1. Introduction

Sea ice is a complex, composite medium that consists of such micro-structures as pure ice, bubbles, pockets of brine, and salt crystals. It also contains impurities, including ice algae, sediment, and organic matter [1,2]. These micro-structures influence the optical properties of sea ice, which determine how solar radiation is distributed in the upper ocean. The micro-structures of sea ice depend on the temperature, which can change the size, volume, distribution, and properties of brine, bubbles, and salt in it. Such changes alter the absorption and scattering of sea ice, influencing the radiation transfer of sunlight in the ice, and causing changes in the apparent optical properties of sea ice [1–4].

The characteristics of transfer of radiation in sea ice are important for understanding the transfer of solar energy among the atmosphere, sea ice, and the ocean. They can be used to evaluate the balance of energy and mass in ice regions in the context of research on the global climate. We can use the apparent optical properties of reflectance and transmittance to describe how the light field of sea ice changes to estimate its inherent characteristics of absorption and scattering. This can in turn help us analyze the physical structure and material composition of sea ice, including the impurities and micro-structures in it. It can also help us better understand the distribution of the algae in sea ice and the underlying seawater. Therefore, studying the characteristics of radiation transfer in sea ice can benefit research on the polar ecological environment.

The albedo and transmittance of sea ice are important optical parameters for research on the polar regions of the world. The albedo can be easily obtained by measuring the incident and reflected sunlight on the surface of ice. Measuring the transmittance of sea ice, which is the amount of light that passes through it, is more complex than measuring its reflectance. It requires inserting an optical detector below the ice through a small hole drilled into it. The hole should be as small as possible, and the point of measurement should be far from it to reduce disruptions to the light field. The above method has been widely used to measure values of the transmittance and Kd of sea ice [5–9]. However, the micro-structure and material composition of sea ice vary along its depth, so the value of Kd obtained from this method cannot be used to determine its values (Kd(z)) at different depths inside the body of ice.

It is difficult to measure the distribution of light inside sea ice because optical detectors cannot be freely set at different positions in it. This has led to a lack of in-situ data on radiation transfer inside sea ice, which hinders research on its optical properties. Some researchers have measured the irradiance profile Ed(z,λ) of sea ice in the Arctic by employing divers and using the vertical increment method to calculate the coefficient of attenuation Kd (λ) of different layers of ice [10]. Pegau and Zaneveld [11] designed a device that can collect data on the downward radiation based on the Lambertian reflection to measure the radiation profile inside ice. This device is easy to use, and is placed at different depths inside a hole drilled through the ice to measure the downward radiation (Ed(z,λ)). Similar devices have been used in many field observations of the optical properties of sea ice [7,8,12,13]. However, this method requires breaking the sea ice, and the measurement is affected by the shadow of the optical detector. The accuracy of measurement of Ed(z,λ) by using this method has yet to be confirmed as well. Another method involves using ice-tethered buoys fixed in the layers of ice for the long-term observation of its optical properties. This method has been applied to field observations of sea ice in recent years [14–21]. However, such buoys are equipped with large detectors that cannot measure the profile of sea ice at a high resolution in environments in which the layer of ice is thin. The shadow of the detector also affects the measurements in this case.

The Kd is recognized to be closely related to the light transparency sea ice, which plays a critical role in the energy balance and biological processes of the upper ocean [22–24]. However, due to the limitations of observation techniques as the above discussed, few studies have reported the spatiotemporal variability of Kd [25,26]. To overcome this challenge, we developed a special instrument for measuring Kd inside sea ice and conducted three experiments with it, two in the field and one in the laboratory. We analyzed the spatial variation of Kd along the vertical axis, and then examined the temporal variation of Kd at different depths. Finally, the driving mechanism of the spatiotemporal changes of Kd was elucidated, by analyzing the microstructure and the internal bubbles of sea ice.

