Fast Radiative Transfer Model for All-Weather Radiance Assimilation Developed for Joint Center for Satellite Data Assimilation (JCSDA)

Contact: Al Gasiewski

Improvements in satellite radiance assimilation techniques for driving numerical weather prediction (NWP) models have been the basis for steady improvements in NOAA forecast skill for many years. In data sparse regions of the globe, (for example, the southern hemisphere) satellite radiances provide virtually the only source of information for NWP purposes. However, most of the satellite data that is currently assimilated is from either infrared or microwave sensors observing over clear or nearly-clear regions. As a result, assimilation over the most economically significant weather conditions, such as frontal zones and hurricane rainbands, has been stymied due to the presence of clouds and precipitation. A major difficulty in being able to assimilate satellite data under such conditions is the current inability to rapidly compute the tangent linear observation operator in the presence of scattering clouds and precipitation.

Drawing upon their extensive analytical capabilities ETL's Microwave Systems Development Division has recently addressed this need by developing a practical (fast) tangent linear forward radiative transfer model applicable for arbitrary wavelengths. A key feature of the model is its ability to accommodate the effects of all scattering and absorbing hydrometeors, including liquid clouds, rain, ice clouds, graupel, and snow. Previous tangent linear models addressed only the absorbing component of clouds but could not properly accommodate the effects of scattering - which requires inversion of poorly-conditioned matrices. The new technique is based around the use of a discrete ordinate tangent linear radiative transfer (DOTLRT) model but with a unique symmetrization of the radiative transfer equation performed within each atmospheric layer. The symmetrization permits analytic factorization of key matrices that would otherwise lead to numerical truncation errors and overall solution instability.

The DOTLRT v1.0 model has been extensively tested at ETL and was delivered to the Joint Center for Satellite Data Assimilation (JCSDA) for incorporation in their Community Radiative Transfer Model. Subsequent incorporation of the DOTLRT model into an NWP assimilation scheme using NOAA satellite microwave data under all weather conditions is ongoing. The assimilation of microwave data over heavy clouds and precipitation is widely expected to improve forecast model accuracy within and in the vicinity of severe precipitating weather by reducing model error in latent heating and hermodynamic structure within cloudy regions.

More Information

• Microwave Systems Development Division