Cryosphere

In October 2014, the Norwegian Ice Service began the beta-testing of Sentinel-1 Synthetic Aperture Radar (SAR) data covering the European Arctic and along the eastern coast of Greenland. The dual polarisation (HH+HV) Extended Wide Swath (EW) mode capability of the Sentinel-1 SAR data has shown qualitative improvements for operational ice charting in the detection of sea ice ridging and deformation features, and new ice in leads. The potential of automatic SAR detection of these features would provide a valuable resource for better routine classification of sea ice types, thus providing a proxy for sea ice thickness, and statistics on pressure ridge distributions for data assimilation with the operational and modeling communities.

Data from the Ku-band (13.575 GHz) SAR Interferometric Radar Altimeter (SIRAL) instrument from Cryosat-2, is used to quantitatively assess the capability of Sentinel-1 to identify these features based on sea ice thickness measurements based on the freeboard retrieval. The methods used to derive sea ice types based on freeboard retrieval with the Cryosat-2 radar altimetry have shown to have corresponding uncertainty estimates with first-year ice types during the spring and autumn seasons. However, different threshold trackers will be further investigated for snow cover on multi-year ice types, particularly during spring. Despite the limitations with sea ice type detection for multi-year ice, we can use the Cryosat-2 sea ice tracking algorithms as a ground-truth for small-scale features detected in the SAR images.

In this study, we concentrate on fast ice located on the north-eastern coast of Greenland. Although this is anchored by grounded ridges (stamukhas) and icebergs in the shallow water of the offshore Belgica Bank, and therefore in some places not in hydrostatic equilibrium, it is a large area of varying sea ice types that is stationary and thus negates the issue of trying to co-locate different satellite data sources that occurs when working with drifting sea ice. Because the ice is not moving, it also allows investigation of whether sea ice parameter retrieval techniques using microwave sensors can be improved during the melt season, when data from such sensors is normally unavailable.

A positive outcome will allow us to design a methodology that will optimize the combined use of Sentinel-1 SAR, the Ocean and Land Colour Instrument (OLCI) during cloud-free conditions and the advanced SAR Altimeter (SRAL), both on Sentinel-3, to improve the automatic classification of sea ice types for near-real time operational use and data assimilation.

Delay Doppler Altimetry (DDA), proposed by R.K. Raney (1998), offers improved altimetric precision and better along-track resolution than conventional pulse limited altimeters. Delay-Doppler altimeters have a high pulse repetition frequency to ensure pulse-to-pulse coherence, leading to an along-track resolution about 300 meters, improved signal-to-noise ratio and enhanced altimeter ranging performance. The delay doppler maps, the beam stacking capabilities and the improved along-track resolution  offers new possibilites for the detection and the determination of the characteristics of both icebergs and ships. Using Cryosat SAR L1b and L2 date, a method of detection of target emeging the sea has bee developed.  This method is based firstly on the analysis of the pseudo_lrm waveforms obtaining without stacking. Within these waveforms, the signature of a target emerging from the sea is  a parabola that can be easily dtected   using the metehod developed by Tournadre et al 2008 for the classical pulse limited altimeters. Using the estimate of the iceberg's position  the Delay Doppler Maps at 85 Hz are then used to estimate the area of the icebergs by averaging the DDM taking into account the iceberg displacement in both range and doppler.  This allows to compute an image of the iceberg backscatter at high resolution. The results are then  compared to the SAR (refocused) waevforms to estimate the precision of the area estimte using the waveforms alone.   

The new ESA SARVATORE project nows allows to obtain full resolution stack data. These data include more than 300 range bins before the mean sea level that can be exploited to improve the iceberg and ship detection by increase more than twofold the detection swath of the altimeter. Several examples of such detection of icebergs and ships have been analyzed and will be presented.

DDA altimetry offer improved capabilities of both iceberg and ship compared to classical altimetry as it is demonstrated using the limited Cryosat SAR dataset.  In the perspective of Sentinel 3 it can be of importance for the scientific community whose interest for icebergs and its impact on both the southern ocean circulation and ecosystems has increased during the recent year to develop an operational processing chain design to detect icebergs.

