Comparison of HRD Airborne-Doppler radar quality control and analysis to independent data sources

Since 2004, the Hurricane Research Division has been using an in-house software package to quality control and analyze, in real time, the airborne Doppler radar data collected aboard the NOAA WP-3D aircraft. The package consists of 1) an automatic quality control that automatically de-aliases the Doppler-radial velocities and removes the sea-surface reflection and other non-precipitation artifacts, and 2) three-dimensional and two-dimensional variational wind analysis techniques to display the data. Work on the three-dimensional variational analysis package began in 1994, and results from it were first reported in 1994 at the TOGA-COARE workshop in Toulouse, France. Further descriptions of the analysis techniques have been made by Gamache (1997) and Reasor et. al (2009). The original development of the automatic quality control, and its marrying to the variational analysis techniques, was developed with funding from 2003-2005 by the Joint Hurricane Testbed, and a final report was made to the JHT in 2005 (Gamache 2005).

In that 2005 report, Gamache did a quick, rough comparison of the analyses with dropsonde observations that fell within the relatively small inner-core three-dimensional wind analysis. No attempt was made to filter the dropsonde observations to make them represent a similar depth to the 3D Doppler analyses, and the differences were not broken down with respect to height. The results showed very good agreements in the mean between the winds from analyses and Doppler observations except in the comparisons confined to below 1 km, where the radial inflow diagnosed by the sondes was much greater.

Meanwhile, since 2008, an ensemble Kalman filter (EnKF) is also being developed at the HRD for HRD's experimental version of the Hurricane Weather Research and Forecasting (HWRF-X) model to assimilate inner-core observations (i.e., flight-level, dropsonde, Doppler-radial velocities, and SFMR surface wind speed) and obtain high-resolution vortex-scale analyses.  Presumably, the EnKF will provide near-optimal analyses for the hurricane core as it will utilize all available flight data and take advantage of flow-dependent model covariance information valid at analysis time.  Such analyses, along with the built-in observation forward operators, offer an excellent medium of fair comparison between the above-described observation-space diagnostics and equivalent model-space diagnostics, and make it possible to evaluate the model representation of hurricane structure at the same spatial and temporal characteristics as the observed data.

Between now and 15 April, further, more detailed, comparisons will be made between the analyses and the observations, as well as between independent observations and Doppler radial measurements. The dropsonde or flight-level observations will be filtered to help them represent similar volumes as the three-dimensional winds fields, and the differences between these observations and the Doppler data and analyses will be examined as a function of height, as well as proximity to the radius of maximum winds.