Gaussian model adaptive time domain filter (GMAT) for weather radars

This paper presents an adaptive time domain filter for ground clutter filtering and signal parameter estimation for dual-polarization capable weather radars. The auto-covariance function of radar signal can be expressed as a sum of auto-covariance functions of the clutter, precipitation and noise that follow Gaussian forms. The filter matrix is designed such as when it is applied to the time series data, clutter component in the signal will be transformed to noise (i.e. the auto-covariance matrix is diagonal). However, weather echoes overlap clutter are also suppressed. An interpolation procedure then is used to recover the transformed part of the weather. The proposed design overcomes limitations of current spectral processing method caused by finite length of the data. A unique filter can be designed to use for both H and V channels for dual-pol parameter estimation. This way ensures the correlation between the two channels and minimizes estimate errors. In addition, the filter can be directly extended for staggered PRT 2/3 sampling scheme. The filter performance analysis was done using simulated time series radar data and CSU-CHILL measurements.