The ESA Sentinel-3 Ocean Land Color Instrument (OLCI) data will enable production of 1) snow/ice sheet albedo and 2) snow cover extent products. The commitment that ESA makes through its Copernicus program for long-term operational and environmental monitoring adds great value to earlier and ongoing data gathering after the ESA Envisat MERIS and AATSR era (2002-2012), the NASA EOS MODIS era (beginning in 2000) and the NOAA AVHRR APP-x era (1982-2011). Especially since MODIS is way beyond its design life, and the carry-on mission Suomi NPP VIIRS has not proven to extend the MODIS capability to produce an albedo climate data record, Sentinel 3 will be crucial in providing continuity of science products. Ongoing since 2007 in Greenland, the PROMICE.org automatic weather station data and year 2015 planned aerial spectral imagery support OLCI calibration, and development of bare ice and snow albedo and snow cover extent products. The ultimate goal is to ensure continuity and consistency of the climate data records currently relying on MODIS, in addition to complementing the ongoing ESA Sentinel-3 calibration and validation efforts. Our effort is further in cooperation with the International Snow Working Group Remote Sensing (iSWGR) and the World Meteorological Organization’s Global Cryosphere Watch (WMO-GCW).

In the context of quantifying Arctic sea ice volume at a global scale, altimetry provides a unique tool to estimate sea ice thickness through the freeboard method. This method mainly consists in evaluating the height of the sea ice emerging above water.

Sea ice has been continuously over flown since 1991 by altimeters onboard ERS-1, ERS-2, ENVISAT, IceSat, Cryosat-2, SARAL and soon, Sentinel-3. However, during these nearly 25 years of observation the technologies have evolved  resulting in various instruments (radar, laser), frequencies (radar in Ku and Ka bands) and measurement principles (LRM, SAR, SARin).

In order to be able to exploit this long time series without discontinuities, the impacts of the technological changes from one mission to another must be assessed.

The presence of snow on sea ice may have particularly important consequences as the radar pulse may penetrate more or less inside the snow cover according to the wavelength and the characteristics of the snow itself (depth, density, water content).

In this study we compare Ka vs Ku  and LRM vs SAR measurements over surfaces with various snow extent or properties, to better understand the behavior of each technology and estimate their biases in preparation for the upcoming Sentinel-3mission.

This research has been done in the framework of CNES TOSCA SICKAYS and IDEX Transversality InHERA projects.

Mass balance is a key variable to describe the state of health of glaciers, their contribution to sea level rise and, in a few dry regions, their role in water resource. We explore here a new method to retrieve seasonal glacier mass balances from low resolution optical remote sensing.

We derive winter and summer snow maps for each year during 1998-2014, using the Normalized Difference Snow Index (NDSI) computed from visible and SWIR channels available with SPOT/VEGETATION. The NDSI dynamic is directly linked to the area percentage of snow in the VGT kilometric pixel. The combination of 15 years of 10-daily NDSI maps with the SRTM DEM allows us to calculate the altitude of the transition between bare soil and snow. Then, we compare the interannual dynamic of this altitude with in situ measurements of mass balance available for 60 alpine glaciers (Huss et al., 2012; Zemp et al., 2009, 2013) and find promising relationships for winter mass balance. We also explore the possibility of a real-time monitoring of winter mass balance for a selection of alpine glaciers. Then, we discuss the robustness and genericity of these relationships for their future application in regions where in situ glaciers mass balances are scarce or not available.  

Finally we compare the Alps snow cover and the glacier mass balance estimations obtained with SPOT/VGT and PROBA-V (1km, 300m and 100m). 

Estimating mass balance trends from measurements of ice sheet elevation change from satellite altimetry requires knowledge of the density at which volume change occurs.  Typically, density models derived empirically from velocity estimates have been used to partition regions of dynamic and meteoric change, because a suitable firn compaction/column density model has not been available.  The limitation of this approach is that density models do not account for SMB driven elevation changes and they are difficult to employ in regions where dynamic and meteoric signals overlap.

In recent years, firn modelling has become more sophisticated and is now potentially a viable option for interpreting altimetry data.  Here, we investigate the range of approaches to firn modelling that are currently available, and assess their suitability for interpreting CryoSat-2 measurements of elevation change on the Greenland and Antarctic ice sheets.  In particular we consider the suitability of the RACMO/Ant regional climate model, which has recently been developed to include a firn densification model, for correcting radar altimeter elevation changes over Antarctica.

The results of our investigation are used to re-analyse historical mission data (ERS1, ERS2, EnviSat, IceSat, and CryoSat-2) and assess the suitability of firn modelling as a correction for the SAR altimeter measurements that will be acquired by Sentinel-3.

Coincident satellite radar and laser altimetry data have been desired by the cryospheric research community for years. Differences in elevation from these two sensor types may yield estimates of the snow depth on sea ice, which in itself is a major geophysical unknown and is also a major error term for altimetry sea ice thickness retrievals. The combination of these data should also enable the reduction of ice and snow surface penetration errors on ice sheets. 

