Toward Reducing Cloudy-sky CRTM Biases Against Satellite Observations at High-frequency Microwave Channels

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Thursday, 6 February 2014: 4:15 PM
Room C111 (The Georgia World Congress Center )
Jie Gong, USRA, Greenbelt, MD; and D. L. Wu

Assimilation of satellite radiance requires accurate and fast radiative transfer models (RTMs) to link between satellite observations and model physical variables. JCSDA CRTM is a centerpiece of many operational data assimilation (DA) systems to fulfill such a task. Although successful under clear-sky scenes, it remains challenging and inefficient to produce compute cloudy-sky radiances for high-frequency (150-183 GHz) channels, showing large biases for optically thick/precipitating cloudy scenes. To reduce these biases, we developed a fast scattering model to replace the current look-up-table (LUT) in CRTM at 157, 183±3 and 190 GHz for ice cloud cases. Computationally efficient, this cloudy-sky model is constrained by CloudSat cloud ice and Microwave Humidity Sounder (MHS) radiance measurements. We will show the performance of this fast scattering model and discuss the plan to incorporate it into CRTM.

Based on more than 5 years of collocated and coincident NOAA-18 MHS-CloudSat measurements, this fast model contains a set of empirical coefficients, or LUT, to describe the relationship between MHS cloud-induced radiance depression (Tcir) and CloudSat ice water path (IWP) at 157, 183±3 and 190 GHz. In the tropics, the Tcir-IWP relationships are also dependent slightly on cloud top height (CTH), while at the mid-latitude, the tropospheric temperature lapse rate plays a key role in modulating the variability of the relationships. A set of LUT with analytical solutions of the Jacobian matrices was then built-up. The retrieved IWP from MHS was further validated against CloudSat observations, which shows broad agreement between 300 and 104 g/m2.

We further compare the cloudy-sky biases of CRTM and the fast scattering model using MERRA reanalysis and a higher-resolution analysis (National Climate Assessment Run, or NCA run) as the input. We will show that the current CRTM produces Tcir-IWP curves that are too shallow at these microwave channels compared with the observation, unable to close the radiance bias with reasonable IWP in optically thick/precipitating scenes during DA processes. The fast scattering model is able to greatly reduce the biases at these scenes. Together with CRTM “clear-sky” module, it runs 3X faster than the current CRTM “cloud scattering” mode. By swapping the current LUTs in CRTM with those generated from this study, the improved CRTM is expected to provide a critical step toward accurately representing cloudy radiances, enabling effective assimilation of all-sky radiances from NASA and NOAA future cloud/precipitation missions using high-frequency microwave.