2. Methods

2.1 Design instrument

Sea ice is a solid substance, because of which measuring its optical properties is challenging. Researchers cannot set detectors at any position inside sea ice without drilling holes into its surface. One method that circumvents this requirement involves setting multiple optical detectors at different depths in seawater before it freezes. Once the water freezes, the irradiance of sea ice can be measured to obtain the characteristics of optical attenuation of its profile. However, conventional spectral equipment includes large optical detectors that cast shadows on one another when they are frozen in sea ice. This causes errors in measurement that increase in magnitude as the thickness of sea ice decreases. Our research team has developed an instrument with a high spectral resolution to measure the profile of sea ice and the irradiance in atmosphere to address this issue. This instrument uses Maya 2000 spectrometers (manufactured by Ocean Optics Corporation, USA), which have a spectral range of 200–1100 nm, a resolution of 0.5 nm, and an integration time of 5–3000 ms. A micro-industrial computer is used to control the spectrometers and a rechargeable lithium battery power the entire system. We designed software to automatically set the time of spectral integration to improve the signal-to-noise ratio of the spectrometers. All spectrometers are connected to each optical detector, which is fixed on a bracket. This instrument can be used to simultaneously measure the incident solar irradiance in the atmosphere as well as at several points inside the sea ice. Figure 1 shows a schematic diagram of this system.

 

Fig. 1. Schematic diagram of the system.

Download Full Size | PDF

The authors of this study develop a system of instruments with a high spectral resolution to measure the profile of sea ice. Traditional spectral acquisition detectors are large because they integrate gratings, photosensitive elements, and motors. We use a fiber optic spectrometer as the core unit of our instrument and connect it to the detector by using an optical fiber. This reduces the size of the detector and eliminates spectral drift caused by the low temperature of the environment. The irradiance detector is waterproof down to a depth of 50 m. The detector and the spectral acquisition unit are separated, as shown in Fig. 2(a). The detector has a diameter of 14 mm and a length of 12 mm, as shown in Fig. 2(b). From left to right, the figure shows a coin, a micro-irradiance detector (4 detectors were used in this study), a micro-radiance detector, a mini-radiance detector, a mini-irradiance detector, and a mini-downward irradiance detector for sea ice. The optical detectors were arranged in order of increasing size. All these detectors were designed by our research team. The micro-irradiance detector was used here to simultaneously measure the incident solar irradiance in the atmosphere and at several depths inside sea ice.

 

Fig. 2. a. Mechanical design of the micro-irradiance detector. b. From left to right: a coin, a micro-irradiance detector (used in this study), a micro-radiance detector, a mini-radiance detector, a mini-irradiance detector, and a mini-downward irradiance detector for sea ice. All the irradiance about the Ed(z1, λ), Ed(z2, λ), Ed(z3, λ), and Ed(z4, λ) were measured by the kind of micro-irradiance detector.

Download Full Size | PDF

2.2 Field site

The Liaodong Bay is the northernmost part of the Bohai Sea in China, and the region of ice region at the lowest latitude in the Northern Hemisphere [27]. Water in the area freezes from December to February [28]. The sea ice in the area is only 5–30 cm thick, and is much thinner than that polar sea ice [29]. The sea ice in Liaodong Bay is young, and does not form clear vertical boundaries like the multi-year bare ice in the Arctic [13]. We used flat and fixed shore ice as the experimental object to ensure the safety of researchers, although most of the sea ice in Liaodong Bay is drift ice. The location of the experimental site is shown in Fig. 3.

 

Fig. 3. Map of the study area in Liaodong Bay, China. The area in which the values of Kd were observed in 2018 and 2022 is marked by the star.

Download Full Size | PDF

3. Results and discussion

3.1 Spatiotemporal variability of Kd in field in 2018

We conducted field experiments on the sea ice in the Shanhai Square in Liaodong Bay from January 18 to January 27, 2018. We used five micro-irradiance detectors that were fixed on a stainless-steel bracket to measure the irradiance at depths of 3.8 cm, 10.1 cm, 15.1 cm, and 31.5 cm below the surface of sea ice. A large hole was drilled into the ice and the bracket was lowered into the seawater beneath it (Fig. 4). The sea ice was then allowed to grow naturally, and the irradiance Ed(z,λ) was measured once or twice a day between 9 am and 5 pm.

