The surface albedo of sea ice in the Arctic has a large impact on the radiation and energy balance of the region. During the melting season the sea ice covered areas experience vast changes, e.g., the albedo of sea ice drops significantly. With a lower albedo, more radiation is absorbed, leading to enhanced melting. During the melt season there are also large changes in the extent of the sea ice.
The AROME-Arctic forecasting system uses an estimate based on the ice surface temperature to represent the sea ice albedo. In order to study the accuracy of this estimate and to determine the potential for assimilating satellite-based sea ice albedo products, we have made a comparison of the CM SAF (Satellite Application Facility on Climate Monitoring) CLARA-A2 SAL (CM SAF cLoud, Albedo and surface RAdiation dataset from AVHRR data – Edition 2 Surface ALbedo) data record and the surface albedo from the operational AROME-Arctic. Consecutive 6-hour forecasts from the operational AROME-Arctic starting at 00, 06, 12 and 18 UTC were used. The comparison is made for the period from April to August in 2016 with weekly averaged data for both sources.
The sea ice albedo derived from the SAL product varies significantly on a monthly scale, whereas the model albedo has a lower variance. The results show differences in sea ice albedo over the whole comparison area, especially for the summer months. The differences are the largest over the marginal ice zone.
These findings have been used for improving the albedo parametrisation in the model system. A smaller dataset of one week in April 2016 has been made to study some of the improvements, particularly an update in the way sea ice concentration is handled. A large improvement in the sea ice albedo is found on the marginal ice zone, i.e. in areas where the ice concentration is small. Elsewhere the effect is small.
This work has been conducted in the Copernicus Arctic Regional Reanalysis project.

Area D - Board 449

The dynamic changes in the cryosphere, such as the ice calving fed by seaward flowing glaciers, have a significant impact on the society and environment, by contributing to the increase of the sea-level rise. This requires a continuous monitoring of the processes occurring within the ice-sheets to predict the dynamics and stability of the polar ice-caps.

Radar sounders can acquire profiles of the subsurface (radargrams) that provide rich information for estimating the Essential Climate Variables (ECV) and better understanding the processes and structures in the ice. Subsurface targets detected from the radargrams can be used to determine: (i) the surface and the basal interface topography, ii) the ice sheet thickness and the internal layering, (iii) the characteristics of ice shelves and the position of the grounding line, (iv) the basal boundary conditions and ice flow regime, and (v) the ice mass balance.

Currently, there are a number of airborne radar sounders (e.g., POLARIS [1] and MCoRDS [2]) that have been acquiring data over the Earth’s icy areas over the last decades. Significant information on the glacier structure and processes in Greenland and Antarctica have been extracted from the analysis of these radargrams [5-7]. However, these data are acquired seasonally and over specific regions, and thus they do not provide uniform coverage in space and time nor have homogeneous data quality.

These limitations can be addressed by a satellite-mounted radar sounder, which would provide continuous, homogeneous, and uniform coverage of the cryosphere. The Earth-orbiting radar sounder should be designed to operate at a central frequency in the HF-VHF range with a bandwidth of 10 MHz, to have a penetration depth up to 5 km and a vertical resolution about 10 meters in ice. However, data acquired by orbiting radar sounders are subjected to the cosmic microwave background, path loss, relatively low transmit power, ionospheric effects, and poor resolution due to the altitude of the platform and the constraints on the bandwidth.

In this paper, we analyze the detectability of subsurface targets in icy radargrams acquired by an orbiting radar sounder taking into account these limitations. To this purpose, we defined a simulation procedure for estimating radargrams acquired by such a radar sounder. Further, we analysed the simulated radargrams with existing techniques developed for airborne acquisitions, in order to understand the detectability of the subsurface targets.
The simulation approach is based on reprocessing the available airborne radargrams to match the range of possible characteristics of the orbiting radar sounder. The simulation technique accounts for the radar and target parameters (e.g., the central frequency, bandwidth, sampling frequency, and the medium dielectric properties) that are related to the radargram characteristics (e.g., the resolution, received power, and sample spacing). Moreover, the cosmic microwave background noise is stochastically added to the simulated radargram to represent the real scenario.

Simulated radargrams of a set of important ice subsurface targets are analyzed using state-of-the-art analysis techniques developed for airborne or planetary radar sounder data. The algorithms are applied to detect the important radargram targets, such as (i) the bedrock and the surface [3] - useful for estimating the ice sheet thickness; (ii) the number and strength of the layers [4, 6] - useful for understanding the paleoclimatic records and glacier aging; (iii) the ice-shelves for better understanding processes in the grounding zone; (iv) the basal features such as subglacial lakes [8] and refreezing ice [7] - useful for understanding the basal conditions and the mass balance.

Our analysis confirms that an EO orbiting radar sounder can image important subsurface targets. From these data, we can extract useful information to better understand the cryosphere processes and structure. Moreover, the multi-temporal, homogeneous, and uniform coverage provided by the orbiting sounder will be extremely useful for the continuous monitoring of the cryosphere in the framework of climate change.

