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

Sea ice concentration (SIC) derived from 89 GHz data has up to four times finer spatial resolution compared to that from the widely used 19 and 37 GHz data. But it has lower accuracy due to the enhanced weather influences from water vapor, cloud liquid water (CLW), wind, and surface temperature. Here we improve a high‐resolution SIC algorithm, called the ASI algorithm, based on the difference between vertical and horizontal polarization 89 GHz data, by correcting the observed data for these weather influences through a radiative transfer model and geophysical data retrieved by an optimal estimation method. The improved algorithm denoted ASI3, is developed for the Arctic based on the weather‐corrected brightness temperatures and newly identified open water (80 K) and sea ice (14 K) tie‐points. The most important component of this correction is the inclusion of CLW, the largest weather influence contributor. ASI3 results are evaluated over pure surface sites of 0% and 100% SICs under various weather conditions, showing a much lower average standard deviation (1.1%) than ASI (16.2%). ASI3 reduces weather patterns over pack ice resulting in more homogeneous retrievals but biased toward lower values. Comparison to Landsat imagery under clear‐sky conditions shows that ASI3 results in better agreement with the Landsat SIC than ASI. The number of cases where real sea ice is falsely identified as open water is reduced by ASI3 due to its relaxed open‐water mask and wider water/ice dynamic range. Plain Language Summary: The recent decline of Arctic sea ice is a key manifestation of climate change. We use satellite microwave radiometers to monitor sea ice globally because they can observe under all weather conditions and also during the polar night. However, they only provide observations at a coarse spatial resolution of several to tens of kilometers. Sea ice concentration (SIC) describes what fraction of a known area is covered by sea ice. Radiometers that include the 89 GHz channels measure at finer spatial resolutions but are also affected more by atmospheric noise caused by water vapor and cloud liquid water (CLW), which causes errors in the derived SIC. Here, we present a new algorithm that corrects this weather‐induced noise and estimates SIC from the 89 GHz data of the AMSR‐E and AMSR2 satellite microwave radiometers. We estimate the atmospheric influence by modeling it, and then subtract it from the 89 GHz measurements. We particularly focus on correcting the influence from CLW, as it is the most important error source. Compared to the previous 89 GHz algorithm without weather correction, we find that the new algorithm provides SIC data with much lower error while maintaining the fine spatial resolution of 5 km. Key Points: An improved sea ice concentration retrieval algorithm based on weather‐corrected brightness temperatures at 89 GHz is developedWeather influences particularly from cloud liquid water are reduced, lowering the retrieval uncertainty by one order of magnitudeNoise in ice concentration caused by the atmosphere is much reduced over consolidated ice [ABSTRACT FROM AUTHOR]