 

Fig. 4. a. Photograph of detectors used to measure the profile of irradiance of sea ice and the bracket used to fix them. B. Holes drilled on the surface of ice and detectors set in the seawater below it.

Download Full Size | PDF

Having measured Ed(z,λ), we derived the value of Kd(λ) of each layer of ice as follows:

The detector at the bottom layer of water was set too deep, and thus did not freeze within the sea ice such that data for this layer were invalid. We thus analyze only the data on the two upper layers of ice. The surface layer (first layer) had a mean attenuation coefficient (Kd12) of 7.58 m⁻¹ while the sub-surface layer (second layer) had a mean attenuation coefficient (Kd23) of 6.88 m⁻¹. There was thus a large variation in the coefficient of attenuation of sea ice along its depth. For example, the value of Kd12 was 11.23 and that of Kd23 was 4.16 in the final measurement, where this is approximately a threefold difference. We assumed that once the sea ice had grown, there was scant exchange of material composition between sea ice and external environment, and the distribution of the light field in ice could thus be determined based only on its internal material composition. The system of attenuation of sea ice should subsequently have changed only slightly. However, the results of our measurements showed that Kd changed significantly, with a variation in its amplitude of 1 m⁻¹ (Fig. 5(a)). In particular between the 10^th and 11^th instances of measurements, between which the interval was only one day, Kd12 increased from 7.08 m⁻¹ to 24.32 m⁻¹, by 2.44 times, while Kd23 increased by 5.59 times from 3.73 m⁻¹ to 24.59 m⁻¹. The sub-surface layer thus recorded a larger increase in the amplitude of Kd than the surface layer. The variation characteristics of Kd were similar to the previous laboratory experiment conducted by Perovich el.al [30]., but the variation range of our measured values was much larger. The differences in the components of the seawater used for freezing into sea ice and the different experimental conditions, may both cause the variations in the intensity of the Kd change between ours and Perovich el.al.’s experiments. Moreover, the values of Kd at different points in the sea ice were not completely correlated over time due to the influence of the climate environment. For example, the values of Kd12 and Kd23 did not exhibit the same trend of variation. The R² of the values of Kd of the two layers was 0.55, was significantly influenced by the large value of the data from 12^th measurement (as the peak in Fig. 5(a)), and was thus unreliable. Once the data from 12^th measurement had been eliminated and the correlation coefficient was calculated once again, its value of R² was 0.40 (Fig. 5(b)). This shows that variations in Kd over time were not strictly consistent between the layers of ice.

 

Fig. 5. a. Trend of variations in the values of Kd12 on the surface layer and Kd23 in the sub-surface layer of sea ice in 2018, the X-axis is the number of measurements. B. The graph of correlation between Kd12 and Kd23.

Download Full Size | PDF

3.2 Spatiotemporal variability of Kd in laboratory in 2021

Laboratory experiment has the advantage of being artificially controllable. To verify how Kd varied in different layers of sea ice in vertical space and time, we conducted another experiment in the laboratory in 2021. A rectangular tank was insulated with a 5-cm-thick layer of cotton and was placed in a refrigerator. A low-voltage DC LED lamp (100 W) was placed above the tank as the source of incident light. Four micro-irradiance detectors were set at distances of 10.9 cm, 14 cm, 17 cm, and 21.8 cm from the upper edge of the tank to measure the irradiance at different depths inside the sea ice once it had formed. We filled the tank with seawater and set the refrigerator to freezing mode to control the growth of sea ice. The incident irradiance in the atmosphere and sea ice were then synchronously measured. The physical structure of the sea ice had a decisive influence on its diffuse attenuation coefficient, and was altered by adjusting the temperature of the refrigerator at different intervals. The experiment lasted for 12 days, and Fig. 5 shows its results.