References:
[1] Dall, J., Kristensen, S.S., Krozer, V., Hernández, C.C., Vidkjær, J., Kusk, A., Balling, J., Skou, N., Sobjaerg, S.S. and Christensen, E.L., 2010. ESA's polarimetric airborne radar ice sounder (POLARIS): Design and first results. IET Radar, Sonar & Navigation, 4(3), pp.488-496.
[2] Shi, L., Allen, C.T., Ledford, J.R., Rodriguez-Morales, F., Blake, W.A., Panzer, B.G., Prokopiack, S.C., Leuschen, C.J. and Gogineni, S., 2010, July. Multichannel coherent radar depth sounder for NASA operation ice bridge. In Geoscience and Remote Sensing Symposium (IGARSS), 2010 IEEE International (pp. 1729-1732). IEEE.
[3] Ferro, A. and Bruzzone, L., 2012. Analysis of radar sounder signals for the automatic detection and characterization of subsurface features. IEEE Transactions on Geoscience and Remote Sensing, 50(11), pp.4333-4348.
[4] Carrer, L. and Bruzzone, L., 2017. Automatic enhancement and detection of layering in radar sounder data based on a local scale hidden Markov model and the Viterbi algorithm. IEEE Transactions on Geoscience and Remote Sensing, 55(2), pp.962-977.
[5] Xiong, S., Muller, J.P. and Carretero, R.C., 2017. A new method for automatically tracing englacial layers from MCoRDS Data in NW Greenland. Remote Sensing, 10(1), p.43.
[6] Ilisei, A.M., Khodadadzadeh, M., Dalsasso, E. and Bruzzone, L., 2017, October. Automatic detection of subglacial lakes in radar sounder data acquired in Antarctica. In Image and Signal Processing for Remote Sensing XXIII (Vol. 10427, p. 1042718). International Society for Optics and Photonics.
[7] Bell, R.E., Blankenship, D.D., Finn, C.A., Morse, D.L., Scambos, T.A., Brozena, J.M. and Hodge, S.M., 1998. Influence of subglacial geology on the onset of a West Antarctic ice stream from aerogeophysical observations. Nature, 394(6688), p.58.

Area D - Board 450

Greenland Ice Sheet albedo is an Essential Climate Variable that critically influences ice sheet surface mass balance. It varies substantially both over the course of a melt season (especially through the snow-slush-ice transition) but also over daily timescales, which have significant implications upon meltwater generation.

There is increasing evidence that ice algae growing on the ice sheet surface are a highly important albedo reducer. Over regional scales, observations of the inter-annual dynamics of 'dark ice' areas show that the extent, duration and intensity of darkening in these areas vary significantly between years, driven by synoptic climatic conditions controlling the size of the ice algae population and hence the darkening signal that they contribute.

Field observations show that the surface loading concentration of ice algal cells varies by several orders of magnitude over spatial scales of less than one metre. These discontinuous patches of ice algal populations thus have spatially hetereogeneous impacts upon albedo at the sub-satellite-pixel level.

Here we use 5 cm multispectral imagery collected by Unmanned Aerial Vehicle (UAV) in the south-western ablation zone of the Greenland Ice Sheet during July 2017 to characterise how local-scale surface albedo variability aggregates to the 10/20 m pixel size of the Sentinel 2 mission. Our UAV time series spans a 10-day period of rapid meteorological change, including snow and rainfall events, enabling us to examine the magnitude of daily variability in surface albedo. Two days of our UAV acquisitions are coincident with Sentinel-2 overpasses which enables us to explore the extent to which local-scale daily albedo variability is captured at 10/20 m resolution, and to examine the sub-Sentinel-pixel distribution of albedo as the surface is perturbed away from its mean bare-ice state which occurs under shortwave-radiation-dominant conditions. These findings have implications for how to more accurately estimate surface melt rates using surface energy balance approaches, and for understanding the emergent properties of bare-ice surfaces which need to be captured by regional climate models running at km resolutions.

Area D - Board 451

Recent studies of the North East Greenland Ice Stream suggest that surface melting is linked to acceleration in marine-terminating sectors (e.g. Rathmann et al., 2017). This is particularly important in the context of a warming climate since North East Greenland is likely to experience the greatest proportional increase in melt area in coming years (Igneczi et al., 2016). Despite this, a paucity of observational data from this region limits our understanding of the spatial and temporal variability of melt here. The scarcity of data has also limited the development of a rigorous method for the quantifying of melt events. Here, we use high resolution 5 x 20m SAR backscatter data from the Sentinel 1 satellite to map changes in melting over time at, and inland of, Zachariae Isstrom and 79N glacier. Initial preprocessing steps are applied (eg Winsvold et al.,2017) to establish a stable data set. We present a new method which uses advanced time series analysis (e.g. changepoints theories and edge detection) with iterative optimisation to delineate melt vs non-melt area in backscatter images. We show that our method is robust in space and in time and that it can be used to assess the spatiotemporal variability in melt area in North West Greenland between 2015 and 2018. We correlate our findings with a range of potential meteorological and glaciological drivers with a view to understanding the controls on melting in this, climatically sensitive, region. We present a technique for measuring the formation of supraglacial lakes and other defined edge surface features. These are correlated spatiotemporally with melt events. Finally, we show preliminary results from an assessment of melt amount over this period using similar method. Our intention is to make our method available to other users who are interested in similar science questions, for example studies of melting in other parts of Greenland and in Antarctica.

Area D - Board 452

A long and unbroken record of ice sheet surface elevation change (SEC) have been recognized as an essential climate variable (ECV). Such a record is provided by the ESA Greenland ice sheet CCI project, as 5-year running means, spanning back to 1992. This long record directly shows the imprint of climate change on the Greenland ice sheet. The 5-year running mean limits the imprint of weather variability and serves as a baseline for detailed process studies.

We focus on the novel results from ESA Cryosat-2 and the French-Indian SARAL/ALtiKa satellite, as this latest generation of radar altimeters provides new opportunities in monitoring Greenland ice sheet SEC. The orbit of Cryosat-2 enables higher temporal-resolution data product from ku-band radar altimetry; a 2-year running mean SEC. This 2-year product shows the inter-annual variability in weather and its imprint in surface elevation. Since SARAL/ALtiKa at ka-band radar altimetry has less surface penetration in firn covered areas, it will map changes in snow-fall in the interior region of the Greenland ice sheet. During the Cryosat-2-SARAL era, 11 maps of Greenland SEC are available from the Greenland CCI, and they all give insight into the complex nature of the Greenland ice sheet changes.

