Sea-ice thickness on a global scale is derived from different satellite sensors using independent retrieval methods. Due to the sensor and orbit characteristics, such satellite retrievals differ in spatial and temporal resolution as well as in the sensitivity to certain sea-ice types and thickness ranges. Satellite altimeters, such as CryoSat-2, sense the height of the ice surface above the sea level, which can be converted into sea-ice thickness. Relative uncertainties associated with this method are large over thin ice regimes. Another retrieval method is based on the evaluation of surface brightness temperature in L-band microwave frequencies with a thickness-dependent emission model, as measured by the Soil Moisture and Ocean Salinity (SMOS) satellite. While the radiometer-based method loses sensitivity for thick sea ice (> 1m), relative uncertainties over thin ice are significantly smaller than for the altimetry-based retrievals. In addition, the SMOS product provides global sea-ice coverage on a daily basis unlike the altimeter data.
In the framework of the ESA project “SMOS & CryoSat-2 Sea Ice Data Product Processing and Dissemination Service”, we present the first operational merged product of complementary weekly Arctic sea-ice thickness data records from the CS2 altimeter and SMOS radiometer. An optimal interpolation scheme is used to produce weekly Arctic-wide sea-ice thickness fields. This data product is publicly available and has been already used in several scientific studies that investigate the sea-ice cover in the context of climate change. Moreover, it supports operational use of remotely sensed sea-ice thickness information for international sea-ice monitoring programs and climate forecasting systems.

A long climate data record (CDR) of Arctic and Antarctic ice Surface Temperatures from thermal Infrared satellite sensors (AASTI) is produced. The AASTI CDR covers high latitude Seas, the Marginal Ice Zone, Sea Ice and Ice Cap surface temperatures North of 50 degree North and South of 50 degree South for the period 1982 to 2015. The CDR is based on Global Area Coverage (GAC) data from AVHRR instruments on board the NOAA and METOP satellites. AASTI is produced as a swath based surface temperature CDR in satellite view projection (level 2) and as a daily mean surface temperature record in a regular grid (level-3). The resolution of the level-2 data set is multiple daily swath files with approximately4 km spatial resolution and the level-3 data are daily values in ~0.25 arc degree resolution. Beside the daily mean temperature, the level 3 dataset also contains 3 hourly mean temperature fields in order to preserve information on diurnal variability.

The Ice and Sea Surface Temperature (IST/SST) retrieval algorithms are adapted from the operational IST/SST production at EUMETSAT’s Ocean and Sea Ice Satellite Application Facility (OSISAF). The data set errors (STD) and biases are estimated for three environments against traditional in situ instruments, i.e. weather station air temperatures and drifting buoys: Arctic sea ice, STD=3.0 K and bias=-2.1 K; North Atlantic water, STD=0.7 K and bias=-0.3 K; Greenland Summit, STD=2.6 K and bias=-3.1 K. It has previously been documented that temperatures from these platforms may deviate from the actual surface temperature and so the bias and STD in the comparison also stems from the validation data variability. AASTI uncertainties using in situ radiometer data as a reference are estimated between 1 and 2 K, depending on location.

Based on the AASTI data set the distributed surface temperature climatology is calculated along with the first order temperature trends for the Arctic Ocean and Southern Ocean and ice caps. These trends show large geographical variations.

