Search

Sea ice information for the near coastal areas of the Greenlandic waters is of high importance forthe local communities and the maritime industry. The “truth” within sea ice information hastraditionally been associated with Manual Ice Charts; however, the demand for accurate forecastsis increasing.At first, this study will introduce a variety of satellite-based Copernicus marine service productswaters with a special focus on a novel automated ice chart that runs on a daily basis at the DanishMeteorological Institute (DMI). The new product is based on a Convolutional Neural Network(CNN), which combines passive microwave and SAR imagery in order to optimize retrieval. Bydoing so, it produces the best possible sea ice concentration with a resolution comparable to themanual ice charts.Secondly, this study presents an improved operational forecast system for the Arctic sea icefocusing on the Greenlandic waters. The physical basis of the system is close to the Arctic Marineforecasting system within the Copernicus Marine System. This presentation will present theforecast system and introduce the first attempts to assimilate a combination of level two data fromthe automated ice charts gap-filled with level 2 passive microwave data.We validate the sea ice edge forecast systems and the individual remotely sensed observationalproducts by computing the Integrated Ice Edge Error metric. This comparison is focused primarilyon the initial state and secondly on a comparison with the initial state.

The Arctic region is responding heavily to climate change, and yet, the air temperature of ice-covered areas in the Arctic is heavily under-sampled when it comes to in situ measurements, resulting in large uncertainties in existing weather and reanalysis products. This paper presents a method for estimating daily mean clear-sky 2 m air temperatures (T2m) in the Arctic from satellite observations of skin temperature, using the Arctic and Antarctic ice Surface Temperatures from thermal Infrared (AASTI) satellite dataset, providing spatially detailed observations of the Arctic. The method is based on a linear regression model, which has been tuned against in situ observations to estimate daily mean T2m based on clear-sky satellite ice surface skin temperatures. The daily satellite-derived T2m product includes estimated uncertainties and covers the Arctic sea ice and the Greenland Ice Sheet during clear skies for the period 2000–2009, provided on a 0.25∘ regular latitude–longitude grid. Comparisons with independent in situ measured T2m show average biases of 0.30 and 0.35∘C and average root-mean-square errors of 3.47 and 3.20 ∘C for land ice and sea ice, respectively. The associated uncertainties are verified to be very realistic for both land ice and sea ice, using in situ observations. The reconstruction provides a much better spatial coverage than the sparse in situ observations of T2m in the Arctic and is independent of numerical weather prediction model input. Therefore, it provides an important supplement to simulated air temperatures to be used for assimilation or global surface temperature reconstructions. A comparison of T2m derived from satellite and ERA-Interim/ERA5 estimates shows that the satellite-derived T2m validates similar to or better than ERA-Interim/ERA5 against in situ measurements in the Arctic.

Day-to-day variations in surface air temperature affect society in many ways, but daily surface air temperature measurements are not available everywhere. Therefore, a global daily picture cannot be achieved with measurements made in situ alone and needs to incorporate estimates from satellite retrievals. This article presents the science developed in the EU Horizon 2020–funded EUSTACE project (2015–19, www.eustaceproject.org) to produce global and European multidecadal ensembles of daily analyses of surface air temperature complementary to those from dynamical reanalyses, integrating different ground-based and satellite-borne data types. Relationships between surface air temperature measurements and satellite-based estimates of surface skin temperature over all surfaces of Earth (land, ocean, ice, and lakes) are quantified. Information contained in the satellite retrievals then helps to estimate air temperature and create global fields in the past, using statistical models of how surface air temperature varies in a connected way from place to place; this needs efficient statistical analysis methods to cope with the considerable data volumes. Daily fields are presented as ensembles to enable propagation of uncertainties through applications. Estimated temperatures and their uncertainties are evaluated against independent measurements and other surface temperature datasets. Achievements in the EUSTACE project have also included fundamental preparatory work useful to others, for example, gathering user requirements, identifying inhomogeneities in daily surface air temperature measurement series from weather stations, carefully quantifying uncertainties in satellite skin and air temperature estimates, exploring the interaction between air temperature and lakes, developing statistical models relevant to non-Gaussian variables, and methods for efficient computation.