The laboratory experiments yielded 110 sets of measurement data. The above figure shows that Kd increased gradually during the initial growth of sea ice. The mean value of Kd12 of the surface layer was 11.88 m⁻¹ while the mean value of Kd23 of the sub-surface layer was 11.35 m⁻¹ (see Fig. 6(a)). This pattern was similar to the results of our outdoor experiment in 2018. The 68th measurement yielded the most significant difference in values of Kd along the depth of sea ice, with values of Kd12 and Kd23 of 9.37 m⁻¹ and 12.62 m⁻¹, respectively. Kd23 was thus about 1.35 times larger than Kd12, and this confirms the variation in Kd along the depth of sea ice. Kd also varied over time. The first layer of sea ice came into direct contact with the air and exchanged substances with it, because of which its values of Kd fluctuated to a greater extent than that of the lower layer. The R² between Kd12 and Kd23 was only 0.27, which suggests that the physical structures of the two layers had changed inconsistently such that there was no correlation between values of Kd over time. This confirmed that the values of Kd of the surface and sub-surface layers were not strictly consistent over time. However, both the field and the laboratory experiments lacked data on the micro-structure of sea ice, and thus could not be used to explain the relationship between its physical structure and values of Kd over time.

 

Fig. 6. a. The trend of variations in Kd12 and Kd23 on the surface layer of sea ice and in its sub-surface layer in 2021, the X-axis is the number of measurements. b. The graph of correlation between Kd12 and Kd23.

Download Full Size | PDF

3.3 Spatiotemporal variability of Kd in field in 2022

We confirmed the spatiotemporal variability of sea ice Kd through a field experiment in 2018 and a laboratory experiment in 2021. However, the mechanism behind this variability remained unclear. To address this issue, we conducted another field experiment in Liaodong Bay in 2022. In this experiment, we not only repeated the previous measurements of the spatiotemporal changes of Kd, but also measured the physical microstructure of sea ice as an additional measure. Five detectors were inserted at depths of 4 cm, 10 cm, 15 cm, 23 cm, and 32 cm below the seawater according to the method of installation used in the experiments conducted in 2018 that has been described above. Measurements were conducted two to eight times a day between 9 am and 5 pm, with an interval of more than one hour between them. Samples of sea ice near the research station were collected on each day following the measurements. All samples were brought back to the laboratory at the State Key Laboratory of Coastal and Offshore Engineering of the Dalian University of Technology, and were analyzed to determine their physical microstructure structure.

We obtained 48 sets of valid data on the radiation profile of sea ice by using the above method. The mean value of Kd12 of the surface layer of sea ice was 8.73 m⁻¹ while that of Kd23 of the sub-surface layer was 12.41 m⁻¹(see Fig. 7). The value of Kd of the surface layer was thus smaller than that of the sub-surface layer, where this is the opposite result to that of experiments conducted in the field experiments in 2018 and the laboratory in 2021. In addition, the value of Kd changed significantly along the depth of sea ice, especially in the later stage. For instance, the value of Kd12 of 39th measurement was 5.70 m⁻¹ and that of Kd23 was 17.7 m⁻¹, where this is nearly a threefold difference (see Fig. 7(a)). Figure 7(b) shows the correlation between Kd12 and Kd23 over time. The value of R² was only 0.008. Our results showed that there was no correlation between the values of Kd of the adjacent layers over time. This result is similar to those of experiments in the field experiments in 2018 and the laboratory in 2021.

 

Fig. 7. a. The trend of variations in Kd12 and Kd23 on the surface and the sub-surface layers of sea ice in 2022, the X-axis is the number of measurements. B. The graph of correlation between Kd12 and Kd23.

