Editorial for the Special Issue “Remote Sensing of the Polar Oceans”

Abstract

This Special Issue gathers papers reporting research on various aspects of the use of satellites for monitoring polar oceans. It includes contributions presenting improvements in the retrieval of sea ice concentration, extent and area, and concerning error information; the interannual and decadal variability of sea surface temperature and sea ice concentration in the Barents Sea; validation and comparison of Arctic salinity products; melt pond retrieval applying a Linear Polar algorithm to Landsat data; the characterization of surface layer freshening from sea surface salinity and coloured detrital matter in the Kara and Laptev Seas; multi-sensor estimations of chlorophyll-a concentrations in the Western Antarctic Peninsula; and enhanced techniques for detection and monitoring of glacier dynamics and iceberg paths.

1. Introduction

Polar regions have undergone significant changes over the past 40 years, impacting the global climate [1,2]. Sea ice is rapidly changing (in terms of extent, concentration, and, especially, thickness), both in the Arctic and in the Southern Ocean [3,4,5,6]. This affects climate and biogeochemical cycles, ocean circulation and stratification, as well as the ocean–atmosphere exchange of momentum, heat and gases [7,8]. Light and the nutrient environment of the upper ocean, and hence biological productivity, are also influenced since sea ice variability is a critical component of the polar ecosystem, providing habitats, refuge, and a source of food and nutrients for many dependent species [9,10].

The Arctic is warming much more rapidly than Antarctica and other parts of the world, and Northern Hemisphere summer sea ice is thinning and shrinking dramatically, so that an ice-free Arctic is a possible future scenario [11]. Beyond the dramatic impacts on polar biodiversity, in the Arctic these changes also have notable implications for global trade and economies linked to polar shipping corridors [12,13], and for coastal communities exposed to multiple climate-related hazards (e.g., permafrost thaw, local pollution, and offshore methane release) [14].

On the other hand, the recent dramatic reversal in the changes occurring in Antarctic sea ice, i.e., from relatively gradual increases to rapid regional decreases that rival those observed in the Arctic [5], emphasizes the urgency of improving our capability to monitor its formation and evolution. More icebergs are now present in this region, affecting the growth, structure, and, hence, the identification of sea ice [15,16].

In this context, obtaining detailed information on the physical and biogeochemical characteristics of polar oceans, as well as on sea ice formation/deformation, drift, distribution, melting and variability, represents a strategic challenge to improving the quantification of ocean and sea ice feedbacks, the representation of the cryosphere in climate simulations, and the understanding of its coupling with atmospheric, oceanic and ecosystem processes. The vastness of harsh and difficult-to-access marine polar areas and scarcity of in situ data make remote sensing observations the only tools that can provide this key information on a continuous basis and with better spatio-temporal coverage than traditional field measurements [17,18].

This Special Issue gathers contributions on research related to various aspects of the remote sensing of both polar oceans. The topics covered in this Special Issue include improvements in the retrieval of sea ice concentration (SIC), extent (SIE) and area at the regional and hemispheric scales; spatial correlation length scales of sea ice concentration errors for high-concentration pack ice; an analysis of the interannual and decadal variability of sea surface temperature and sea ice concentration in the Barents Sea; the use of saildrones for the validation of Arctic salinity products; the characterization of surface layer freshening from sea surface salinity and colored detrital matter in the Kara and Laptev Seas; the development of a new algorithm for the multi-sensor estimation of chlorophyll-a (Chl-a) concentration in the Western Antarctic Peninsula; the application of a Linear Polar algorithm for melt pond retrieval to very high-resolution Landsat data; the detection and long-term path visualization of the A-68 iceberg in the Ross Sea; and an enhanced methodology to extract glacier ice front positions from multi-polarization SAR imagery applied to the d’Iberville glacier. The next section reports a short summary of the varied contributions to the Special Issue.

2. Overview of Contributions

This Special Issue contains ten contributions, i.e., eight research articles and two technical notes, dealing with sea ice cover, sea surface oceanography, and glaciers and iceberg monitoring.

2.1. Sea Ice Cover

The SIC seasonal variability is a key indicator of global climate change [19]. Satellite passive microwave observations have been used to monitor SIC and SIE since the 1970s by measuring sea ice and open water brightness temperature (TB). Since then, several products have been developed to improve our understanding of the sea ice cover in both polar hemispheres [3,4], allowing us to estimate its decrease in the Arctic Ocean [20] and its enormous variability of the Southern Ocean [5,6]. However, this information is characterized by a certain level of inaccuracy due to retrieval and sensor noise, and may be less accurate during summer when sea ice and open-water TB values are similar.