The Greenland Ice Sheet (GIS) is subject to amplified impacts of climate change and its monitoring is essential for understanding and improving scenarios of future climate conditions. Surface temperature over the GIS is an important variable as it regulates processes related to the exchange of energy and water between the surface and the atmosphere. As few key local observation sites exist, an important alternative to obtain surface temperature observations over the GIS is space-borne sensors that carry thermal infrared instruments. These offer several passes per day with a wide view and are the basis of deriving Ice Surface Temperature (IST) products. The aim of this study was to compare several satellite IST products for the GIS and develop and validate the first multi-sensor, gap-free (Level 4, L4) product for 2012. High resolution Level 2 (L2) IST products from the European Space Agency (ESA) Land Surface Temperature Climate Change Initiative (LST_cci) project and the Arctic & Antarctic Ice Surface Temperatures from Thermal Infrared Satellite Sensors (AASTI) dataset, were assessed using observations from the PROMICE stations and IceBridge flight campaigns. The AASTI data showed overall better performance compared to LST_cci data, that in return had superior spatial coverage and availability. Both datasets were further utilised to construct a daily, gap-free, L4 IST product using the optimal interpolation (OI) method. The resulting L4 IST product performed satisfactorily in terms of quality when compared with surface temperature observations from the PROMICE stations and IceBridge flight campaigns. By combining the advantages of the upstream satellite datasets, the gap-free L4 IST product allowed for the analysis of IST over the GIS during the year 2012, when a significant melt event occurred. Mean summer (June–August) IST over the GIS was −5.5 °C ± 4.5 °C, with an annual mean of −22.1 °C ± 5.4 °C. Mean IST during the melt season (May–August) ranged from −15 °C to −1 °C, while almost the entire GIS experienced at least between 1 and 5 melt days when temperatures were −1 °C or higher. Finally, this study assessed the potential for using the satellite L4 IST product to improve model simulations of the GIS surface mass budget (SMB). The new L4 IST product was first used to evaluate, and then it was assimilated into an SMB model of snow and firn processes for the year 2012, to assess the impact of including a high resolution IST product on the SMB model. Compared with independent observations from the PROMICE stations and IceBridge flight campaigns, inclusion of the L4 IST dataset improved the model performance during the key onset of the melt season, where model biases are typically large. Reassuringly, internal model parameterisations performed well compared to the L4 IST dataset outside of this season, providing more confidence in modelled GIS surface mass budget (SMB) estimates.

The Greenland Ice Sheet (GIS) is subject to amplified impacts of climate change and its monitoring is essential for understanding and improving scenarios of future climate conditions. Surface temperature over the GIS is an important variable, regulating processes related to the exchange of energy and water between the surface and the atmosphere. Few local observation sites exist; thus spaceborne platforms carrying thermal infrared instruments offer an alternative for surface temperature observations and are the basis for deriving ice surface temperature (IST) products. In this study several satellite IST products for the GIS were compared, and the first multi-sensor, gap-free (Level 4, L4) product was developed and validated for 2012. High-resolution Level 2 (L2) products from the European Space Agency (ESA) Land Surface Temperature Climate Change Initiative (LST_cci) project and the Arctic and Antarctic Ice Surface Temperatures from Thermal Infrared Satellite Sensors (AASTI) dataset were assessed using observations from the PROMICE (Programme for Monitoring of the Greenland Ice Sheet) stations and IceBridge flight campaigns. AASTI showed overall better performance compared to LST_cci data, which had superior spatial coverage and availability. Both datasets were utilised to construct a daily, gap-free L4 IST product using the optimal interpolation (OI) method. The resulting product performed satisfactorily when compared to surface temperature observations from PROMICE and IceBridge. Combining the advantages of satellite datasets, the L4 product allowed for the analysis of IST over the GIS during 2012, when a significant melt event occurred. Mean summer (June–August) IST was −5.5 ± 4.5 ∘C, with an annual mean of −22.1 ± 5.4 ∘C. Mean IST during the melt season (May–August) ranged from −15 to −1 ∘C, while almost the entire GIS experienced at least between 1 and 5 melt days when temperatures were −1 ∘C or higher. Finally, this study assessed the potential for using the satellite L4 IST product to improve model simulations of the GIS surface mass balance (SMB). The L4 IST product was assimilated into an SMB model of snow and firn processes during 2012, when extreme melting occurred, to assess the impact of including a high-resolution IST product on the SMB model. Compared with independent observations from PROMICE and IceBridge, inclusion of the L4 IST dataset improved the SMB model simulated IST during the key onset of the melt season, where model biases are typically large and can impact the amount of simulated melt.