 

With Sentinel-3 scheduled for launch in late 2015 and ICESat-2 in 2017 there is a high likelihood of coincident Sentinel-3 SRAL and ICESat-2 GLAS data collection. Given the health of Cryosat-2 it is likely that SIRAL data may be available. Although Cryosat-2, ICESat-2, and Sentinel-3 all have different orbit parameters there will be many orbit cross-overs, especially at high latitudes, that are essential for collaborative calibration and validation efforts. Coordinated efforts will enable improved ice sheet and sea ice retrievals and also will aid the development of a combined radar-laser altimetry time series through cross-calibration of data and derived geophysical products. Furthermore, Operation IceBridge is planned to operate through 2019 so that there is a potential for a Cryosat-2/IceBridge/Sentinel-3/ICESat-2 data set. In collaboration and coordination with ESA, Operation IceBridge is already conducting dedicated flights that are aligned with Cryosat-2 ground tracks.  Comparisons of Cryosat-2 data with IceBridge data have shown good agreement for the retrieval of sea ice thickness and demonstrated the utility and benefits of coordinated efforts.

This presentation will give an overview of ICESat-2 and Operation IceBridge, and lay out possible joint and coordinated data analysis and cal/val scenarios for further discussion. The timing for these discussions is ideal because specific cal/val plans for ICESat-2 are currently in development.

The CryoSat-2 satellite is now in the 5th year of data acquisition. With its synthetic aperture radar altimeter, CryoSat-2 achieves great improvements in the along track resolution compared to previous radar altimeter missions like ERS or Envisat. The latitudinal coverage contains major parts of the Arctic marine ice fields where previous missions left a big data gap around the North Pole and especially over the multiyear ice zone north of Greenland. With this unique data set, changes in sea-ice thickness can be investigated in the context of the rapid reduction of the Arctic sea-ice cover which has been observed during the last decades. Nevertheless, long time series of data retrievals are essential to estimate trends in Arctic sea-ice thickness and volume. The scheduled Sentinel-3 mission is dedicated to continue radar altimetric measurements over sea ice and therefore monitoring of Arctic sea-ice thickness. Experiences in interpreting CryoSat-2 waveforms over sea ice will be beneficial for this upcoming radar altimetry mission so that Sentinel-3 retrievals can be incorporated into a merged data set.       

The examination of retrievable information from the SAR waveforms is a field of ongoing research. Different methods can be applied to CryoSat-2 data: threshold retrackers that use the leading edge, or fitted forward models, which are applied to full waveforms. The uncertainty of these methods propagates into the uncertainty of the final sea-ice thickness estimate via the freeboard to thickness conversion. Theoretical considerations show that the magnitude of uncertainties in the radar retracking may be a major if not dominating contribution to the uncertainty budget of sea-ice thickness retrieval from CryoSat-2.

We present the current state of the CryoSat-2 sea-ice thickness retrieval that is processed at the Alfred Wegener Institute and investigate the ability to retrieve additional physical properties of the sea-ice surface. We discuss the uncertainty budget in the context of threshold retrackers which are a fast and robust method to determine the ice surface elevations. Though biases in sea-ice thickness may occur due to the interpretation of waveforms, airborne and ground-based validation measurements give confidence that the retracker threshold algorithm enables us to capture the actual distributions of sea-ice regimes.

Seasonal snow is a key element of the water cycle in high and mid latitudes, characterized by high spatial and temporal variability. Melt water is an important water resource in cold regions, in many mountain areas and also in lowlands downstream.. Accurate observations of snow cover extent and physical properties are not only of interest for climate change research, but are of great socio-economic importance. Medium resolution optical sensors enable to monitor the snow extent from regional to global scale with high temporal sampling. Automatic processing lines including rectification, calibration, cloud masking and snow detection have been implemented for generation of snow information and tested with various multispectral satellite sensors. Ongoing work is related with adapting and optimizing the snow retrieval algorithm for Sentinel 3 SLSTR and OCLI, making use of the full spectral capabilities of these sensors for generating high quality snow maps. The algorithm for mapping snow makes use of the typical spectral signature of snow in the visible (VIS) and short wave infrared (SWIR) region of the spectrum, which enables a clear discrimination against other surfaces. The baseline products include binary snow extent maps derived from combinations of VIS and SWIR bands and maps of fractional snow extent. This algorithm applies a multispectral linear un-mixing technique, which requires the selection of local end-members. In open forest areas correction for transmissivity of the canopy layer is applied for retrieval of the fractional snow extent. In spite of significant developments of optical snow retrieval methods over the last several years, there is still need for improvement, particular for forested regions and for steep mountains requiring the compensation of illumination effects. Another critical issue for the quality of the snow products is the performance of cloud screening including snow/cloud discrimination. We apply an algorithm based on multi-spectral data in the visible, near infrared, short wave and thermal infrared. The preliminary version of a retrieval algorithm for future use of Sentinel 3 SLSTR and OCLI data synergistically for mapping the snow extent has been developed, applying the multi-spectral un-mixing method and cloud screening and making use of the various spectral channels. The dual-sensor concept is tested using ENVISAT MERIS and AATSR data acquired over Alpine and lowland regions. The performance of these snow products is tested by comparison with snow maps from high resolution sensors.

