A New Bias Correction Approach for Better Assimilation of Microwave Sounding Data Over Winter Sea Ice in the Korean Integrated Model

Microwave sounder observations are essential for numerical weather prediction (NWP) systems, but utilizing channels sensitive to surface over sea ice has been challenging due to difficulties in estimating the sea ice surface radiance. This study presents a preprocessing method to assimilate near-surface microwave-sounding observations over winter sea ice, including an estimation of a real-time surface emissivity from satellite radiance and a bias correction scheme to minimize the radiance discrepancy between observation and model simulation. Our results show that the radiance simulated using dynamic emissivity exhibits a much better agreement with the measured one, although a significant negative bias of about 0.61–1.18 K remains over the winter sea ice. Thus, a new bias correction procedure, based on the regression relationships between the residual bias and potential bias sources such as the surface temperature and surface emissivity, is added. When it is applied, the remained bias is successfully estimated. Moreover, the sea ice observations from all temperature-sounding channels have been better utilized in the Korean Integrated Model (KIM). The additional information on the polar regions has increased the analysis increment and reduced the ensemble spread. In addition, a neutral to slightly positive impact on temperature analysis errors in layers sensitive to surface radiance encourages further utilization of microwave sounder data over sea ice.