Monday, 28 August 2023: 2:00 PM
Great Lakes BC (Hyatt Regency Minneapolis)
Feng Nai, CIWRO,
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Phased-array radars are being evaluated as potential replacements for the WSR-88Ds, which are near the end of their expected service life. The Advanced Technology Demonstrator (ATD) has been developed and deployed as part of the research effort to show the ability for a phased-array radar to collect calibrated, dual-polarization weather data with rapid update rates. Since becoming operational in 2021, the ATD has collected data supporting both engineering and meteorological research. The ATD utilizes an overlapped-subarray architecture, which allows for digital beamforming on receive. Thus, the ATD can use adaptive beamforming to adjust the receiving antenna radiation pattern to improve the accuracy of estimated radar variables, especially in the presence of large reflectivity gradients. One well-known adaptive beamforming technique to minimize contamination from discrete targets is Capon beamforming. However, Capon beamforming is not suitable for weather observations because Capon adaptive beams are difficult to calibrate. The conventional reflectivity calibration process involves assumptions on the two-way antenna radiation pattern that Capon beamforming often violates. In this work, additional constraints to the Capon optimization problem are developed in order to shape the main lobe of the adaptive beam to allow for calibration while providing flexibility to adjust the sidelobes to reduce contamination. To evaluate the proposed technique, a subarray IQ data simulator was developed to take archived NEXRAD data as inputs and produce subarray IQ data as though collected by the ATD. These simulated data are used in the adaptive beamforming technique to form optimized beam patterns. Simulation results show that the proposed technique leads to calibrated radar data with reduced biases from sidelobe contamination. To reduce the computational complexity of this technique, a deep neural network is being trained to optimize adaptive beamforming coefficients that result in patterns that are similar to those obtained from the optimization solver at a speed acceptable for real-time operations.