Further, Cryosat-2 alone, the combination of the 5-year and 2-year SEC products, shows the interplay between surface mass balance and internal ice sheet dynamics. Whereas the combination of elevation changes from the two instruments provide high spatial-resolution and insights into changes in firn properties, which ultimately can lead to a better understanding of how ice sheet mass balance estimate can be derived directly from satellite radar altimetry. The interpretation of dual-frequency altimetry is supported by firn modeling. Here, the model of firn driven elevation change difference between Ka- and Ku-band may provide unique insights in the possible exploration of combined Cryosat-2 and ICESat-2 data. In the future, a dual-band radar altimeter operating from space may provide ice sheet wide measurements of firn densities, and thus providing a self-sustained system for direct determination of ice mass balance from satellite altimetry.

Area D - Board 453

Quantifying the distribution and abundance of ice algae is fundamental for understanding the evolving processes of algal blooms in supraglacial environments, particularly over the Greenland ice sheet, given the role of algal impurities in modulating surface albedo and meltwater production. This is especially true along the ablation zone in southwest Greenland where ice algae play a vital role in controlling surface melting. Despite the importance of these processes, the spatial distribution and abundance of ice algae at a regional scale remain poorly known and understudied.
Here, we show results concerning the dynamic changes of dark ice and ice algae in southwest Greenland during the summers of 2016 and 2017 building on the spectral characteristics of ice algal cells and the advanced band designations of Sentinel-3 Ocean and Land Colour Instrument (OLCI). We show that the dark ice expanded rapidly over a few days in July and was most prevalent from late July to the middle of August. The OLCI spectra of the dark ice present the typical spectral signatures of chlorophyll, consistently with the field spectra collected over algae-concentrated areas. To study the spatio-temporal evolution of algae, we define an algal index based on the reflectance ratios between the 709-nm and 673-nm bands of OLCI to detect the algal hot spots. The algal abundances show altitudinal dependency on the bare ice surface, consistently with previous observations in Greenland and other glaciated regions. In addition, we found a significant exponential relationship between the darkening duration and algal index, further demonstrating the close linkage between algal growth and dark ice expansion. We laslty discuss results concerning the links between the meteorological drivers including air temperature, meltwater production, and surface runoff obtained from the regional climate model MAR and algae spatio-temporal variability, to study how environmental factors might influence the spatiotemporal changes of the algae-covered surfaces.

Area D - Board 454

Water stored at the ice-bed interface has been identified as a key component of the hydrological systems of ice caps and ice sheets in the Polar Regions. Drainage of subglacial lakes have been observed to influence ice dynamics, which increases uncertainty in projections of the contribution of ice sheets and ice caps to future sea level (Stearns et al., 2008). Subglacial lakes also have the potential to be rare and isolated ecosystems (Christner et al., 2014), which may prove useful as analogues for extreme environments elsewhere in the Solar system. While almost 400 significant bodies of stored water have been discovered to date beneath the Antarctic ice sheets (Wright & Siegert, 2012), fewer than ten lakes have so far been detected beneath the Greenland Ice Sheet, or elsewhere in the Arctic (Palmer et al., 2013, Rutishauser et al., 2018, Willis et al., 2015). The few subglacial lakes that have been detected in the Arctic are thought to be primarily fed by drainage to the bed of surface meltwater, so they are unlikely to be isolated ecosystems. However, it is likely that the cycle of filling and drainage of these lakes occurs on shorter timescales than compared with Antarctic lakes, due to the extensive and increasingly intense surface melting that occurs throughout the Arctic during the boreal summer. Because of this, they have the potential to significantly alter the flow dynamics of the overlying and downstream glaciers.

Here we combine observations from a range of remote sensing approaches, including ICEsat, ArcticDEM, TanDEM-X, Landsat-8, Sentinel-2 and airborne ice penetrating radar, to investigate several possible subglacial lakes in the Arctic, and to evaluate their influence on the flow dynamics of nearby glaciers. We derive time series of elevation data and ice flow around and downstream of candidate subglacial lakes to investigate the relationship between the timing of inferred lake drainage and variations in ice flow at a range of temporal scales. Our preliminary results indicate that the ice dynamical signature of subglacial lake drainage in the Arctic is highly site specific and is generally low magnitude in comparison with their Antarctic counterparts. We suggest this is due to the relatively low volumes of stored Arctic sub-ice water resulting from generally high horizontal gradients in ice thickness and corresponding lateral gradients in subglacial hydropotential, which limits the extent to which water can pond at the bed. Our results highlight that combining these established and emerging remote sensing approaches holds great promise for investigating a range of ice dynamical processes.

References
Christner, B. C., Priscu, J. C., Achberger, A. M., Barbante, C., Carter, S. P., Christianson, K., ... & Vick-Majors, T. J. (2014). A microbial ecosystem beneath the West Antarctic ice sheet. Nature, 512(7514), 310.
Palmer, S. J., Dowdeswell, J. A., Christoffersen, P., Young, D. A., Blankenship, D. D., Greenbaum, J. S., ... & Siegert, M. J. (2013). Greenland subglacial lakes detected by radar. Geophysical Research Letters, 40(23), 6154-6159.
Rutishauser, A., Blankenship, D. D., Sharp, M., Skidmore, M. L., Greenbaum, J. S., Grima, C., ... & Young, D. A. (2018). Discovery of a hypersaline subglacial lake complex beneath Devon Ice Cap, Canadian Arctic. Science advances, 4(4), eaar4353.
Stearns, L. A., Smith, B. E., & Hamilton, G. S. (2008). Increased flow speed on a large East Antarctic outlet glacier caused by subglacial floods. Nature Geoscience, 1(12), 827.
Willis, M. J., Herried, B. G., Bevis, M. G., & Bell, R. E. (2015). Recharge of a subglacial lake by surface meltwater in northeast Greenland. Nature, 518(7538), 223.
Wright, A., & Siegert, M. (2012). A fourth inventory of Antarctic subglacial lakes. Antarctic Science, 24(6), 659-664.