The sea ice on the polar oceans is a small part of the global climate system but it controls the regional exchange of heat, mass, and momentum between ocean and atmosphere. Changes in sea ice concentration can affect patterns of atmospheric circulation even at lower latitudes and are key information to initialize models for reliable prediction of its evolution on scales from days or seasons to decades. Systematic and long-term observations of the major sea ice variables are only possible using past and present satellite Earth observation data. Sea ice data from satellites has been collected for more than four decades; sea ice mapping is one of the most successful applications of Earth observation satellite data in climate change studies. During the last 25 years, the increase of the Arctic surface air temperature has exceeded the global warming by a factor of two. This Arctic amplification results in dramatic changes of a variety of climate parameters. In particular, the Arctic sea ice has declined and got significantly thinner, also affecting the biodiversity. The summer sea ice extent reduction is one of the major climate change extremes. Coupled regional/global climate models still fail reproducing this decline: They underestimate the observed sea ice reduction, showing that the physical processes and feedback mechanisms are not yet well represented in Arctic models. The so-called Arctic amplification has repercussions to lower latitudes but prediction of these models is yet inadequate.
Passive microwave sensors with their independence to daylight and ability to penetrate clouds have been used for sea ice applications since the 70s, and the global Sea Ice Concentration (SIC) from microwave radiometers is one of the longest continuous climate records. However, the retrieval accuracy of the SIC from satellite observations is still limited essentially due to the spatial resolution of current microwave sensors at low frequencies. The high microwave frequencies already provide good SIC spatial resolution, but are stronger affected by the atmospheric disturbance which the so-called weather filters cannot completely compensate for (Spreen et al., 2008).
The Copernicus Imaging Microwave Radiometer (CIMR) mission is a high priority candidate mission within the European Copernicus Expansion program, primarily designed to observe the Arctic environment. It is a conically scanning microwave radiometer imager with incidence angle ~55° in a sun-synchronous polar orbit with no observation holes at the poles. It will include channels at 1.4, 6.9, 10.65, 18.7, and 36.5 GHz, with spatial resolution of better than 5 km at 18.7 and 36.5 GHz, better than 15 km at 6.9 and 10.65 GHz, and around 60 km at 1.4 GHz. It will provide SIC and Sea Surface Temperature with an increased precision and/or spatial resolution, along with other sea ice, ocean and land parameters (Kilic et al., 2018).

The current SIC algorithms (e.g., Spreen et al., 2008; Tonboe et al., 2016; Lavergne et al., 2018) are based on key works from Cavalieri et al. (1984, the ‘NASA Team’ algorithm), Comiso (1986, the ‘Bootstrap’ algorithm), and Smith et al. (1994, the ‘Bristol’ algorithm).They rely on regression formulas that use empirically derived so-called tie-points (i.e. typical radiometric signatures of ice or water) to scale the relatively large contrast between ice and water at microwave frequencies. They can include an estimate of the uncertainty, along with a regular update of the tie-points to better account for the variability of the surface emissivity and for the changes in the instruments over the time-record. However, these algorithms have difficulties optimizing the use a large number of channels and to handle the spatial resolution differences.

In order to fully benefit from the multi-frequency observations of CIMR at high spatial resolution, we propose to revisit the SIC algorithm methodology. The scheme is also based on the use of the empirically derived tie-points provided by the Round Robin datasets, a collection of satellite observations at known ground conditions like 100% and 0% SIC (Pedersen and Saldo, 2016). The here derived linear model is embedded in an optimal estimation framework that facilitates the combination of multi-channels and easily provides uncertainty estimates. It can handle dynamic tie-points to better fit the varying surface emissivities. It can directly use calculation of open ocean sea surface emissivity estimates, using ancillary information (from reanalysis for instance). The combination of the spatial resolutions of the different channels is rigorously treated, within the optimal estimation framework. The algorithm maximizes the potential of the low microwave frequency sensitivity to sea ice while maintaining the high spatial resolution of the higher microwave frequency. The algorithm will be presented. It will be applied to AMSR data for evaluation by comparison with MODIS observations.

References:
Cavalieri D. J., P. Gloersen, and W. J. Campbell (1984), Determination of sea ice parameters with the NIMBUS 7 SMMR., J. Geophys. Res., 89, D4, 5355-5369.
Comiso J.C. (1986) Characteristics of arctic winter sea ice from satellite multispectral microwave observations. J. Geophys. Res., 91(C1), 975-994.
Kilic, L., Prigent, C., Aires, F., Boutin, J., Heygster, G., Tonboe, R. T. et al. (2018). Expected performances of the Copernicus Imaging Microwave Radiometer (CIMR) for an all-weather and high spatial resolution estimation of ocean and sea ice parameters. J. Geophys. Res., 123, doi: 10.1029/2018JC014408.
Lavergne, T., Sørensen, A. M., Kern, S., Tonboe, R., Notz, D., Aaboe, S., et al. (2018), Version 2 of the EUMETSAT OSI SAF and ESA CCI Sea Ice Concentration Climate Data Records, The Cryosphere Discuss., doi :10.5194/tc-2018-127, in review.
Pedersen, L. F., and R. Saldo (2016), Sea ice concentration (sic) round robin data package, sea ice climate initiative: Phase 2, Tech. Rep. SICCI-RRDP-07-16 Version: 1.4, ESA, DTU.
Spreen, G., L. Kaleschke, and G. Heygster (2008), Sea ice remote sensing using AMSR-E 89 GHz channels J. Geophys. Res.,113, C02S03, doi:10.1029/2005JC003384.

