Extended Abstract (152K)1.9
Atmospheric Radiative Transfer Adjoint Models for the Regional Atmospheric Modeling and Data Assimilation System (RAMDAS)
Thomas J. Greenwald, CIRA/Colorado State Univ., Fort Collins, CO; and T. Vukicevic
An important aspect of evaluating a 4DVAR system that assimilates satellite radiance data is investigating the characteristics of the adjoints of the atmospheric radiative transfer (RT) models. This study examines the properties of the forward linearized versions (i.e., tangent linear) and adjoints of the RT models and the conditions under which these linear approximations are valid for the new Regional Atmospheric Modeling and Data Assimilation System (RAMDAS). This system uses the Regional Atmospheric Modeling System (RAMS), a mesoscale model that explicitly predicts clouds and precipitation species. RAMDAS is unique in that it will have the ability to incorporate both clear and cloudy satellite radiance data.
As defined here, the RT models include the RT solver, cloud optical property models, and gas extinction model. At visible wavelengths we apply the Spherical Harmonics Discrete Ordinate Method (SHDOM) to compute radiances and at infrared wavelengths we use an Eddington 2-stream model. Cloud extinction and single-scatter albedo are derived from modified Anomalous Diffraction Theory (MADT). A simple parameterization is used to estimate the asymmetry parameter, which is the third required cloud optical property. The Optical Path TRANsmittance (OPTRAN) approach is used to compute gas extinction. These models are applied to all five Geostationary Operational Environmental Satellite (GOES) imager channels. We plan to use GOES data in our future 4DVAR experiments with RAMDAS.
Early results show that the SHDOM tangent linear model for the GOES imager visible channel is generally valid, depending on the cloud conditions, for perturbations in cloud extinction of 15-25% (roughly corresponding to 25-45% perturbation in liquid water mixing ratio). The adjoint models of the MADT model and asymmetry parameter parameterization also show for the visible channel that the sensitivity of cloud optical properties to changes in cloud liquid water mixing ratio generally increases with decreasing mixing ratio. Complete results will be presented at the conference.
Session 1, Effective Assimilation of the Vast Observational Datasets Becoming Available
Monday, 14 January 2002, 9:30 AM-2:44 PM
Previous paper Next paper