Identifying and Quantifying Temperature Dependent Biases in the CRTM Ocean Emissivity Model using the NCEP Global Data Assimilation System.

An accurate ocean infrared surface emissivity model is an integral part of the National Centers for Environmental Prediction (NCEP) Global Data Assimilation System. It is an anchor point for the variational bias correction scheme and crucial for the surface and near-surface infrared radiance assimilation. Identifying, quantifying and reducing the systematic errors in the ocean surface emissivity model will improve the current assimilation system and the future coupled ocean-atmosphere assimilation system.

Specifically, we have identified and quantified two types of temperature-dependent errors which are not fully accounted for in the current Community Radiative Transfer Model (CRTM) IR Sea Surface Emissivity (IRSSE) model, 1) the spectral temperature dependence and 2) the scan angle dependence. The spectral temperature dependence is primarily within the 800–900 cm -1 window region, where previous studies found the actual emissivity to decrease at colder temperatures. The temperature dependence is also found to be exacerbated at large scan angles. Hyperspectral radiances from the Cross-track Infrared Sounder (CrIS) and Infrared Atmospheric Sounding Interferometer (IASI) instruments will be used to compute observation minus background (O − B) statistics for the range of scan angles and ocean surface temperatures to quantify temperature-dependent emissivity biases. We find significant, first-order temperature-dependent biases in global O − B statistics, indicating that ocean emissivity models (e.g., the CRTM and SARTA IRSSE) will need to include temperature dependence.