4.3
Use of cloudy radiance observations in mesoscale data assimilation
Tomi Vukicevic, Colorado State Univ., Ft. Collins, CO; and T. Greenwald, R. Hertenstein, and M. Ghemires
Mesoscale atmospheric data analysis is challenging in many respects. These challenges result from: 1) insufficient temporal and spatial frequency of observations, 2) inaccurate direct observations of moisture and 3D flow and 3) complex relationships between observed quantities and those we wish to analyze. To address and hopefully overcome these problems it is necessary to develop methods to utilize observations with either high temporal or spatial resolution, or both, and with small measurement errors. The remote sensing observations have high potential to satisfy these requirements. Important consequence of this choice of observation type is that the problems in category (3) are then emphasized. They could be addressed by means of using mesoscale numerical forecast models in the data analysis. The four dimensional variational (4DVAR) data assimilation method provides framework for generating atmospheric analysis from indirect observations and the complex mesoscale models.
We investigate the implementation of 4DVAR data assimilation method in mesoscale data analysis using satellite radiance observations. We use the GOES regional radiance observations and the Regional Atmospheric Modeling System (RAMS) to study impact of the high frequency radiance observations on the mesoscale data analysis of moisture for a case of continental stratocumulus evolution. We first developed forward model to simulate time evolving radiance data using RAMS integrations of dynamical, thermodynamical and cloud microphysical quantities for selected locations. This involves use of radiative transfer (RT) models controlled by the mesoscale model solutions. The preliminary results show relatively large discrepancy between the modeled radiance and observations. We also developed an adjoint model associated with the RAMS to analyze model solution dependencies on the initial conditions and to perform optimization of the initial condition in the 4DVAR data assimilation of the GOES observations. The data assimilation of the GOES radiance data will be performed after adjoints of the RT models are also developed.
We wish to present the comparison of modeled and observed radiance data and the adjoint analysis of mesoscale model cloud simulations. The results demonstrate the scope of modeling effort involved in the mesoscale data assimilation, provide insight into physical links between the cloud evolution and initialization as represented in the mesoscale model simulation and illustrate computational challenge of the mesoscale data assimilation.
Session 4, Assimilation
Tuesday, 16 January 2001, 2:15 PM-5:45 PM
Previous paper Next paper