Two retrieval algorithms developed as a part of the European Space Agency Climate Change Initiative (ESA-CCI) project are used to assess the effects of withholding observations from selected frequency channels on the retrieved subskin Sea Surface Temperature (SST) from AQUA's Advanced Microwave Scanning Radiometer—Earth Observing System (AMSR-E) and to evaluate a Copernicus Imaging Microwave Radiometer (CIMR) like channel configuration. The first algorithm is a statistical regression-based retrieval algorithm, while the second is a physically based optimal estimation (OE) algorithm. A database with matching satellite and drifting buoy observations is used to test the performance of each channel configuration using both retrieval algorithms to identify the most optimal channel selection for accurate SST retrievals. The evaluation against in situ observations allows identification of the strengths and weaknesses of the two retrieval algorithms, and demonstrates the importance of using in situ observations to evaluate existing theoretical retrieval uncertainty studies. Overall, the performance increases as expected when more channels are included in the retrieval. In particular, more channels allow a better performance for the range of different observing conditions (e.g. cold waters). The two retrieval algorithms agree that for a three-channel configuration, the 6, 10, 18 GHz (V and H polarization) is better than the 6, 10, 23 GHz configuration (V and H polarization). This is demonstrated for different geographical regions and throughout all seasons. Of the different combinations tested here, it is evident that withholding observations from the 23 and 36 GHz channels from the retrieval has the least impact on the SST performance. Overall, this analysis shows that the CIMR like channel configuration performs very well when compared to an AMSR-E like constellation using both retrieval algorithms.

We establish a methodology for assimilating satellite observations of ice surface temperature (IST) into a coupled ocean and sea-ice model. The method corrects the 2 meter air temperature based on the difference between the modelled and the observed IST. Thus the correction includes biases in the surface forcing and the ability of the model to convert incoming parameters at the surface to a net heat flux. A multi-sensor, daily, gap-free surface temperature analysis has been constructed over the Arctic region. This study revealed challenges estimating the ground truth based on buoys measuring IST, as the quality of the measurement varied from buoy to buoy. With these precautions we find a cold temperature bias in the remotely sensed data, and a warm bias in the modelled data relative to ice mounted buoy temperatures, prior to assimilation. As a consequence, this study weighted the modelled IST and the observed IST equally in the correction. The impact of IST was determined for experiments with and without the assimilation of IST and sea-ice concentration. We find that assimilation of remotely sensed data results in a cooling of IST, which improves the timing of the snow melt onset. The improved snow cover in spring is only based on observations from one buoy, thus additional good quality observations could strengthen the conclusions. The ice cover and the sea-ice thickness are increased, primarily in the experiment without sea-ice concentration assimilation.

The Arctic region is responding heavily to climate change, and yet, the air temperature of Arctic, ice covered areas is heavily under-sampled when it comes to in situ measurements, and large uncertainties exist in weather- and reanalysis products. This paper presents a method for estimating daily mean 2 meter air temperatures (T2m) in the Arctic from satellite observations of skin temperature, using the Arctic and Antarctic ice Surface Temperatures from thermal Infrared (AASTI) satellite dataset, providing spatially detailed observations of the Arctic. The method is based on a linear regression model which has been developed using in situ observations and daily mean satellite ice surface skin temperatures combined with a seasonal variation to estimate daily T2m. The satellite derived T2m product including estimated uncertainties covers clear sky snow and ice surfaces in the Arctic region during the period 2000–2009. Comparison with independent in situ measured T2m gives average correlations of 95.5 % and 96.5 % and average root mean square errors of 3.47 °C and 3.19 °C for land ice and sea ice, respectively. The reconstruction provides a much better spatial coverage than the sparse in situ observations of T2m in the Arctic, is independent of numerical weather prediction model input and it therefore provides an important alternative to simulated air temperatures to be used for assimilation or global surface temperature reconstructions. A comparison between in situ T2m versus T2m from satellite and ERA-Interim shows that the T2m derived from satellite observations validate similar or better than ERA-Interim estimates in the Arctic.