Area D - Board 455

With Sentinel-1A launched in April 2014, followed by its twin satellite Sentinel-1B in April 2016, a continuous highly resolved C-band SAR dataset is freely available for most places on Earth including the polar areas. Here we investigate in how far this dataset can be utilized for seasonal analyses over snow and ice in north-eastern Greenland with a special focus on supraglacial lakes. During the melt season, numerous supraglacial lakes form at the surface with the potential to lubricate the base of the ice sheet after drainage. We therefore attempt to monitor the lake evolution from the dense time series of Sentinel-1 acquisitions. We base our approach on thresholding methods in order to classify lake areas in the radar images. Here we focus on an automatic processing chain. In a first step Sentinel-1 radar acquisitions are downloaded, calibrated and geocoded. Based on this product optimal thresholds are found and the lake area is classified. Widespread wet surfaces in the surrounding of the lakes are the most challenging aspect in this concept. Our results are validated by optical satellite imagery from Sentinel-2 and Landsat-8 and further analyzed to detect the timing and duration of lake filling and draining. In addition to the dense time series of supraglacial lakes we present a continuous record of glacier surface velocities derived from intensity offset tracking in order to investigate the seasonal behavior of glacier flow in the study area.

Area D - Board 456

Ice dynamics over large areas can be captured thanks to remote sensing. Compared to previous decades of glaciological research, frequent observations are now available through the relatively recent satellites Landsat-8 and Sentinel-1/-2, allowing to study the short-term ice dynamic.
Here, we present our technique to reconstruct and represent the sub-seasonal ice flow variations of 3 major greenlandic glaciers from 2013 up to now. Different types of glaciers are examined in this study: a land-terminating glacier (Russell Gletscher, Southwest Greenland), an outlet glacier with an ice-shelf (Petermann Gletscher, North Greenland) and a tidewater glacier (Upernavik Isstrøm). These glaciers are all subject to seasonal fluctuations of their flow, but are affected by different external forcings. Seasonal ice speed changes of land-terminating glaciers are mainly due to change in the basal friction, while tidewater glaciers are also affected by changes in buttressing due the advance or retreat of their front.
We use imagery from a number of optical and radar satellite-based instruments, such as Landsat-8 (NASA-USGS) and Sentinel-1/-2 (ESA), to measure the surface ice velocity field at weekly time-scale and to map the ice front and grounding line positions. The obtained data are analyzed through time-space dimensions to get a better understanding of short-time dynamic of the selected glaciers. To manipulate these large datasets, we have implemented our own geographic database to easily access ice velocity products derived from different platforms. The geo-database is constructed with netCDF cubes covering the greenland ice sheet. This staging minimizes disk space encumbrance and enables a natural merging of data from different platforms on a common grid. Observations are staged as multiple layers, with their own metadata. We also created a set of Python functions to access, process, store and manage the cubes in the geo-database, visualize it as maps or graphs, design and calculate statistics. The obtained data are complemented with the most recent BedMachine ice thickness reconstruction and number of surface Digital Elevation Model from the ArcticDEM project that together provide the geometrical evolution of the glaciers. These detailed velocity/geometry time-series will be used to assess the effect of different forcings on seasonal glacier dynamic and to evaluate the impact of seasonal ice speed fluctuations on mass balance estimates. Finally we will discuss the respective role of the different forcings (such as surface melt, subglacial hydrology, calving, submarine melt) affecting the glaciers. Understanding the dynamic response of different greenlandic glaciers to seasonal forcings can help us identify more precisely the main physical processes and quantify their importance for the long-term evolution of the ice sheets.

Area D - Board 457

The geodetic mass balance is a method to describe the development of glaciers or ice sheets without the need for modelling of precipitation and melting. The measurement is based on changes in surface elevation that occur over the entire area attributed to a single glacier or sector of an ice sheet. Most recent geodetic mass balance measurements on ice sheets have been performed with laser and radar altimeters. These sensors offer the advantage of a coarse, but complete coverage over the vast regions of ice sheets with a high number of repeating acquisitions. However, spaceborne altimeters have a worse performance in areas with steeper slopes at the margins of the ice sheet than over the flat interior.

The TanDEM-X mission (TDM) provides a different possibility to obtain temporally spaced elevation measurements. With the twin satellite constellation we are able to generate DEMs for a region of interest from bistatic InSAR acquisitions. Due to the swath width of approx. 30 km there are only a limited number of scenes available for each location. Especially for the flat ice sheet interior, the acquisitions from 2010/2011 and 2011/2012 are too close in time to capture surface elevation change (SEC) within the accuracy of the instrument. However, during the effort to create a global high resolution DEM with TDM, additional coverages of the margin of the Greenland Ice Sheet have been acquired in 2013/2014.

We combine the SEC measured with the SIRAL instrument on the CryoSat-2 (CS-2) radar altimeter together with DEM differencing of InSAR DEMs from TanDEM-X in the time period of winter 2010/11 to winter 2013/2014 (Supporting figure). The sensors complement each other in the way that TDM measurements are used at regions of interest at the margin of the Northeast Greenland Ice Sheet. Reprocessed LRM mode data of CS-2 Baseline C is utilised for the remaining area of the ice sheet interior. Missing data near the ice sheet margin is filled with measurements from the CS-2 SARIn mode. This ensures high resolution maps of SEC where dynamic processes are expected on the glacier termini. At the same time, a complete coverage of entire drainage basins can be achieved. Additionally, the signal penetration at the dry snow zone of the ice sheet interior is kept minimal by using retracking procedures during the CS-2 processing.