The delay/doppler mode altimeter carried on-board the ESA Cryosat-2 satellite (and shortly on the Sentinel-3 and Jason-CS missions) has already demonstrated its ability to provide enhanced performances compared to the conventional mode over open ocean, with improved range precision and reduced along track spatial resolution (from ~10 km to 300 m). Its performances over ice regions (sea ice and ice sheet) have now to be demonstrated. This is the aim of this presentation that will give an overview of improvement potentialities and specific interests of this mode but that will also present some limitations and difficulties that will be encountered when processing the Sentinel-3 measurements over these regions.

In the recent years, CLS has conducted several studies on sea ice and ice sheet measurements mainly with conventional altimetry but also with the new delay doppler mode. For the first time, comparisons between Ku and Ka altimeter measurements have been possible thanks to the new AltiKa instrument embarked onboard the Saral mission launched on February 25, 2013. This comparison is of particular interest when dealing with ice observations because both frequencies have different penetration characteristics clearly impacting the echo shapes and thus the retracking algorithms. We will present illustrations of the differences observed in Ku and Ka bands using AltiKa, Envisat/RA-2 but also Cryosat-2 measurements. Different studies have also been conducted to improve the processing of altimeter data over sea ice conditions with two main objectives. The first one consists in improving the sea level estimations in regions mostly covered by the ice. The second one aims at determining the freeboard thickness for environmental and climatological studies. To fulfill these two objectives, the use of altimeter waveform classification via neural network has provided very promising results. Concerning SAR data, the analysis of the distribution of power across the looks in the stack allows to clearly correlate it with geophysical signals in particular over sea ice regions but also over ice sheets.

We propose to synthesize these different results in this talk

Reference and repeat-observations of Glacier and Ice Sheet Margin (GISM) topography are critical to identify changes in ice thickness, provide estimates of mass gain or loss and thus quantify the contribution of the cryosphere to sea level change. The lack of such sustained observations was identified in the Integrated Global Observing Strategy (IGOS) Cryosphere Theme Report as a major shortcoming. Conventional altimetry measurements over GISMs exist, but coverage has been sparse and characterized by coarse ground resolution. Additionally, and more importantly, they proved ineffective in the presence of steep slopes, a typical feature of GISM areas. Since the majority of Antarctic and Greenland ice sheet mass loss is estimated to lie within 100 km from the coast, but only about 10% is surveyed, there is the need for more robust and dense observations of GISMs, in both time and space.

The ESA Altimetry mission CryoSat aims at gaining better insight into the evolution of the Cryosphere. CryoSat’s revolutionary design features a Synthetic Interferometric Radar Altimeter (SIRAL), with two antennas for interferometry. The corresponding SAR Interferometer (SARIn) mode of operation increases spatial resolution while resolving the angular origin of off-nadir echoes occurring over sloping terrain. The SARIn mode is activated over GISMs and the elevation for the Point Of Closest Approach (POCA) is a standard product of the CryoSat mission.

Here we present an approach for more comprehensively exploiting the SARIn mode of CryoSat and produce an ice elevation product with enhanced spatial resolution compared to standard CryoSat-2 height products. In this so called L2-swath processing approach, the full CryoSat waveform is exploited under specific conditions of signal and surface characteristics. We will present the rationale, validation exercises and results from the Eurpean Space Agency's STSE CryoTop study over selected test regions of the margins of the Greenland and Antarctic Ice Sheets.

Results from swath processing of CryoSat's SARIn mode data over test regions in Greenland and Antarctica indicate a factor of 10 to 100 increase in elevation measurements in the ice sheet margins. Swath processing capability also enables SIRAL to overcome the sampling limitations of by more traditional pulse-width limited altimeters or single-beam lidars. The increased data density offers more comprehensive mapping capability in the most dynamic ice margins, and reinforces the advantages of CryoSat's twin-antenna capability for ice sheet monitoring.

Unfortunately, though Sentinel-3 carries the SRAL instrument, its single antenna limits the capability to localise the echo, and to resolve slope-dependent bias in measured elevation. The recommendation is to exploit the limited overlap between the Sentinel-3A and CryoSat missions to cross-calibrate the two instruments, and to understand the elevation biases resulting from S-3 SAR only operation. It is advised to characterise the limitations resulting from excluding SAR inteterferometry, in advance of the design of the Next-Generation Copernicus system.