To facilitate the construction of a satellite-derived 2 m air temperature (T2 m) product for the snow- and ice-covered regions in the Arctic, observations from weather stations are used to quantify the relationship between the T2 m and skin temperature (Tskin). Multiyear data records of simultaneous Tskin and T2 m from 29 different in situ sites have been analysed for five regions, covering the lower and upper ablation zone and the accumulation zone of the Greenland Ice Sheet (GrIS), sea ice in the Arctic Ocean, and seasonal snow-covered land in northern Alaska. The diurnal and seasonal temperature variabilities and the impacts from clouds and wind on the T2 m–Tskin differences are quantified. Tskin is often (85 % of the time, all sites weighted equally) lower than T2 m, with the largest differences occurring when the temperatures are well below 0 ∘C or when the surface is melting. Considering all regions, T2 m is on average 0.65–2.65 ∘C higher than Tskin, with the largest differences for the lower ablation area and smallest differences for the seasonal snow-covered sites. A negative net surface radiation balance generally cools the surface with respect to the atmosphere, resulting in a surface-driven surface air temperature inversion. However, Tskin and T2 m are often highly correlated, and the two temperatures can be almost identical ( ∘C difference), with the smallest T2–Tskin differences around noon and early afternoon during spring, autumn and summer during non-melting conditions. In general, the inversion strength increases with decreasing wind speeds, but for the sites on the GrIS the maximum inversion occurs at wind speeds of about 5 m s−1 due to the katabatic winds. Clouds tend to reduce the vertical temperature gradient, by warming the surface, resulting in a mean overcast T2 m–Tskin difference ranging from −0.08 to 1.63 ∘C, with the largest differences for the sites in the low-ablation zone and the smallest differences for the seasonal snow-covered sites. To assess the effect of using cloud-limited infrared satellite observations, the influence of clouds on temporally averaged Tskin has been studied by comparing averaged clear-sky Tskin with averaged all-sky Tskin. To this end, we test three different temporal averaging windows: 24 h, 72 h and 1 month. The largest clear-sky biases are generally found when 1-month averages are used and the smallest clear-sky biases are found for 24 h. In most cases, all-sky averages are warmer than clear-sky averages, with the smallest bias during summer when the Tskin range is smallest.

A statistical regression-based microwave sea surface temperature (SST) retrieval algorithm has been developed within the European Space Agency Climate Change Initiative (ESA-CCI) SST project. The retrieval algorithm was used to generate a climate data record (CDR) of passive microwave (PMW) SST from the Advanced Microwave Scanning Radiometer – Earth Observing System (AMSR-E) and its follow-on instrument AMSR2 for the period June 2002–October 2017. Multisensor Matchup Datasets (MMDs), which includes satellite orbital data collocated with in situ and auxiliary data, was used to derive consistent algorithms for AMSR-E and AMSR2. The retrieval algorithms consist of wind speed (WS) retrievals and SST retrievals, with corresponding uncertainty retrievals. The WS retrieval consists of a two-step regression model, where the second step is a localized algorithm, trained to perform well over restricted WS intervals. A two-step multiple linear regression retrieval with localized algorithms is used to retrieve SST. The first-stage algorithm is trained to perform well over restricted latitude intervals for ascending and descending orbit, respectively, whereas the second-stage retrieval uses localized SST and WS algorithms. Furthermore, a new and effective method for detecting and screening for Radio Frequency Interference (RFI) was developed. Validation of the PMW SSTs against drifter in situ SSTs shows an overall bias of −0.02 K for quality level (QL) 4 and 5 AMSR-E retrievals with a standard deviation of 0.46 K. Validation results for QL 4 and 5 AMSR2 retrievals against drifter in situ SSTs give a bias of 0.002 K and a standard deviation of 0.45 K. The corresponding mean modelled SST uncertainties, including in situ and sampling uncertainties, are estimated to 0.45 K for QL 4 and 5 AMSR-E retrievals and 0.44 K for QL 4 and 5 AMSR2 retrievals. Validation against near-surface temperature measurements from Argo floats yielded comparable results, confirming the drifting-buoy validation. The validation results demonstrate a consistent PMW SST CDR with accurate SST observations and reliable uncertainty estimates.

