A New Methodology for Separating Cloud and Drizzle Signatures from Radar Observations

Description

The separation of radar signatures depicting cloud and drizzle within a pulse radar volume is a long standing problem and also very important since the separation enhances our understanding of the microphysical and dynamical processes introduced by turbulence and leads to the manifestation of other meteorological observables . In the case when only clouds are present, the Doppler spectrum is well approximated by a Gaussian model. When clouds and drizzle are coexisting in the same radar volume, the signal spectrum is usually bimodal or asymmetrical. In this research, a new methodology to detect, estimate and separate cloud and drizzle echoes from vertically pointing Doppler spectra data is introduced. First, a parametric model in the time domain is developed using the recorded radar Doppler spectra data to retrieve the signal spectral moments. Goodness of fit parameters are defined that specify features of the Doppler spectrum. If the detection parameters exceed pre-determined thresholds, the signal contains a mixture of cloud and drizzle. It is shown that these new detection parameters are robust to noise contamination and variations of the signal spectrum width. As a result, the proposed method can be applied to signals with shorter integration time; this significantly reduces the effect of small-scale fluctuation present in the Doppler spectrum. Next, the drizzle map is processed to reduce uncertainty due to turbulence and to accommodate the location of the cloud base. At the locations where cloud and drizzle co-exist, the model is modified to include drizzle and cloud spectral parameters. Finally, a similarity-based classifier is implemented to specify which echoes are associated with cloud or drizzle. The proposed method is demonstrated using data collected by the W-band ARM cloud radar (WACR) on Graciosa Island in the Azores. The results show the efficacy of the method in separating cloud from drizzle.