To identify the area over which SEC is integrated to calculate volume change, a method has been developed to generate individual drainage basins for outlet glaciers. This approach employs a modified watershed algorithm, that uses the TDM global DEM as source of slope information and takes ice surface flow directions into account for areas moving faster than 20 m/a.

We calculate the volume change for several outlet glaciers in the NE Greenland region with focus on Zachariæ Isstrøm, 79North and Daugaard Jensen. Subsequently the volume change is converted to a mass change taking modelled firn densification processes from the RACMO FDM model [1] into account.

[1] Ligtenberg, Stefan RM, et al. ”Brief communication: Improved simulation of the present-day Greenland firn layer (1960-2016).” Cryosphere 12.5 (2018): 1643-1649.

Area D - Board 458

Both the Antarctic and the Greenland Ice Sheet have been defined as Essential Climate Variables (ECV) by the World Meteorological Organisation’s (WMO) Global Climate Observing System (GCOS). Among others, ice mass change is one ECV products required to characterise the Earth’s changing climate. Operational between 2002 and 2017, the Gravity Recovery and Climate Experiment (GRACE) and its successor GRACE-FO, launched in 2018, are the only space-born sensors which are directly sensitive to changes in ice mass. In 2018, processing centres started to roll out their latest GRACE monthly solutions releases. These releases comprise the entire GRACE observational period and are based on improved Level-1 data, updated background models and enhanced processing schemes.
This study evaluates the quality improvement achieved by the most recent GRACE releases, both on a global scale and for applications on the Antarctic Ice Sheet (AIS). We make use of monthly solution series provided by the official processing centres of the GRACE Science Data System (e.g. GFZ RL06, CSR RL06) as well as by independent research institutions like TU Graz (ITSG-Grace2018). Gravimetry Mass Balance (GMB) products, as provided by the Climate Change Initiative project on the AIS, were generated from the different releases. These products comprise (a) time series of monthly mass changes for the entire ice sheet and for individual drainage basins, and (b) gridded mass changes covering the entire ice sheet. The gridded product reveal spatial patterns of mass changes at a formal resolution of about 50 km.
Our assessment of the solution quality is based on the quantification of the noise level inherent to the derived GMB basin and gridded products. In addition, their signal content is compared and validated against independent observations. We make use of basin-scale mass change time series derived from surface elevation changes (SEC) observed by multiple radar altimetry missions. On seasonal time scales, our validation is complemented by surface mass changes predicted by the regional atmospheric climate model RACMO. Our analysis allows us to identify the GRACE release most suitable for the generation of GMB products. Finally, we present updated mass balance estimates for the entire AIS and its drainage basins.

Area D - Board 459

Comparisons between climate models and reanalysis show persistent biases in surface temperature over polar regions, particularly in the wintertime Arctic. Recent studies using the Community Earth System Model (CESM) suggest that incorporating a more realistic spectrally varying surface emissivity can substantially reduce discrepancies.
These studies also highlight the importance of the far-infrared region (wavelengths > 15 µm), especially in cold and dry regions, where, under clear-sky conditions, surface emission in this spectral band can reach the top of atmosphere.
In this work we investigate whether a similar response is seen in UK HadGEM-GA7.1 by replacing the current simplified assumption of surface grey body emissivity with band resolved emissivities. In this model version, the land model JULES (Joint UK Land Environment Simulator) decomposes the surface into 9 different tile types (ice, urban, 5 various plant function types, bare ground and water), providing to the atmospheric model information about the surface temperature and grey body emissivity for each tile, and the area fraction for each tile. However, the land model only considers broadband short wave and long wave fluxes while the radiative transfer model SOCRATES (Suite Of Community Radiative Transfer codes based on Edwards and Slingo) considers 9 bands (3 in the far-infrared).
We assess the impact of making these changes by analysing results from simulations with:
1 - Standard grey body model
2 - Simulated band-resolved emissivities of snow and ice from Huang et al., 2016.
3 - Far-infrared snow and ice emissivities retrieved from unique aircraft measurements over the Greenland ice-sheet (Bellisario et al., 2017).
The spectrally resolved emissivities are used in the computation of the surface energy exchanges combined with the downwelling fluxes provided by SOCRATES on the respective bands. Because these surface energy exchanges also include the thermal emission, an effective emissivity for each surface type is also computed as a function of temperature. This maintains the computational performance of the model by avoiding the calculation of the Planck function inside the land model.
The radiative impact of these changes on the upwelling flux is then estimated as well as their influence on long-term trends in surface characteristics. Top of atmosphere flux changes are investigated, especially in the far infrared so as to assess the capability of the ESA Earth Explorer 9 candidate mission FORUM (Far-infrared Outgoing Radiation Understanding and Monitoring) to observe the impacts of high latitude surface emissivity change with time. Surface temperature and sea ice extent trends are also highlighted and compared to earlier studies using CESM.