Despite its importance (e.g., in forecasting sea ice conditions or assimilating sea ice values in numerical models), limited details are generally provided about this uncertainty. The EUMETSAT OSI SAF ESA-CCI Level-4 SIC data records, for example, provide gridded SIC error estimates but do not provide an error covariance matrix, so that the correlation scales of these error contributions, and the total error, are unknown. Kern [21] attempts to fill this gap by deriving SIC error correlation length scales for high-concentration pack ice over the Arctic and the Southern Ocean. His approach seems to work and provides reasonable correlation length scales, for most of the freezing season and into spring, which are smaller for the total error (i.e., less than 200 km) than the SIC error (i.e., between 200 and 700 km) in both hemispheres.

Han et al. [22], instead, developed a summer SIC retrieval model for the Pacific Arctic Ocean using Advanced Microwave Scanning Radiometer 2 (AMSR2) observations and European Reanalysis Agency-5 (ERA-5) reanalysis based on Random Forest (RF) regression. This method takes into account TB changes caused by atmospheric effects. Consequently, it shows higher performance in retrieving summer SIC values in the Pacific Arctic Ocean relative to the Bootstrap and Arctic Radiation and Turbulence Interaction STudy (ARTIST) Sea Ice (ASI) algorithm, under various atmospheric conditions, when compared with ice/water maps generated from the Korean Multi-purpose Satellite-5 synthetic aperture radar images from July to September in 2015–2017.

Furthermore, Meier et al. [23] focus on SIE and areas comparing hemispheric and regional trends from three passive microwave-based sea ice products archived at the National Snow and Ice Data Center (NSIDC): the NASA Team (NT) algorithm product, the Bootstrap (BT) algorithm product, and Version 4 of the NOAA/NSIDC Climate Data Record (CDR) product. The presented results indicate that the three products are consistent in showing strong statistically significant negative trends in the Arctic and small positive trends in the Antarctic. The patterns are also similar when looked at on a regional scale, except for a steeper trend in the Central Arctic observed in NT sea ice area, likely related to increasing surface melt. Small differences in trends (i.e., between NT trends and BT or CDR), can also be due to the inter-sensor calibration approach, which yields small discontinuities between the products. Such differences are not critical in the Arctic, but can be important in the interpretation of Antarctic data where overall trends are near-zero and borderline statistically significant.

Using an empirical orthogonal function analysis, Mohamed et al. [24] investigated the interannual and decadal variability of sea surface temperature (SST) and SIC in the Barents Sea, as well as the possible relationships among them and local and large-scale atmospheric parameters over the last 39 years (1982 to 2020). They estimated the SST warming over the study period in the ice-covered and ice-free regions of the Barents Sea and found a strong decline in the SIC and a shortening of the sea ice season. On the interannual and longer-term scales, a strong coherent spatio-temporal variability is reported in both SST and SIC. Both parameters were highly correlated with air temperature, meridional winds, and two climate indices (i.e., the Atlantic Multidecadal Oscillation and the East Atlantic Pattern) on the interannual time scale.

Another important factor that influences the Arctic climate is the formation and distribution of melt ponds. However, even though accurate information on melt ponds by remote sensing is highly desirable, available large-scale products, such as the melt pond fraction (MPF), still require additional validation and verification. In this scenario, Qin et al. [25] discuss the applicability of the Linear Polar algorithm, generally used for Sentinel-2 data, to the very high-resolution Landsat 8 (L8) optical imagery. To this aim, they determine the most suitable band combination for the Linear Polar algorithm in L8 data through sensitivity experiments, and demonstrate that this algorithm can identify 100% of dark melt ponds and relatively small melt ponds. Still, they suggest that MPF from Landsat longer time series are more efficient than those from Sentinel-2 for verifying large-scale MPF products or conducting the long-term monitoring of a fixed area.

2.2. Sea Surface Oceanography

The Arctic Ocean is one of the most challenging regions to observe because of its remoteness and extremely cold temperatures. On the other hand, it is a critical area for both climate and biodiversity [7]. Among other things, it experiences the most significant changes from climate warming [2], being strongly affected by the above-mentioned sea ice retreat and the enormous growth of freshwater intake due to glacier melt and river runoff increases. Thus, monitoring sea surface salinity (SSS) over the Arctic Ocean is fundamental for improving our efforts in climate monitoring and forecasting.

Although SSS is a key component of the water cycle and of oceanic circulation, in situ observations are still sparse due to limited accessibility and the rough environment of most oceanic regions, especially in the high-latitude oceans [26]. SSS has been monitored from space since 2010, thanks to the European Space Agency’s Soil Moisture and Ocean Salinity (SMOS) mission, the National Aeronautics and Space Administration (NASA) Aquarius mission and the NASA Soil Moisture Active Passive (SMAP) mission. However, remotely sensed SSS products showed limitations in some regions, especially at latitudes higher than 45–50° [27]. Thus, additional studies aimed at improving the satellite calibration/validation in these regions are warmly welcome.

