A weather signal detection algorithm based on EVD in elevation for airborne weather radar

Interest is growing in the application of the spatial location information from weather scenario to the monitoring of meteorological signal for airborne weather radar system. Here, a new algorithm utilizes the differences between ground clutter and meteorological signal in terms of the spatial location in elevation dimension to achieve weather signal detection (WSD). Specifically, weather signal submerged in ground clutter background is detected over a so-called diagram of the second eigenvalue to improve weather observation. Generally, this diagram is acquired by implementing eigenvalue decomposition (EVD) operation on raw data which is collected via dual-channel in elevation. For WSD, the second eigenvalue in the diagram is employed as the test statistic. And this study investigates the statistics of the second eigenvalue in the case that weather signal component, ground clutter component and Gaussian noise are all complex Gaussian distributed and they are statistically independent. Based on the statistics, a constant false-alarm rate (CFAR) detector is also designed and then the second eigenvalue of the cell under test undergoes the CFAR detector to screen out the pixels containing meteorological signal component. Simulations, as well as experimental results, are presented to demonstrate the theoretical analysis and to evaluate the detection performance of the proposed EVD-based algorithm. As compared to most of the current WSD approaches, the presented EVD-based method really shows increased detection capability and great robustness. (C) 2021 Elsevier Inc. All rights reserved.

Automated summary

A new algorithm is proposed to detect weather signals in airborne weather radar systems, utilizing spatial location information and eigenvalue decomposition. By analyzing the statistics of the second eigenvalue in complex Gaussian distributed components, a CFAR detector is designed to improve detection capability. Experimental results demonstrate increased detection performance and robustness compared to current approaches.