Area D - Board 460

The grounding line is the transition between the grounded ice sheet and the floating ice shelf. Its position depends on the ice thickness, the sub-glaciological topography and the ocean tide level. The delineation of an ice sheet grounding line is critical to ice sheet mass budget calculations, ice sheet dynamics modeling and the understanding of ice-ocean interactions.
The grounding line location (GLL) is difficult to identify in situ. Within ESA's Antarctic Ice Sheet (AIS) project which was carried out within the framework of phase 2 of the Climate Change Initiative (CCI) programme, we used Synthetic Aperture Radar (SAR) data from TerraSAR-X, Sentinel-1 and ERS-1/2 to map GLL for major ice streams and outlet glaciers in Antarctica. The vertical movement due to the tidal motion is separated from the horizontal displacement of the ice through the well-established double difference interferometry technique. The upper limit of flexure provided by a characteristic fringe belt is taken as a very good approximation of the grounding line and has been provided as GLL product. For the continuation of the project in the next years within CCI Extension we are planning to take advantage of the ongoing dedicated data acquisition campaign of the Sentinel-1 A/B constellation over the margins of Antarctica to derive time series of GLL throughout the year and monitor its short term migration in respect to different tidal and atmospheric conditions.
The AIS_cci GLL product (available as kml, ESRI shapefile and WKT formats) contains attributes (metadata) which includes information about model based ocean tide level and air pressure at satellite acquisition times for meaningful and interpretable comparison of GLLs. The line itself has been mapped manually for the products generated during CCI. With increasing data availability a key challenge and improvement in the future is the development and implementation of a (semi-) automatically GLL delineation procedure in the processing chain. We are presenting an approach based on the gradient of ice flow associated to the surface deformation in the radar line of sight which is applied in various situations starting with high coherent interferograms and simple shapes to more complex situations (less coherence, presence of noise, complicated geometry).

Area D - Board 462

It is well known that glaciers are retreating in West Antarctica, particularly along the rapidly-deteriorating Amundsen Sea Sector. Within this region, the importance of local-scale atmospheric, oceanic and topographic influences to this glaciological drawdown is now well resolved. However, recent research has made it clear that glaciers all over West and parts of East Antarctica are also undergoing recession, suggestive of a higher-order forcing divorced from stochastic glacier variability. Through assimilation of new and recently documented Landsat 7-, Landsat 8-, ERS-1/2-, ENVISAT- and CryoSat-2-based observations, we present a new, spatially complete synthesis of glaciological change across the entire Central Pacific-facing margin of West Antarctica for the period c.2003-2015, and compare this dataset to contemporaneous estimates of the state of the Southern Ocean and its atmosphere. We detect spatially coherent gradients in the magnitude and nature of glaciological change along this coastline through time, which are consistent with the Rossby radius scale of atmospheric perturbation across coastal West Antarctica. Signifying an intimate relationship between West Antarctic glacier change and regionally-contrasting atmosphere-ocean variability across inter-decadal timescales, our findings yield important implications for predicting the future evolution of the West Antarctic Ice Sheet.

Area D - Board 463

The collapse of four major Antarctic Peninsula ice shelves since the 1950s, including the catastrophic collapse of the Larsen B Ice Shelf in 2002, has highlighted the need for a greater understanding of ice shelf instability drivers under current climate warming trends. In particular, surface and basal melting, firn densification, water ponding, vertical hydrofracturing, horizontal fracture propagation, and ice shelf edge retreat due to increased calving are all factors that may have contributed to past ice shelf collapse events. We are using remotely sensed data from several Antarctic ice shelves to investigate the potential precursors to ice shelf instability. The data will be mostly optical imagery from Sentinel-2 and Landsat, and Synthetic Aperture Radar (SAR) imagery from Sentinel-1 and ERS 1 and 2. Preliminary investigations have been conducted on the Nivlisen Ice Shelf, which is understudied in comparison to other nearby ice shelves, including several that have shown recent signs of instability. We have developed a semi-automated method for tracking the formation, growth, shrinkage, and demise of surface meltwater features across the ice shelf, building on an existing algorithm that identifies lake drainage events on the Greenland Ice Sheet. Preliminary results have identified marked inter- and intra-annual variations in the number, area, depth, volume, and pattern of meltwater features between 2013 and 2017. In February 2017, the extent of surface melt peaked at 541 km², which was 46% greater than the next greatest surface melt extent identified throughout the study period. During periods of intense melt production, meltwater often propagates up to ~ 24 km towards the ice shelf edge through a series of interconnected melt ponds and defined meltwater channels. The results can be used to validate ice shelf surface melt/routing models, and for calculating patterns of mass loading and unloading which control ice shelf flexure and potential fracture.

Area D - Board 464

The position of the hinge-line is an important parameter in the study of the glaciers. Due to its fine sensitivity to vertical motion Synthetic Aperture Radar interferometry has been used to detect the tide induced uplift of the ice and consequently to locate the transition zone indicated by the hinge-line. This operation is typically performed manually using the differential SAR interferograms as reference. The possibility of exploiting physical models to identify such parameter has been investigated in the past [1]. However, due to the strong deformation gradients, the solution of phase ambiguities can be particularly prone to unwrapping errors, limiting the possibility of exploiting the technique in a more automatic fashion.
The scope of this work is to investigate the possibility of an unwrapping-free inversion of the hinge-line position using as baseline the model in [1], in order to move towards a more automatized detection algorithm. The proposed strategy is to estimate the deformation gradients directly from complex interferograms and fit them with the derivative of the model. The gradient of the 2D deformation phase ∇ ϕ permits to identify first the direction of the maximum variation, given by ∠ ∇ ϕ. The gradient amplitudes |∇ ϕ| can then be interpolated along the identified direction, and fitted with the derivative of the displacement model. The idea is to go towards a more automatic way to estimate the position of the hinge-line. The method has been tested using an ERS Tandem Pair acquired over the Schirmacher area in Antarctica. The input is a coarse polygonal chain plotted as reference on the interferogram as shown in Figure 1. Starting from that the algorithm check for the direction of maximum variation, interpolate the gradients that along that and fit the model in order to derive position of the transition zone. Figure 3 shows an example of the direction identification and the model fitting. The retrieved positions are finally qualitatively compared with available data [2][3][4], as shown in Figure 2.

