Polar Regions | Remote Sensing Approach

By: Alvarinho J Luis
Remote sensing is a space-based satellite technique preferred for its repetitive coverage of inaccessible and rugged terrain for surface characterisation. This paper showcases climate change in the vulnerable polar realms by adapting different algorithms to the satellite technology to infer surface signatures.

Remote sensing is a tool for data acquisition through two primary techniques-active and passive. Active remote sensing is employed in polar regions during consistent cloud cover and darkness, where the sensor in orbit (eg., LIDAR and RADAR) sends its own signal and records the backscatter signal. During cloud-free conditions, however, the passive remote sensing measures the radiation reflected from the surface, eliminating the need to send its own signals. Researchers working on polar regions use this data to study temporal sea ice concentration (SIC) changes, estimate glacier surface velocity, map blue ice areas, detect crevices and surface melt on the ice sheet etc.

Sea Ice Variability in Polar Regions

Sea ice is frozen seawater that floats on the ocean surface. It forms in winter and melts in summer and has a high albedo (>90 per cent) which alters the atmospheric heat budget. In the polar winter, high convective heat loss in dark conditions freezes the freshwater leaving dense brine, which sinks down to the abyssal depth. Through thermohaline circulation, this dense brine plume eventually makes its way to the global oceans, as the Antarctic bottom water. Measurements from NASA’s Scanning Multichannel Microwave Radiometer (SMMR) during 1978, and the Special Sensor Microwave/Imager (SSM/I) sensors from 1987 onwards, provide a time series of sea ice concentration spanning 38 years, which helps climate scientists to monitor the interannual variations and trends through the use of remote sensing in polar regions.

Interannual variability of antarctic sea ice

sea ice concentration

The satellite-observed sea ice concentration data from the polar regions was used to derive sea ice extent by summing up only those pixels with sea ice content greater than 15 per cent. The sea ice covers an area of about 14-16 million sq km in late February/March (winter) in the Arctic and 17-20 sq km in the Southern Ocean in September (austral winter). On an average, seasonal decrease is much larger in the Southern Ocean, with only about 5-6 million sq km remaining at the end of summer. In terms of overall trend, the Antarctic sea ice extent exhibits a positive trend of about 2 per cent per decade during 1979-2015. In winter (summer) the trend is weaker at 1.48 (3.68) per cent per decade. However, some regions such as Bellingshausen-Amundsen Sea show a decrease of 3700 sq km per year which is due to high surface and basal melt during December to May. Likewise, Weddell Sea shows a decrease by 1500 sq km per year from June to November. Nevertheless, the Antarctic sea ice extent exhibits
Geolocation of polar recird glacier

high interannual variability which is enhanced post 2000 (Fig. 1). There is also large seasonal variability in the trend depicted in Fig. 2 for all the sectors. For example, after a record-high in September (20 million sq km) during 2012 to 2014, the Antarctic sea ice has decreased by 6.82 million sq km during September to November, 2016. This amounts to 18 per cent more loss than in any previous September-November months during the satellite era (Turner et al., 2017). Though this increase is statistically not significant, what causes it is ambiguous.

This small increasing trend in Antarctic is contrary to results from coupled climate models as well. The positive sea ice trend is attributed in part to the stratospheric ozone depletion over Antarctica that promotes a dip in the mean sea level pressure in the Amundsen Sea (Amundsen Sea Low, ASL), West Antarctica (Sigmond and Fyfe, 2010). The small overall Antarctic increase in sea ice extent appears to be the residual of a coherent pattern of a much larger regional increase and decrease that almost compensates each other (Fig.1). These large local areal changes can also be viewed as changes in the length of the ice season (Stammerjohn et al., 2012).

