879 Real-Time Mobile Radar Hurricane Wind Retrievals during Landfall

Tuesday, 14 January 2020
Hall B (Boston Convention and Exhibition Center)
A. Addison Alford, Univ. of Oklahoma, Norman, OK; and M. I. Biggerstaff and G. D. Carrie

The mobile C-band University of Oklahoma Shared Mobile Atmospheric Research and Teaching (SMART) radars have sampled five landfalling tropical cyclones (TCs) during 2016-2018. Data were collected in coordination with nearby coastal WSR-88Ds to perform dual-Doppler wind retrievals to study dynamic processes during landfall. However, as was demonstrated with the Harvey (2017) data, the low-level hurricane wind fields can also be extrapolated to estimate near-surface winds affecting coastal and inland communities. In the past, those analyses required many months of time to quality control before processing to produce the dual-Doppler analyses.

To assist emergency managers and federal forecast partners, a technique to perform dual-Doppler analysis in near real-time has been developed. A Python-based framework integrates the use of the Python Atmospheric Radiation Measurement (ARM) Radar Toolkit (Py-ART) and a three-dimensional variational data assimilation (3DVAR) dual-Doppler analysis technique. A modified Py-ART package is used to perform quality control, Doppler velocity dealiasing, and interpolation to a Cartesian grid. The 3DVAR dual-Doppler technique is used to heavily weight Doppler velocity observations over mass continuity to retrieve estimates of the horizontal flow. To decrease time needed to perform all steps for each analysis, the dual-Doppler analysis step for an analysis at time t is run in parallel with the quality control, dealiasing, and interpolations steps for the next dual-Doppler analysis at t+Δt.

A detailed overview of the real-time technique will be given. A successful test of this technique was implemented during the landfall of Hurricane Florence (2018). Results from real-time versus research quality dual-Doppler analyses will be shown. In Florence, analyses were often available two to three hours after data collection. However, improvements to the technique now result in five to ten-minute latency. The improvements will be tested during the 2019 hurricane season, and real-time analyses from hurricane landfalls in 2019 (if any) will be shown.

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner