Mesoscale field experiments are one of essential procedures to advance our understanding of physical processes and its parameterization into numerical prediction model. We performed two majors field experiments (summer and winter experiments) to achieve this goal. In this presentation we will introduce these experiments and recent advance of retrieval of microphysical parameters and understanding of these processes.
Conventional retrieval of microphysical parameters from polarimetric radars, such as polynomial regressions, have been widely used but struggle to capture the inherent nonlinearity between drop size distributions (DSD) parameters and radar measurements. We developed a machine learning (ML) algorithm to retrieve DSD parameters from polarimetric radar variables within the framework of double-moment normalization. This normalization provides a stable and potentially invariant representation of DSDs, enabling consistent applicability of the retrieval algorithm across different precipitation regimes.
The developed algorithm was employed with X-band polarimetric radar observations to investigate the spatiotemporal evolution of DSDs within convective cells. A well-developed convective cell was intensively observed through a unique scanning strategy, yielding a Lagrangian perspective on microphysical evolution throughout the storm lifecycle. The retrieved DSD parameters revealed characteristic variation across the storm lifecycle, with condensational growth dominating the developing stage, collision?coalescence processes prevailing the mature stage, and the dissipating stage characterized by smaller mean diameters, weakened coalescence, and potential evaporation. These findings highlight deeper insights into the spatiotemporal variability of microphysical processes and its application into numerical models such as model verification and parameterization.
|