A New Approach to Real-time Automated Processing of Hurricane Reconnaissance Radar Data

An automated processing method for airborne tail-Doppler radar observations has permitted near-real time transmission of quality-controlled wind data for assimilation into operational hurricane models over the past decade. It has also led to increased availability of three-dimensional wind and reflectivity analyses to forecasters and the general research community for assessment of tropical cyclone structure. Essential components of the automated method include quality control (QC) to remove non-meteorological noise, de-aliasing to recover Doppler velocities that fall outside of the Nyquist interval, and wind synthesis to estimate the flow state from overlapping fore- and aft-pointing Doppler radials. In this presentation, we discuss initial progress on the development of a new algorithm for automated processing that replaces a rules-based QC with machine-learning QC. The necessity of having to employ a low-wavenumber background wind field for de-aliasing is evaluated through extension of the Nyquist interval using dual-Pulse Repetition Frequency observations collected in Major Hurricane Lee (2023). Faster and more flexible synthesis methods are also discussed. The advantages of a new algorithm in the near term are retention of greater amounts of meteorological data and, potentially, processing that is not contingent on execution of an entire transect through a storm and a low-wavenumber analysis being done first. In the 2030 timeframe, as the WP-3D reconnaissance aircraft are replaced with C-130Js, a new Doppler-radar capability will be required. The new algorithm should be more adaptable to substantial changes in the airborne radar system used for hurricane reconnaissance, whether that be Airborne Phased Array Radar (APAR) or mechanically steered Doppler radar.