[1] Rignot, E. (1998). Hinge-line migration of Petermann Gletscher, north Greenland, detected using satellite-radar interferometry. Journal of Glaciology, 44(148), 469-476.
[2]Rignot, E., J. Mouginot, and B. Scheuchl. 2011. Antarctic Grounding Line Mapping from Differential Satellite Radar Interferometry, Geophyical Research Letters. 38. L10504.
[3]Rignot, E., J. Mouginot, M. Morlighem, H. Seroussi, and B. Scheuchl. 2014. Widespread, rapid grounding line retreat of Pine Island, Thwaites, Smith, and Kohler glaciers, West Antarctica, from 1992 to 2011, Geophysical Research Letters. 41. 3502-3509.
[4]R. Bindschadler, H. Choi, A. Wichlacz, R. Bingham, J. Bohlander, K. Brunt, H. Corr, R. Drews, H. Fricker, M. Hall, R. Hindmarsh, J. Kohler, L. Padman, W. Rack, G. Rotschky, S. Urbini, P. Vornberger, N. Young Getting around Antarctica: new high-resolution mappings of the grounded and freely-floating boundaries of the Antarctic ice sheet created for the International Polar Year The Cryosphere, 5 (2011), pp. 569-588

Area D - Board 465

The grounding line (GL) of a glacier is a fundamental variable for outlet glaciers with ice shelves. Accurate knowledge of the GL position allows for precise measurements of ice flux into the ocean, improved ice shelf melt rate estimates, and the detection of glacier retreat or advance. The most accurate way to determine the GL location to date is synthetic aperture radar (SAR) double difference interferometry (DInSAR), which exploits tidal changes between acquisitions. The Copernicus Sentinel-1 mission has fundamentally changed the way we can monitor ice sheets today. Year round, short repeat observations of coastal Antarctica provides many more opportunities to measure ice flow as well as to generate double difference interferograms, thus increasing measurement points in time. This increase in the number of measurements allows us to measure the grounding zone, i.e. the area where tidal changes cause the grounding line to move particularly when a flat bed is present. The repeat orbit of Sentinel-1 (6 days if both satellites are used) is still not ideal to capture the grounding line of some fast moving glaciers due to decorrelation, here, dedicated acquisition campaigns of SARs with shorter revisit time are essential. In areas south of 78 degrees South, where left looking capability is needed to acquire SAR images, longer running left looking campaigns are an asset. We use Sentinel-1 and dedicated campaigns from other SAR missions like Cosmo SkyMed, RADARSAT-2 and TerraSAR-X to map the grounding zone of Antarctica. This zone, together with improved knowledge of the bed will provide a better tool for the interpretation of long term grounding line migration. Looking at the much more reliable data collection now available in coastal zones compared to only a few years ago, we also identify areas, which still require additional contributions from other missions for a more complete monitoring of the Grounding zone in Antarctica.

Area D - Board 466

Ice velocity is a key parameter for ice sheet models and a vital indicator of dynamic processes driving cryospheric change. Synthetic Aperture Radar (SAR) imagery, obtained independent of weather and light conditions, is increasingly utilised to track ice velocity. Sentinel-1a/1b have provided a wealth of high temporal and spatial resolution data allowing near real time processing. The resolution of Sentinel-1 requires understanding of how data processing impacts the output datasets. Therefore, we present a case study of sensitivity analysis on the SAR intensity tracking technique to estimate surface velocity of Pine Island Glacier (PIG). PIG was selected because within a single scene it illustrates some of the highest velocities in Antarctica as well as the relatively low velocities of surrounding regions, thus allowing differing velocity rates to be evaluated. We assessed the key input parameters of temporal baseline, window and step size, tracking offset threshold, as well as post processing filtering and dusting.

Results illustrate altering the temporal baseline can strongly impact velocity outputs due to a loss of coherence in areas of high flow (>3000 m yr-¹) over 18 days. Low velocity regions (<1000 m yr-¹) demonstrated increasing and raised velocities, resulting from more slow moving features being distinguished but also an apparent increase in noise. When selecting window and step sizes there is a trade-off between data coverage and resolution. Importantly, the resolution directly relates to the size of feature that can be tracked. Small step sizes produce anomalously different values to larger step sizes with the same window size. At higher velocities the offset threshold for the tracking algorithm only has minimal impact on mean values. However, at lower velocities there is greater variability particularly with smaller step and window sizes. For example, at <100 m yr-¹ threshold means have a maximum range of 446 m yr-¹. Whereas in regions of >3000 m yr-¹, the range is only 71 m yr-¹. A high level of filtering can strongly impact velocity outputs. Regions of 200 – 499 m yr-¹ experience the largest reduction in mean velocity with 458 m yr-¹ from the original image associated with a loss of 48% data coverage. Dusting can then further lower the mean by up to 203 m yr-¹. Low velocity regions also experience the greatest amount of data loss through dusting with <100 m yr-¹ losing 71% and a reduction in mean values by up to 185 m yr-¹.

Sentinel-1 SAR intensity tracking produces the most robust velocity results in regions of high velocity. However, the input parameters can cause variability in the realms of hundreds of metres for velocity outputs. There is a trade-off between coverage, quality and the resolution of the data impacting the output velocity values, particularly in slow flowing regions. The highest resolution data which has high data coverage should be utilised and conservative filtering applied to produce reliable velocity values. Consideration of the purpose of the data is needed to select the optimal parameters, which will vary for each new study area.

Area D - Board 467

Mapping the velocity fields of the continental ice sheets and their outlet glaciers is important in order to monitor and model the response of the cryosphere to global climate change. Since the mid 1990s, space-based SAR data have enabled measurement of ice velocities on a continental scale. Compared to interferometry, Offset Tracking techniques excel in terms of robustness and ease of automation. With the launch of Sentinel-1A in 2014 and Sentinel-1B in 2016, the potential coverage and revisit frequency have greatly improved, allowing continuous monitoring of temporal changes in the ices sheet velocity fields [1].
DTU Space has implemented an operational ice velocity processor, IPP, with a highly automated offset tracking capability, developed in the framework of ESA’s Climate Change Initiative [3]. The IPP processor ingests Sentinel-1 Interferometric Wideswath (IW) bursted SLC products, as well as SLC products from several other past and current sensors.
Several error sources affect SAR ice velocity measurements, among them orbital errors, timing errors, processing artifacts and ionospheric artifacts. The first three are traditionally handled by calibration with ground control points (GCPs), but for ice applications, this approach is challenging, for several reasons:
1. Finding ice-free stable areas with no motion can be difficult in the polar regions, especially in Antarctica.
2. Even if stable areas can be identified and used in calibration, these are often not distributed evenly throughout the scene. The presence of ionospheric artifacts can affect the measured displacements in the stable region differently from the rest of the scene, resulting in a bias on the calibrated velocities.
Orbital data are generally of high quality for Sentinel-1, but systematic differences in the geolocation of Sentinel-1A and Sentinel-1B IW SLC products have been observed, leading to a bias on the velocity measurements when combining Sentinel-1A and Sentinel-1B products. We estimate this bias using timeseries of stable areas and compare to the geolocation accuracies reported in literature [2]. Less systematic scene-to-scene variations will also be examined, especially ionospheric artifacts which mainly affect the measured azimuth displacements, and are prevalent in high latitude regions. Finally, the accuracy of velocity measurement in moving areas will be compared with newly available GPS data from Greenland coinciding with several years of Sentinel-1 operation.
[1] T.Nagler et. al, ”The Sentinel-1 Mission: New Opportunities for IceSheet Observations”, Journal of Remote Sensing, 2015, 7, 9371-9389
[2] A. Schubert et. al, “Sentinel-1A/B Combined Product Geolocation Accuracy”, Remote Sens. 2017, 9, 607, doi:10.3390/rs9060607
[3] A.Kusk, et. al., “An automated System for Ice Velocity Measurement from SAR”, Proceedings of EUSAR 2018; 12th European Conference on Synthetic Aperture Radar, pp. 929-932

Area D - Board 468

The consistent and global acquisition strategy of the Sentinel-1 radar mission has opened new opportunities to monitor the glacier flow with very high spatiotemporal details. In this study, we present seasonal ice velocity changes of 256 Greenlandic glaciers using Sentinel-1 C-band synthetic aperture radar data of 2017. We also combine Sentinel-1 radar backscatter data in our analysis to pinpoint the onset of surface melt and the extent of melt period on individual glaciers. Our results suggest that the seasonal velocity fluctuations of half of the glaciers (128) are primarily controlled by surface melt induced changes in the subglacial hydrology. However, the glaciers respond differently to the surface melting conditions and therefore, are classified in two categories. Roughly half of these glaciers (67), classified as type-A, speedup with the onset of surface melt and slowdown as the melt season tapers off. The type-A glaciers generally attain minimum velocity at the end of melt season with consistent or slight speedup during winters again. For instance, Zachariae Isstrøm in the northeast Greenland follows the type-A pattern and shows ~ 19 % velocity increase in the early melt season compared to late winter velocity. Other half of these glaciers, (61), classified as type-B, show comparable late winter and early melt season velocities showing significant slowdown during most of the melt season and speedup during winter. For example, Upernavik Isstrøm-IV in the west Greenland follows type-B pattern and shows only a slight velocity increase (~ 4%) with the melt onset as it bounces back during last winters compared to previous late-summer velocity minimum. We expect that type-B glaciers receive consistent supply of basal meltwater helping them to develop channelized networks, which further grow via surface melting. On the other hand, type-A are more likely to have underdeveloped and inefficient hydrological systems and therefore, they are more sensitive to surface meltwater. This study reports new observations of Greenlandic glaciers, which are further useful to model the sensitivity of glaciers to hydrological switching.

Area D - Board 469

The Sentinel-1 satellites provide the main data source for polar ice-sheet velocity measurements, and these are primarily carried out by applying intensity offset tracking. One of the main error sources affecting C-band SAR offset tracking, as well as Multi Aperture InSAR, measurements in polar regions are the so called azimuth streaks. These are azimuth registration errors caused by spatial variations of ionospheric Total Electron Content on scales smaller than the synthetic aperture length. Such effects become particularly relevant in the high accumulation areas of the ice-sheets, where velocities are lowest and thus comparable in magnitude to the artefacts.

Currently no general-purpose methodology has been demonstrated for the correction of azimuth streaks affecting ice-sheet velocity measurements. However, at mid-latitudes the split-spectrum technique, which exploits the range-frequency dependence of ionospheric phase delay, has been successfully used to correct line-of-sight interferograms. It is also well established that the azimuth spatial derivative of ionospheric phase delay is proportional to the azimuth registration errors, i.e. the streaks. Thus computing the azimuth derivative of the ionospheric contribution to the line-of-sight interferometric phase, estimated with the split-spectrum method, would seem to be a viable correction approach for the azimuth streaks. Indeed such a strategy was successfully applied to Multi Aperture InSAR co-seismic deformation measurements carried out with the L-band ALOS-2 PALSAR-2 sensor, although it was to our best knowledge never previously applied to ice-sheets and to Sentinel-1 data

In this work we describe the application of azimuth ice velocity correction approaches based on the split-spectrum principle to Sentinel-1 Interferometric Wide-swath (IW) data acquired over the Greenland ice-sheet. Application to this kind of data, acquired in Terrain Observation with Progressive Scans (TOPS) mode, is complicated by the strong coupling, which exists between azimuth registration and line-of-sight interferometric phase for non-stationary surfaces. We consider regions characterized by different ice-velocity and interferometric correlation properties and assess feasibility, requirements and performance of the correction approaches.

Area D - Board 470