The Microphysics and Kinematics of GPM's Satellite Radar Profiles

In this study we use polarimetric, Doppler radars to get an x-ray like view into the reflectivity profiles observed with a spaceborne precipitation radar. The core satellite of NASA’s Global Precipitation Measurement (GPM) Mission consists of a Dual-frequency Precipitation Radar (DPR) that detects light rain and falling snow and retrieves the raindrop size distribution to provide a more accurate depiction of global precipitation. To validate these retrievals, GPM leverages ground-based, polarimetric, Doppler radar networks and geometrically matches them with the DPR rays in the framework of the GPM Validation Network (VN).

This study applies a K-means cluster analysis to the DPR reflectivity profiles collected over the VN’s 118 ground radar domains between 2016-2020, from which twelve precipitation regimes sampled by the DPR were identified. These clusters are characterized by several modes of stratiform and convective precipitation, including isolated cells as well as different regions of mesoscale convective systems. We use hydrometeor identification and vertical motion retrieved from ten dual-Doppler radar domains in the VN to examine the characteristic shape of each cluster of DPR reflectivity profiles and gain further insight about the the DPR algorithm (2ADPR) profile classifications of “stratiform,” “convective,” and “other.” A quick check confirms that stratiform clusters are characterized by weak vertical motion and convective clusters by much stronger vertical motion, especially above the melting level. The deepest convective cluster consists of the strongest vertical motions (>12 m/s) and greatest fraction of heavily rimed ice (>30%) in the precipitation column. However, this most kinematically intense cluster does not include profiles with the greatest reflectivity below the melting layer. Instead, its profiles have much larger volumes of graupel and hail distributed across a 5 km layer above the melting level with lesser amounts of rain below it. This result implies that the convection with the most kinetic energy may not produce the most intense precipitation at the Earth’s surface, which has implications for depicting convective precipitation in large scale models and precipitation conversion rates used in cloud models. Additionally, the dataset assembled in this study is ripe for machine learning applications to extract additional information from DPR observations and future missions that use satellite radars to study convection.