Assessment of Radar-Derived Tropical Cyclone Structure using Different Quality Control Methods

A novel airborne Doppler radar dataset, The Tropical Cyclone Radar Archive of Doppler Analyses with Re-centering (TC-RADAR), made publicly available by the National Oceanographic and Atmospheric Administration’s Atlantic Oceanographic and Meteorological Laboratory, is increasingly being used by the research community for investigations of TC structure and intensity change. Recent multi-case composite studies using TC-RADAR have suggested that deficiencies in low-level data coverage in particular may be responsible for discrepancies in structure with studies employing other data sources. The quality control (QC) method applied to the dataset, NOAA-QC, has been shown to accurately identify non-meteorological data, but is aggressive at removing meteorological data (MD) especially in the lower and upper troposphere. The purpose of this study is to assess the benefits of reprocessing TC-RADAR with a less aggressive QC method. We first employ a dataset comprised of a few human-edited cases (with maximum MD retention) to create a baseline for comparison with NOAA-QC. The impact of increased MD retention at lower and upper levels is assessed through structural diagnostics (e.g., azimuthal-mean flow) derived from the radar wind analyses. In addition, we employ a recently-developed machine-learning quality control method, which has been shown to retain more MD than NOAA-QC. Although the machine-learning method retains less MD than human editing, its automation facilitates construction of a larger database of cases suitable for compositing. A subset of TC-RADAR cases is reprocessed here using the machine-learning QC and structural composites are compared with those of recent studies (using NOAA-QC) to highlight the advantages of a better QC for future research applications.