting and formation of ice crystals. There are several indicators of seasonal variations in the oceanic flux in the Van Mijen fjord. The measurements of Kangas [2000], and our own measurements of warm ice under the consolidated layer late in the season, indicate no flux in the cold season and some flux in the melting period. The ridge keel temperatures in an instrumented stamucha in the Van Mijen fjord in 1997 showed that it was heated from below as well as from above in late May and June. Temperatures up to 0.6 C were recorded in the lower parts of the keel. The lowest temperature was found in the middle of the keel [Løset et al., 1998]. Finite element simulations of ice growth in the Van Mijen fjord also suggest zero flux during the growth period [Høyland, 2000]. This does not mean that there is no energy transfer between the ice in the keel and the surround- Figure 8. The seasonal development of R avg versus the level ice thickness, temperatures T, and drillings D.

12 15-12 HOYLAND: CONSOLIDATION OF FIRST-YEAR SEA ICE RIDGES Table 5. Summary of Ridge Experiments a R avg h c /h i Field/ Laboratory Ridge Age, Days Salinity, ppt Porosity, % Density, kg L 1 Blanchet [1998] field Coon et al. [1995] 1.24 D 1.87 D field Croasdale et al. [1990] laboratory Frederking and Wright [1982] field, grounded Kankaanpää [1997] 1.42 D... field Leppäranta et al. [1995] 1.75 D 1.85 D field Timco and Goodrich [1988] 1.39 T 2.77 T 1.79 T 12 T laboratory Veitch et al. [1991a, 19991b] 1.25 T 1.54 T /1.35 D field Veitch et al. [1991c] 1.98 T /2.13 b laboratory a Superscripts T and D refer to whether temperatures or drillings have been used. b Visual inspection of the ridges after each test was used to estimate the thickness of the consolidated layer. ing water. If there are some currents present, it is reasonable to expect that there will be some mechanical and thermal erosion Consolidated Zone [36] Table 5 summarizes some results from the literature concerning ridge consolidation. In general, the results presented in Tables 4 and 5 correspond well; the thickness of the consolidated layer has been found to be <2 times the thickness of the level ice for the ridges examined in situ. However, Beketsky [1998] includes reports of ridge measurements from the Sea of Okhotsk. In the ridges that were 3 months old they found average values of the thickness of the consolidated zone of about 3.5 m and a porosity ranging from 15 to 28%. The thickness of the level ice is not given in the paper. [37] The warm winter on Spitsbergen in 1999 resulted in a thinner consolidated zone and a shorter duration of growth than in The growth per day was also less for the Camp Morton ridge. The ridge at Marjaniemi had a higher growth per day; this is probably due to less insulation caused by a thinner snow cover. The ridge examined by Leppäranta et al. [1995] was also situated off the Marjaniemi weather station, and their results are similar to ours. Their higher growth rate corresponds to lower porosity; a higher final value for R avg in our case is probably due to the rafted layers that gave a higher initial growth rate. [38] Examination of Table 4 reveals an important distinction. The three top rows comprise figures related to the growth of the consolidated layer, and in the three subsequent ones the values are related to the total thickness of the consolidated layer. In the latter case, there seems to be a systematic difference between the numbers derived from temperature measurements and the ones obtained from drillings. The temperature measurements predict a thinner consolidated zone. The average ratio, h c T /h c D, is 0.78 for the three ridges. This difference is not found in the three top rows. Drilling results are generally more uncertain because they rely more on the person operating the drill. However, there is no reason to believe that this uncertainty should result in a systematic overestimation. Two possible explanations can account for this difference: first, that the given temperature-derived data are minimum values and, second, that the drilling results include a partly consolidated layer. The first explanation is not sufficient to explain the difference; even if the vertical resolution of the thermistor string (0.21 m) is added to h c T, it does not reach the value of h c D ; there is a still a ratio of h c T /h c D of Timco and Goodrich [1988] and Veitch et al. [1991c] also found that the temperature estimates of the thickness of the consolidated layer were lower than what was found in a visual inspection (h c T /h c D = 0.79 and 0.92, respectively). A better examination of the mechanical definition of the consolidated layer, such as the compression tests presented in Figure 5, corresponds well with the temperature-derived values. It is generally more difficult to compare drilling results than temperature measurements and standardized mechanical testing. However, it looks as if the partly consolidated layer is easily incorporated into the estimate of the consolidated layer by drilling. [39] In Figure 8 the seasonal development of R avg for in situ ridges shows that it increased during the season, approaching a value of about 1.5 for the temperature measurements and 1.8 for the drilling results. Laboratory work has shown the opposite trend. Timco and Goodrich [1988] measured temperatures in laboratory-made ice rubble during the consolidation process. They reported the initial level ice thickness (equal to the block thickness), the final level ice thickness, and the thickness of the consolidated layer for each test. Some results derived from their report are presented in Figure 9. Figure 9 shows that R avg decreased with time. However, it approached the temperature-derived field values. Results from experiments with laboratory-made ridges at Hamburgische Shciffbau- Versuchsanstalt GmbH in Hamburg by Jensen et al. [2000] fit well into Figure 9. This demonstrates the effect of the internal redistribution of energy during the initial phase, and it strongly indicates that the initial growth of the consolidated zone may easily be overestimated in the laboratory. Høyland et al. [2001] suggest that this is a function of the temperature profile in the level ice from which the ridge is made. The temperature gradient in the thin laboratory ice may become too steep, creating a higher flux through the submerged ice blocks and thereby possibly also a higher initial growth of the consolidated layer. Different oceanographic conditions in ice tanks compared to the real world may also contribute. [40] So far, only average values for the thickness of the consolidated zone have been discussed. However, there were variations in the thickness throughout the ridge, and this is important for ridge-structure interaction predictions. As Table 4 shows, there was considerable variation in the