Climate change is amplified in the Arctic region relative to elsewhere, and there is growing evidence that accelerated change is happening at high northern latitudes. This Arctic amplification has also been found in past changes in warm and glacial climates, as well as in historical simulations. The phenomenon is often explained by retreating snow and ice leading to more solar surface warming (surface albedo feedback). However, by analyzing climate model simulations Pithan and Mauritsen (2014) found that the largest contribution to Arctic amplification comes from temperature feedbacks, due to the smaller increase in heat loss by longwave emission per unit of warming at colder temperatures compared to tropical conditions. Satellite remote sensing offers the best way of deriving surface temperatures in the Arctic, but given that the surface temperature retrieval algorithms in the infrared are designed to compensate for the effects of the atmosphere, mainly water vapor, satellite-derived surface temperatures have larger uncertainties at high latitudes because the atmosphere is very dry and cold. When the water vapor concentrations are low, the correction algorithms tend to over-compensate leading to warm biases. So, the motivation of the study is to improve the algorithms to obtain more accurate surface temperatures which can be used to research the feedback mechanisms. To undertake the study, we use collocated, simultaneous satellite measurements of brightness temperature at the top of atmosphere and in situ measurements of surface temperature. We have analyzed the matchup databases for MODIS on Aqua, Terra and for VIIRS on S-NPP to characterize the differences between satellite retrieved temperatures and in-situ measurements, and to identify the main causes of the discrepancies. This leads to regional optimization of the surface temperature retrievals. We will report on the progress towards improving the satellite-derived surface temperatures with the expectation that the near two-decadal time series of MODIS surface temperature fields will contribute to studying climate change in the Arctic.

Pithan, F., & Mauritsen, T. (2014). Arctic amplification dominated by temperature feedbacks in contemporary climate models. Nature Geoscience 7, 181. 10.1038/ngeo2071

Manual ice charting from multi-sensor satellite data analysis has for many years been the method at the National Ice Centers around the world for producing sea ice information for marine safety. Ice charts are today made by ice analysts that use all relevant imagery available; primarily Copernicus Sentinel-1 SAR imagery, due to the radar sensor high resolution and capability to see through clouds and in polar darkness, but also optical imagery, thermal-infrared and microwave radiometer data from eg. AMSR-2. Ice charting is a time-consuming method that is increasingly challenged by the vast amount of free and availiable satellite imagery these years. Along with a growing number of users accessing wider parts of the Arctic due to the retreat and thinning of the Arctic sea ice, this calls for a more effective way of producing detailed and timely ice informations to the users.

In the ASIP project (Automated downstream Sea Ice Products) our aim is to partially or fully automate the generation of sea ice information from Sentinel-1 SAR imagery. This danish research and development project is carried out in collaboration between the Danish Meteorological Institute (DMI), the Technical University of Denmark and Harnvig Arctic and Maritime.These partners together holding a wide expertise within active and passive microwave signature analysis, sea ice applications development and advanced computing methods. The project is funded by the Innovation Fund Denmark and partners.

The design of an automatic and robust sea ice classification scheme has been studied for many years and many attempts have been made with limited success. Here we present our vision for adressing the issue; merging the Sentinel-1 active microwave data with other satellite sensor data that have complementary capabilities, such as passive microwave data (AMSR-2) to resolve the ambiguities that can occur in SAR imagery eg. at high wind speeds.

We use Convolutional Neural Networks (CNN) to generate ice maps from the Sentinel-1 imagery. The CNNs have been trained with a vast data set of approx. 900 ice charts manually produced by the ice analysts in the DMI Ice Service based on Sentinel-1 imagery. The ice chart data set spans a wide area around the Greenland coast and cover all seasons from November 2014 until December 2017. We will present the results of this study. We will present our plans for merging the Sentinel-1 active microwave data with AMSR-2 passive microwave data and possibly present preliminary results from using AMSR-2 data as input to the CNN also.

If able to automate the manual ice charting method, ASIP can make ice services more efficient and able to fully utilize the vast amount of Sentinel-1 data that is available, for delivering better and faster ice services to users. Ultimately, the maritime users want reliable ice forecasts for safe navigation and planning. And to be able to provide these, there is a need for these future automated, detailed and standardized ice observations from satellite data for integration in forecast systems and models. It is a project goal to test an assimilation of the future automated ASIP ice information products in the DMI HYCOM-CICE model and to do demonstration services to users.