Download Full Size | PDF

4. Discussion

The driving mechanism of spatiotemporal variability of Kd will be discussed in this section. The microstructure of sea ice was observed by using the Fedorov stage in laboratory. The samples of ice were divided into segments, each with a length and width of 5 cm. One side of each sample was ground flat, attached to a piece of glass with a temperature slightly higher than 0 °C, and was left in a low-temperature environment to freeze. The ice sample froze a few minutes later, and was fixed on the glass. This fixed sample was then cut into thin slices (thinner than 1 mm) with a planer. These slices were placed on the Fedorov stage to observe their micro-structure, and were photographed under polarized light and natural light.

We examined the micro-structure of samples of sea ice exhibiting the largest variations in Kd recorded on adjacent days (number 7th and 8th, as shown in Table 1). A Fedorov stage was used to observe the slices of ice as described above. Figure 8 shows slices 7B-5 and 8B-5, which were collected from the surface layer of sea ice in the 7th and 8th instances of sampling, respectively. The crystals of 7B-5 (Fig. 8(a)) were smaller and more elongated than those of 8B-5 (Fig. 8(c)). The ice crystal sizes of both samples decreased slightly with depth. The crystals of 8B-5 were also strip shaped but were irregular, and became more regular with increasing depth below the surface. It is clear from images of slice 7B-5 that the high environmental temperature at the time at which the sea ice was collected caused the wall of brine and bubbles to melt, and even the middle ice of this medium would melt, and then these mediums connected together.

 

Fig. 8. a. Photograph of the crystal structure of the surface layer of sea ice, with a thickness of 5 cm, at the 7th instances of sampling (slice 7B-5). b. Photograph of the distribution of brine and bubbles in slice 7B-5. c. Photograph of the crystal structure of the surface layer of sea ice, with a thickness of 5 cm and a volume fraction of bubbles of 6.41% (The volume fraction of bubbles refers to the ratio of the volume of bubbles inside the sea ice to the volume of the sea ice), at the 8th instance of sampling (slice 8B-5). d. Photograph of the distribution of brine and bubbles in sea ice in slice 8B-5, with a volume fraction of bubbles of 0.96%.

Download Full Size | PDF

Table 1. Values of Kd of the surface and sub-surface layers of sea ice, and the volume fractions of bubbles in these layers.

View Table

Sea ice is a material that can caused light to become highly scattered [25,26,28]. Its value of Kd depends on both its internal, highly absorptive substances (such as particles, dissolved colored materials, and algae) and highly scattering bubbles in it. However, sea ice exchanges brine and air bubbles with the atmosphere above it and the seawater below it. The particles in ice generally maintained their original positions. Therefore, the total absorption of sea ice changed only slightly, and its coefficient of scattering mainly determined the changes in the values of Kd. Both slices 7B-5 and 8B-5 mainly contained columnar and needle-shaped bubbles (Figs. 8(b) and (d)), while circular and nearly circular bubbles were rarely observed. We also calculate the percentage content of bubbles in the slice. The bubble volume fraction of slice 7B-5 was 6.41%, which was the highest among all the samples collected, and the corresponding Kd was 12.14 m⁻¹ (Table 1), also the maximum attenuation coefficient among all collected samples. By contrast, the bubble volume fraction of slice 8B-5 was 0.96% and its value of Kd was 5.70 m⁻¹.

Figures 9(a) and (c) show graphs of the trends of changes in values of Kd and the volume fraction of bubbles in the samples of sea ice. The two parameters exhibited similar trends of change. Figures 9(b) and (d) show the results of the analysis of the correlation between the values of Kd and changes in the volume fraction of bubbles in the sea ice. The surface layer had a value of R² of 0.52 between values of Kd12 and the content of bubbles (Fig. 9(b)), while the sub-surface layer had a value of R² of 0.47 (Fig. 9(d)). This reflects a suitable correlation between the values of Kd and the volume fraction of bubbles in the sea ice. Once the sea ice had grown such that the climate environmental changes could not influence its inherent characteristics of absorption, the volume fraction of bubbles inside the sea ice dominated temporal variations in the values of Kd.