To facilitate the combined use of traditional 2 m air temperature (T2m) observations from weather stations in the Arctic and skin temperature (Tskin) observations from satellites the relationship between high latitude snow and ice Tskin and T2m is quantified. Multiyear data records of simultaneous Tskin and T2m from 20 different in situ sites have been analysed, covering the Greenland Ice Sheet (GrIS), sea ice in the Arctic Ocean, and coastal snow covered land in North Alaska. The diurnal and seasonal temperature variabilities and the impacts from clouds and wind on the T2m-Tskin differences are quantified. Considering all stations, T2m is on average 1.37 °C warmer than Tskin, with the largest differences at the GrIS stations (mean of diff. of 1.64 °C). Tskin and T2m are often highly correlated, and the two temperatures are almost identical (−1, likely due to the katabatic winds, which are most prominent at this wind speed. Clouds tend to reduce the vertical temperature gradient, by warming the surface, resulting in a mean T2m-Tskin difference of 0.53 °C considering all stations. Following that the influence of clouds on Tskin has been assessed by comparing clear-sky Tskin observations with all-sky observations averaged for the time windows of: 24 h, 72 h and 1 month. The largest clear-sky biases are generally found when 1 month averages are used and smallest for 24 h. The mean clear-sky bias for the 24 h average is −0.28 °C, ranging from −0.10 °C in summer to −0.95 °C in winter. The expected clear-sky biases and the difference between Tskin and T2m are of practical value for researchers and operational users that aim at integrating satellite observations with ocean, sea-ice or atmospheric models.

The Optimal Estimation (OE) technique is developed within the European Space Agency Climate Change Initiative (ESA-CCI) to retrieve subskin Sea Surface Temperature (SST) from AQUA’s Advanced Microwave Scanning Radiometer—Earth Observing System (AMSR-E). A comprehensive matchup database with drifting buoy observations is used to develop and test the OE setup. It is shown that it is essential to update the first guess atmospheric and oceanic state variables and to perform several iterations to reach an optimal retrieval. The optimal number of iterations is typically three to four in the current setup. In addition, updating the forward model, using a multivariate regression model is shown to improve the capability of the forward model to reproduce the observations. The average sensitivity of the OE retrieval is 0.5 and shows a latitudinal dependency with smaller sensitivity for cold waters and larger sensitivity for warmer waters. The OE SSTs are evaluated against drifting buoy measurements during 2010. The results show an average difference of 0.02 K with a standard deviation of 0.47 K when considering the 64% matchups, where the simulated and observed brightness temperatures are most consistent. The corresponding mean uncertainty is estimated to 0.48 K including the in situ and sampling uncertainties. An independent validation against Argo observations from 2009 to 2011 shows an average difference of 0.01 K, a standard deviation of 0.50 K and a mean uncertainty of 0.47 K, when considering the best 62% of retrievals. The satellite versus in situ discrepancies are highest in the dynamic oceanic regions due to the large satellite footprint size and the associated sampling effects. Uncertainty estimates are available for all retrievals and have been validated to be accurate. They can thus be used to obtain very good retrieval results. In general, the results from the OE retrieval are very encouraging and demonstrate that passive microwave observations provide a valuable alternative to infrared satellite observations for retrieving SST.