Development of a Polarimetric Radar Forward Operator for the Bayesian Observationally Constrained Statistical-physical Scheme (BOSS)

 
Poster PDF

Authors

Matthew Kumjian — Pennsylvania State University
Charlotte Martinkus — Penn State University
Olivier P. Prat — North Carolina Institute for Climate Studies
Marcus van Lier-Walqui — Columbia University
Hugh Clifton Morrison — University Corporation for Atmospheric Research

Category

Microphysics (cloud, aerosol and/or precipitation)

Description

To explore the relationship between radar-observed quantities and microphysics, accurate models of the observed features are essential. However, current microphysics schemes are ill-suited for this task because they feature errors caused by structural assumptions and poorly constrained parameter values. We present an alternative approach wherein observations provide a fundamental physical and statistical underpinning in the construction of a novel microphysical parameterization scheme: The Bayesian Observationally constrained Statistical-physical Scheme (BOSS), which facilitates the use of polarimetric radar observations for development and constraint of a parameterization of warm rain microphysics. Typical forward operators use model-prognosed variables to determine a drop size distribution (DSD) that is subsequently discretized; scattering calculations are applied to these individual particles. However, BOSS predicts DSD moments and does not assume an underlying DSD shape, necessitating a moment-based forward operator. We report on progress towards developing this operator. To our knowledge, such an operator is unique, as it eliminates structural errors in the microphysics model’s DSD functional form and provides a natural way to estimate forward operator uncertainties. In order to estimate how DSD moments relate to observable polarimetric radar variables, we analyze a number of “realistic” rain DSDs by employing a tripartite approach, using (i) analytic DSDs, (ii) 1D bin microphysical simulations, and (iii) surface disdrometer measurements. This allows us to explore answers to some key scientific questions: What is the relationship seen in nature between DSD moments and the radar variables? What does the uncertainty in the moment-radar relationships suggest about which DSD moments should be prognosed in the model? Our approach can be extended for use with other advanced radar techniques (e.g., multiple frequencies, Doppler spectral signatures, etc.) routinely available from DOE ARM sites.