The seasonal variability in sea ice concentration results from a combination of winds and ocean circulation. The ASL primarily controls the atmospheric conditions between the Antarctic Peninsula and the Ross Sea and promotes northward-blowing winds over the region (Fig. 2). The interannual sea ice extent variability in the Ross Sea sector is significantly correlated with the strength of these winds and the depth of ASL. Stronger cold winds facilitate coastal polynya formation along the Ross ice shelf boundary and increase the sea ice production (Holland and Kwok, 2012). Other researchers attributed the positive trend to changes in atmospheric circulation induced by Southern Annular Mode (SAM) and El Niño-Southern Oscillation (ENSO), with more La Niña events since the late 1990 (Zhang, 2007). Our understanding of the increase in sea ice extent is fragmentary, as the climate models are unable to replicate the observed scenario.

SAM, which is the index of pressure difference between subtropics and coastal Antarctic, has switched to positive since 2000. This is responsible for a shift in the strong wind region called west


Resultant surface velocity

wind drift in the Southern Ocean towards Antarctica. The surface water cools rapidly by larger net surface heat loss and through upwelling, facilitated by northward Ekman transport. The availability of cold water during summer preconditions the surface for formation of more sea ice in winter.

Monitoring Changes in Glacier Velocity in the Antarctic

Observations of ice motion in glaciers are critical to understand mass balance and its contribution to sea level rise, apart from predicting future changes. Most of the studies use Synthetic Aperture Radar (SAR) and optical data with the methodologies for velocity study through SAR including Interferometeric SAR (InSAR), Digital Elevation SAR (DInSAR), offset tracking, and feature tracking which have reasonable results.

Feature tracking is one of the most effective ways to study glacier as it allows the estimation of displacement between two images-a reference image and a search image. Since the 1980s, image matching technique has been used by manually inspecting the images and identifying the same objects in the images from two different time periods. Scambos et al. (1992) was the first to perform image matching based on normalized cross-correlation. Image matching methods can be either area-based or feature-based. Area based methods operate directly on image quantities like brightness or phase. Feature-based methods match features that are extracted from the images in a pre-processing step. Such features can be crevasses, rocks or other differences in digital numbers. The window size to be correlated has to be large enough to ensure that texture and not noise is matched.

A study was conducted on the Polar Record Glacier, east Antarctic located on the eastern side of the Amery Ice Shelf (Fig. 3). It is the largest outlet glacier along the Ingrid Christensen Coast, bounded by Meknattane Nunataks and
Dodd Island.

The study used Landsat 8 OLI images (panchromatic band) for the estimation of velocity. The study for the estimation of glacier velocity was first conducted on a single image pair using four different tools-Image GeoRectification and Feature Tracking (ImGRAFT), Normalized Cross-Correlation (CIAS), COSI-Corr and image-to-image cross-correlation (IMCORR). The statistical evaluation COSI-Corr method yielded pixel-level velocity with both magnitude and directions. The pre-event and post-event images were selected and ortho-rectified. The images were then correlated with each other with a search window size of 256 x 256 pixels (max value) to 8 x 8 pixels (min value) with the step size of 8 pixels and mask threshold of 0.9 using the frequency correlator option. The procedure is summarized in figure 4.

The velocity of the Polar Record Glacier is observed to be 1-2 m per day. The velocity and the direction presented in figure 5 agree with previous studies (Liu et al., 2017).

Way forward

Future studies should focus on bipolar sea ice variations in polar regions using a record of satellite-based sea ice concentration over 38 years. This will also help decipher teleconnections between the poles. Sea ice volume estimates are crucial for evaluating interannual changes in sea ice and the contributing factors like freshwater released from ice-sheet basal melt etc. The changes in the ice sheet elevation due to surface and basal melt by constructing digital elevation models using Sentenal 1 & 2, ALOS –PALSAR sensors will provide a wealth of information about the health of the glaciers and ice sheet over the west Antarctic in particular, where the basal melt is accelerated. Continuous monitoring of calving events along the coast of Antarctica is required to detect breakaway of icebergs from glaciers and ice sheets to the oceans since they are the major contributor to current sea-level change. With a new satellite NISAR, by NASA and ISRO, geospatial applications to cryospheric research towards understanding the polar regions will receive a fillip.

Leave a Reply

Your email address will not be published. Required fields are marked *