The WSR-88D Inanimate Hydrometeor Class

Abstract

The dual-polarization upgrade to the WSR-88D network of weather radars included the addition of a hydrometeor classification algorithm (HCA) to the Open Radar Product Generator (ORPG). The HCA product categorizes each Level-III radar range–azimuth cell into one of 10 classifications or marks it as “unknown.” However, not all target types fall under the 10 classification options. For this reason, multiple studies have examined adding new classes to the operational HCA. In this study, a new “inanimate” class is developed for additional hydrometeor classification in the ORPG and is described in detail. This class encompasses nonweather phenomena such as chaff (the primary motivation of this work), sea clutter, and some types of combustion debris and radio frequency interference. The design process is detailed, including human truthing, data selection, and optimization with a genetic algorithm. Multiple case examples are presented and analyzed, both qualitatively and quantitatively. Quantification is put into perspective with previous studies for a better understanding of the algorithm’s performance and impact. A discussion of applications, including subclassing, chaff detection, and implementation as an “aviation” classification algorithm in the ORPG, is presented.

Polarimetric Observations of Chaff Using the WSR-88D Network

“…It is not entirely clear why this is the case, although it is postulated that the distribution of VCP 31 cases may be limited to sites that do not have large, widespread chaff dispersions. As noted in [5] and observed repeatedly in the present study, chaff that does not spread out significantly in the horizontal dimension tends to have much lower Z H values and, interestingly, has issues being picked up by the Inanimate class [22] and the chaff detection algorithm [17]. This is somewhat counterintuitive because it might be expected that less dispersion would result in higher concentration (and hence higher Z H values).…”

“…The histogram of W was not included in [5], but is shown here due to its importance in the work from [22], which utilized statistics from [5] to develop a new hydrometeor class (called "Inanimate") for the WSR-88D hydrometeor classification algorithm (HCA; [23]). It was found in [22] that exceptionally low values of W were present in all of the target types encompassed by the Inanimate class, which include chaff, sea clutter, combustion debris, and radio frequency interference. However, in recent attempts to separate chaff from sea clutter, there are distinct differences in the peak and tail of W , making these findings relevant at the current time.…”

Estimating Fire Radiative Power Using Weather Radar Products for Wildfires

“…While this is an elevated feature, since the fire was intense during this period, it coupled vertically to the pyrometeor plume, significantly enhancing the integrated radar products. Thus, future work should integrate chaff classification algorithms (e.g., Kurdzo et al., 2017; Yu et al., 2016) to assess if these signals can be automatically removed.…”

A Deep Learning–Based Velocity Dealiasing Algorithm Derived from the WSR-88D Open Radar Product Generator

“…Currently, only one operational ORPG algorithm utilizes modern machine learning techniques (the chaff detection algorithm; Kurdzo et al 2017a), but due to the wide accessibility to years of archived data, the resources exist to enable other ML-based algorithms (e.g., Jatau et al 2021) truth at all times if we are attempting to emulate/recreate the algorithm, eliminating the need for any manual dealiasing.…”