Experiments with three-dimensional radar reflectivity data assimilation into the COAMPS model

High temporal and spatial resolution of radar measurements enable to continuously observe dynamically evolving meteorological phenomena. Such set of a radar reflectivity data assimilated into the numerical weather prediction model has the potential to improve initial description of the atmospheric model state. Major drawback for the assimilation of radar data is the strong non-linearity of the observation operators and the intermittent nature of the precipitation processes. These result in a violation of the assumption of Gaussian error characteristics in the data assimilation schemes, which manifests itself in unrealistic background error covariance matrices and in unstable solutions.

The paper is concentrated on the development of radar reflectivity assimilation technique into COAMPS mesoscale model using an Ensemble Kalman Filter (EnKF) type assimilation schemes available in Data Assimilation Research Testbed (DART) programming environment. To obtain a well-conditioned background error covariance matrix we introduced concept of Xue to simulate the errors of the equivalent radar reflectivity factor Ze (with units of mm6 m-3) which error probability density function (PDF) approaches Gaussian when there are sufficient independent samples. The reflectivity in dBZ value, i.e. a logarithmic function of Ze, does not have the above error statistics. Proposed error model is more suitable for representing typical radar sampling errors.

Before weather radar data enter into the assimilation system, the measurement errors are eliminated through quality control procedures. Artifacts associated with non-meteorological errors caused by sun or external antenna interfering signals etc. are removed using the algorithms based on a pattern analysis of 3-D reflectivity fields. Then procedures for correction of the reflectivity data are required, especially due to radar beam blockage and attenuation in rain.

After the quality control procedure the quality indices for each radar pixel were estimated. Quality index (QI) is defined as unitless metric that quantitatively expresses quality of the data after corrections basing on analysis of selected quality factors. The factors are determined from radar technical parameters, radar beam geometry, burdening by non-meteorological echoes, and other disturbances in radar observations. The radar pixels successfully verified by means of the QI algorithm were employed in the assimilation.

The proposed method has been applied to simulate intense precipitation events during development of mesoscale convective systems (MCS) in Poland in June 2009, May 2010, and August 2010. Results from one selected case will be presented in details.