Real-time Airborne Radar Data Quality Control and transmission from NOAA Aircraft for assimilation into HWRF
Airborne Doppler radar has been on the NOAA WP-3D aircraft since 1981, and the first tropical cyclone sampled was Hurricane Debby in 1982. Since then, the Doppler radar system has been updated and more fully integrated with the radar system aboard the WP-3D. Airborne Doppler radar was also installed on the NOAA G-IV aircraft, and the first tropical cyclone observed by the radar was Hurricane Katia of 2011. These radars have been a powerful research tool for decades, and since 2005 data and analyses have been sent from the aircraft in real time. Doppler analyses were first received at the National Hurricane Center (NHC) during Hurricane Wilma in 2005. Doppler superobs were first assimilated in a real-time research model in 2008 during Tropical Storm Fay. HWRF (Hurricane Weather Research and Forecasting Model) assimilation of Doppler radial velocities first occurred in a real-time parallel run during Hurricane Tomas of 2010, and since 2013, they have been assimilated in the operational runs of HWRF. At the Hurricane Research Division (HRD) and Aircraft Operations Center (AOC) we have been working to make the process as automated as possible.
Providing analyses and Doppler radial velocities in real time requires an automated analysis and quality control system. The original system was developed under funding from the Joint Hurricane Testbed. To use the Doppler velocities, a great deal of quality control is needed, including removal of side-lobe noise, removal of sea-surface reflection, and de-aliasing. Several passes through the Doppler data are required to be able to quality control the radial velocities correctly.
In the first pass, continuity along the Doppler radial from the aircraft to the target is required for every observation that is accepted. The Doppler velocities are unfolded using the Bargen-Brown method, which de-aliases outward from the aircraft using the wind at the aircraft for a starting guess. A three-dimensional least-squares wind analysis is produced that has no azimuthal variation of radial, tangential and vertical winds.
The second pass uses the winds from the first pass to re-initialize the Bargen-Brown seed at each radial gap, allowing much more data to be included in the second pass. A new cylindrical-coordinate analysis is made that contains only the wavenumber 0 and 1 Fourier components of radial, tangential, and vertical wind.
A least-squares Cartesian wind analysis is produced in the third pass, using the wind field from the second pass to reseed the Bargen-Brown de-aliasing, wherever there is a data gap along the Doppler radial. The quality-controlled Doppler radial velocities are sent to NCEP (National Centers for Environmental Prediction) Central Operations (NCO) for assimilation into HWRF. They are also used to produce superobs for HRD and its research partners. The wind analyses, the superobs, and the quality-controlled Doppler radials are all transmitted off the aircraft in real-time during a tropical-cyclone mission. In the third pass of the data, vertical profiles of radar reflectivity, radial velocity, total wind speed, and vertical wind are also produced and transmitted. The wind-retrieval method for the profiles does not require mass continuity and thus the analyses have much better radial (from storm center) and vertical resolution.
Every measure that can be taken to make the process as automated as possible is required. Once the quality-control process is started, it is fully automated to completion. The procedure does require human inputs, however. A java application was developed this year at HRD to permit entry of the required information, such as storm center and motion, beginning and end time of input data, the radial directions of flight tracks from the storm center, the resolution of the analysis, and the stringency of the quality control (based upon the gradients of wind velocity, the requirements are more stringent in hurricanes, and even more stringent in major hurricanes). Since it is a java application it can be run from numerous platforms, allowing scientists on the ground to supply information from their workplace or home to the computers on the aircraft to run the proper analysis.
AOC has developed the procedures for most efficiently getting these inputs to the onboard workstation, and for sending completed analyses and quality-controlled velocities to the ground and the appropriate agencies. Remote Sensing Solutions wrote the original code for transmission of the data to NCO from the aircraft. Within the last few days before this abstract submission, AOC has developed a procedure where, once the system is started on the aircraft, a person on the ground can submit a job file, the transmission software aboard the aircraft detects a new jobfile, the quality-control/analysis process starts, and once it is finished the appropriate files are transmitted. This is a major step in making this process fully operational.