In this context, Vazquez-Cuervo et al. [28] present an interesting experiment using a Saildrone vehicle for validating SSS in the Arctic Ocean derived from three separate products that use data from the SMAP satellite. To examine possible issues in resolving mesoscale-to-submesoscale variability, comparisons are also made with two versions of the Estimating the Circulation and Climate of the Ocean (ECCO) model. A strong freshening event associated with the Yukon River discharge was detected in all the SMAP products and the Saildrone SSS observations at approximately day 150 of 2019. Overall, both SMAP and ECCO data show statistically significant correlations and negligible biases with Saildrone observations. These results make this approach promising for future high-latitude applications, such as the monitoring of changes in fresh shelf water influence on the large-scale salinity of the Arctic Ocean regional seas.

The discharges of the Ob’, Yenisei, and Lena rivers represent other important sources of freshwater, and of dissolved organic matter and suspended matter, into the Arctic Ocean. River runoff signatures in the Kara and Laptev Seas are also evident in recent satellite SSS products. Umbert et al. [29] focus on this phenomenon and compare the new SMOS Arctic SSS absorption coefficient of colored detrital matter (CDM) in the Kara and Laptev Seas for the period 2011–2019. The SSS and CDM are found to be strongly negatively correlated in the regions of freshwater influence. Then, exploiting this linear correlation, they estimate the SSS back to 1998 and retrieve the freshened surface layer (FSL) for the Kara and Laptev Seas. The yearly mean freshwater content and extension of the FSL, computed from SMOS SSS and optical data, is found to co-vary with in situ measurements of river discharge from the Arctic Great Rivers Observatory database. Following these results, they suggest that SMOS SSS might be safely used to monitor the river discharge changes in Eurasia and to understand the Arctic freshwater system during the ice-free season.

In the austral hemisphere, Chl-a underestimation by global satellite algorithms in the Southern Ocean has long been reported, limiting the potential for evaluating phytoplankton biomass. Several regional Chl-a algorithms have been proposed to overcome this issue. Ferreira et al. [30] analyze the performance of both global and regional satellite algorithms that are currently available for the Western Antarctic Peninsula and investigate which factors are contributing to the underestimation of Chl-a. This underestimation seems to vary with the levels of Chl-a. Nevertheless, no clear link was found with the pigment packaging effect, nor with the phytoplankton community composition and sea ice contamination. Then, they propose a new and more accurate satellite Chl-a algorithm (i.e., the OC4-SO) based on multi-sensor satellite data and the most comprehensive in situ dataset ever collected from the WAP, highlighting its potential to become an important tool for an effective monitoring of ocean color and phytoplankton communities in such a climatically sensitive region.

2.3. Icebergs and Glaciers

In iceberg monitoring, there are two basic objectives: iceberg detection and iceberg drift forecasting. Weather conditions and sunlight absence impose restrictions on the monitoring of Antarctica by satellite remote sensing. During the Antarctic night, both SAR and infrared images can monitor ice coverage. However, since clouds make infrared observations impossible, the continuous monitoring of Antarctica can only be carried out by SAR imaging systems. Lopez-Lopez et al. [31] gives an example of Antarctic iceberg monitoring by analyzing some elements of the drift trajectory of the A-68A iceberg using a set of Sentinel-1 SAR data that cover a period of eighteen months. The applied image processing scheme addresses two relevant problems: how to conduct quasi-automatic analysis using a fuzzy logic approach to image contrast enhancement and the use of ferromagnetic concepts to define a stochastic segmentation. To this goal, the Ising equation and the Bayes equation are used to model the energy function of the process, and the segmentation is the result of a stochastic minimization. The achieved results track the movement of the A-68A iceberg over the study period and provide details of the small variations in its area, perimeter and major axis length up to January 2019.

Finally, Zahriban Hesari et al. [32] present an enhanced methodology to extract glacier ice front positions from multi-polarization SAR imagery. This technique is based on a global threshold constant false alarm rate approach applied on single- and dual-polarization features, namely the HH-polarized normalized radar cross section and a combination of HH- and HV-polarized scattering amplitudes. The aim is to extract the position of the glacier ice front in each SAR scene, so as to estimate its dynamics. This technique is applied to the d’Iberville glacier (Ellesmere Island, Canada), taking advantage of ten C-band SAR images collected by Radarsat-2 and Sentinel-1 missions from 2010 to 2022. The experimental results show that the HH-polarized normalized radar cross ensures a better performance with respect to the dual-polarized features, and that the d’Iberville glacier exhibited, during the study period, a significant retreating trend that resulted in a net loss of 2.2 km² of the glacier’s surface.

3. Conclusions

The contributions reported in this Special Issue highlight different and varied aspects of the research on remote sensing of polar oceans and their applications to a better monitoring of sea ice and hydrological system. Several improvements are presented and discussed here, providing new products, techniques and validations to scientists and the polar community. However, additional efforts are still required for acquiring and exploiting enhanced remotely sensed observations that could allow us to understand better the complex relationships and feedbacks between sea ice formation and evolution, sea surface features and ocean dynamics. Algorithms need sustained research, leading to improved and enhanced products and a better estimation of their accuracy, to the goal of continuity in the regular and global satellite observation of essential climate variables.