Satellite Passive Microwave Sea-Ice Concentration Data Set Intercomparison for Arctic Summer Conditions

Abstract

We report on results of a systematic inter-comparison of 10 global sea-ice concentration (SIC) data products at 12.5 to 50.0 km grid resolution from satellite passive microwave (PMW) observations for the Arctic during summer. The products are compared against SIC and net ice-surface fraction (ISF) -SIC minus the per-grid cell melt-pond fraction (MPF) on sea ice -as derived from MODerate resolution Imaging Spectroradiometer (MODIS) satellite observations and observed from ice-15 going vessels. Like in Kern et al. (2019), we group the 10 products based on the concept of the SIC retrieval used. Group I consists of products of the EUMETSAT OSI SAF and ESA CCI algorithms. Group II consists of products derived with the Comiso bootstrap algorithm and the NOAA NSIDC SIC climate data record (CDR). Group III consists of ARTIST Sea Ice (ASI) and NASA Team (NT) algorithm products and group IV consists of products of the enhanced NASA Team algorithm (NT2). We find wide-spread positive and negative differences between PMW and MODIS SIC with magnitudes frequently 20 reaching up to 20-25 % for groups I and III and up to 30-35 % for groups II and IV. On a pan-Arctic scale these differences may cancel out: Arctic average SIC from Group I products agrees with MODIS within 2-5 % accuracy during the entire melt period from May through September. Group II and IV products over-estimate MODIS Arctic average SIC by 5-10 %. Out of group III, ASI is similar to group I products while NT SIC under-estimates MODIS Arctic average SIC by 5-10 %. These differences, when translated into the impact computing Arctic sea-ice area (SIA), match well with the differences in SIA 25 between the four groups reported for the summer months by Kern et al. (2019). MODIS ISF is systematically over-estimated by all products; NT provides the smallest (up to 25 %) over-estimations, group II and IV products the largest (up to 45 %) over-estimations. The spatial distribution of the observed over-estimation of MODIS ISF agrees reasonably well with the spatial distribution of the MODIS MPF and we find a robust linear relationship between PMW SIC and MODIS ISF for group I and III products during peak melt, i.e. July and August. We discuss different cases taking into account the expected influence 30 of ice-surface properties other than melt ponds, i.e. wet snow and coarse grained snow / refrozen surface, on PMW observations used in the SIC retrieval algorithms. Based on this discussion we identify the mismatch between the actually observed surface properties and those represented by the ice tie points as the most likely reason for i) the observed differences between PMW SIC and MODIS ISF and for ii) the often surprisingly small difference between PMW and MODIS SIC in areas of high meltpond fraction. We conclude that all 10 SIC products are highly inaccurate during summer melt. We hypothesize that the un-35 known amount of melt-pond signatures likely included in the ice tie points plays an important role -particularly for groups I and II -and suggest to conduct further...

Trends, abrupt shifts and interannual variability of the Arctic Wintertime Seasonal Sea Ice from 1979 to 2019

“…The average standard deviation for the Arctic is <6% in winter for both low and high sea-ice concentration (SIC) (Ivanova and others, 2015). Recent results show that the product underestimates SIC by 5–10% in summer (Kern and others, 2020) and overestimates SIC by 0.9% for 100% sea-ice cover (Kern and others, 2019). An additional advantage of using a set of 19 and 37 GHz algorithms is that the dataset extends from fall 1978 through today and into the foreseeable future (Ivanova and others, 2015).…”

Reducing Weather Influences on an 89 GHz Sea Ice Concentration Algorithm in the Arctic Using Retrievals From an Optimal Estimation Method

“…Unfiltered (no weather filter) and untruncated (allowing retrievals outside the 0%-100% range) SIC data are produced in the EUMETSAT OSI SAF and ESA CCI sea ice concentration climate data records (Lavergne et al, 2019), and have been used in recent inter-algorithm comparison studies (Kern et al, 2019(Kern et al, , 2020. Such datasets help to avoid possible inter-sensor bias caused by the choice of weather filter thresholds, and can better represent the algorithm uncertainty around the boundary SIC values.…”

“…(2015) reviewed 30 algorithms, and Kern et al. (2020) compared 10 SIC data products. These studies also provide useful background information that is not repeated here.…”

Assessment of the Stability of Passive Microwave Brightness Temperatures for NASA Team Sea Ice Concentration Retrievals

“…This is particularly prevalent in the Antarctic where nearly all of the seasonal ice cover is lost during the summer. Another factor, particularly in the Arctic, is the well-known surface melt and melt pond effect, e.g., [17]. At the microwave frequencies used by the NASA team algorithm (and most other algorithms), emission comes from the surface or very near the surface.…”

“…Each product has been found to have limitations, particularly during summer melt where surface water and melt ponds on the ice are interpreted by the algorithms as open water [16]. The NASA team is particularly susceptible to underestimation of concentration during summer melt, e.g., [17].…”