 

Fig. 9. Relationship between values of Kd and the contents of bubbles in different layers of sea ice. (a) Graph of the trend of changes in Kd12 and the volume fraction of bubbles in the surface layer of sea ice. (b) Linear regression of values of Kd12 and the volume fraction of bubbles in the surface layer, and the values of R². (c)–(d) Relationship between values of Kd and content of bubbles in the sub-surface layer of sea ice.

Download Full Size | PDF

The results of our three experiments showed that the values of Kd varied significantly in different layers of sea ice due to differences in their material composition. The values of Kd varied by as much as two times with the depth of sea ice. However, the trend of variations in it in different layers of sea ice was inconsistent with changes in the climate environment, and no correlation was observed in certain cases. Once the growth of sea ice had been completed, it became difficult for its components to exchange substances with the environment (such as the atmosphere above it and the seawater below it). In particular, the upper layer of sea ice exchanged brine and bubbles with the atmosphere and seawater, while the particles fixed in the ice generally remained in their original positions. Therefore, the total absorption by sea ice changed only slightly, and could even be regarded as constant. The bubbles in sea ice caused light to scatter significantly, and thus determined the value of Kd. When environmental changes caused the volume fraction of the bubbles to increase, the scattering of light was enhanced and a larger value of Kd was obtained. By contrast, the scattering of light decreased with the volume fraction of the bubbles in sea ice, and this led to smaller values of Kd.

5. Conclusions

The Kd is recognized to be closely related to the light transparency sea ice, which plays a critical role in the energy balance and biological processes of the upper ocean. To study how Kd varies in space and time, the authors of this study designed a radiometric instrument with a high spectral resolution to measure the irradiance profiles of sea ice and the irradiance in the atmosphere. The instrument contained a fiber optic spectrometer as the core unit that was connected to independent spectrometers and detectors by optical fibers. This design reduced the influence of the shadow of the detector on the measurements and eliminated the spectral drift caused by the low temperature of the sea ice. The instrument also contained a bracket to fix the detectors at different positions to enable the measurement of downward irradiance at different depths of sea ice. This design enables us to successfully measure sea ice Kd, which is challenging due to the solid nature of sea ice.

We conducted three Kd experiments with the instrument: two in-situ experiments in Liaodong Bay in 2018 and 2022, and one laboratory experiment in 2021.The results showed that Kd varied significantly within the depth of sea ice, and its values in adjacent layers differed by up to two times. Moreover, Kd changed over time due to changes in the climate environment. For example, its value on the surface of sea ice at different times differed by up to 1.38 times in 2022. Different layers of sea ice also had different trends of temporal variations in the values of Kd. The R² of Kd between adjacent layers over time was only 0.008.

In order to explain the driving mechanism of spatiotemporal variability of Kd, an additional experiment focusing on the physical microstructure of sea ice was conducted in Liaodong Bay in 2022. The result shows that the changing of air bubbles in sea ice maybe the main the reason drove the Kd to change. For example, at the time when the sea ice exchanged brine and bubbles with the atmosphere above it and the seawater below it, its highly absorbent particles in it tend to remain in their original position. Assuming that the total coefficient of absorption remained constant, air bubbles that could significantly scatter light were the dominant factor influencing changes in the values of Kd in sea ice. This conclusion is supported by the fact that the R² between the volume fraction of bubbles, and the values of Kd on the surface and sub-surface layers of sea ice were 0.52 and 0.47, respectively. If climatic changes have led to an increase in the volume of bubbles, the more bubbles will increase the scattering properties of sea ice and lead to an increase in Kd. Conversely, the reduced bubble volume would reduce the scattering properties of sea ice, which in turn would reduce Kd.

The future work is to deploy this instrument in the polar sea ice and conduct a long-term continuous observation. Using the observation data, we will develop a Kd prediction model that can accurately forecast the spatiotemporal variability of Kd in the polar sea ice. That work will contribute to the study of the energy balance and ecological environment in the polar regions.