Gu, Boyan (2019-12). Evaluations and Improvements of the RRTMG and Fu-Liou Radiative Transfer Model Simulations of Clouds. Doctoral Dissertation. Thesis uri icon


  • Clouds play an important role in the Earth's climate system, especially affecting the radiation balance and hydrological cycle. However, the modeled ice cloud radiative effects (CRE) are quite uncertain, because longwave scattering is neglected and scattering properties of ice clouds are not well represented in most general circulation models (GCMs). This study focuses on radiative effects of clouds, and evaluates the performance of the currently widely used broadband radiative transfer models. Furthermore, improvements to these models in single column simulations and GCM applications are introduced in this study by using a similarity principle, state-of-the-art cloud models, and more accurate scattering schemes. We first analyze the strengths and weakness of the Rapid Radiative Transfer Model for GCM Applications (RRTMG) and the Fu-Liou Radiative Transfer Model (RTM) for radiative transfer in single column simulations against the rigorous LBLRTM-DISORT (a combination of Line-By-Line Radiative Transfer Model and Discrete Ordinate Radiative Transfer Model, also called LBLDIS) calculations. In total, 6 US Standard Atmosphere profiles and 42 atmospheric profiles from the Atmospheric and Environmental Research (AER) Company are used to evaluate the RRTMG and Fu-Liou RTM by LBLRTM-DISORT calculations in the longwave spectral region from 0 to 3250 cm^-1 . Ice cloud radiative effect simulations with both models are initialized using the ice cloud properties from MODIS (Moderate Resolution Imaging Spectrometer) Collection 6 products. Simulations of single layer cloud CRE produced by RRTMG and LBLDIS show that RRTMG, which neglects scattering, overestimates the top of atmosphere (TOA) CRE by about 0-13 W/m^2 for ice clouds and 0-4 W/m^2 for water clouds, and underestimates the surface CRE by about 0-3.5 W/m^2 for ice clouds and 0-4 W/m^2 for water clouds, depending on the cloud particle size and optical thickness. The most significant overestimation and underestimation occur when the particle effective radius is small (about 10- 20um) and the cloud optical thickness is intermediate (about 1-10). The overestimation and underestimation is reduced significantly when using the similarity principle scaling of cloud optical properties in the RRTMG. The Fu-Liou RTM, which considers scattering, overestimates the TOA CRE by about 0-12 W/m^2 for ice clouds and 0-15 W/m^2 for water clouds; while underestimating the surface CRE by about 0-3 W/m^2 for ice clouds and 0-14 W/m^2 for water clouds depending on the cloud particle size and optical thickness. We further improve the performance of the column RRTMG and GCM RRTMG by implanting the 2/4-stream approximation scheme into the default RRTMG. The modified RRTMG significantly improves the performance of TOA CRE simulations in a column model. In GCM simulations, the globally averaged surface downward longwave flux is underestimated about 0.5 W/m^2 ; and TOA upward longwave flux is overestimated about 0.8 W/m^2 due to neglecting scattering. Cloud multiple scattering not only influences the longwave radiation, but also modulates the modeled climate. The longwave cloud scattering tends to enhance precipitation over the regions where the precipitation rates are already large, and also nonlinearly enhances the ascending and descending branches of the Hadley Circulation.

publication date

  • December 2019