Matrix Decomposition-Based Spectral Moments Estimators For Airborne Weather Radar Echoes

Meteorological radar is an important means of meteorological detection.Airborne meteorological radar(WXR)is used to detect severe meteorological areas such as thunderstorms,hail,turbulence,and wind shear on the airway ahead of the aircraft.Its performance is directly related to the comfort and safety of flight.In many dangerous weather conditions that affect flight safety,such as clear air turbulence and dry low-level windshear,which are not accompanied by visible weather phenomena,resulting in a low radar echo signal-to-noise ratio,and conventional airborne weather radars have not been able to effectively detect them.There is a greater threat to flight safety.The essence of airborne meteorological radar target detection is the estimation of echo spectral moment.The type of meteorological target and the spectral moment parameters(average Doppler frequency and Doppler spectral width)of radar echo are closely related.The research on airborne weather radar echo spectral moment estimation method is of great significance for improving the accuracy and effectiveness of detecting meteorological targets.This paper aims at the problems of large airborne meteorological radar echo spectral moment estimation methods under low signal-to-noise ratios,including large errors,large calculations,and inaccurate estimation when the data is abnormal.Construct a parametric model of the radar echo covariance matrix and use matrix decomposition to propose two methods for estimating the spectral moment of airborne radar echoes in different backgrounds.The airborne weather radar echo spectral moment estimation method under simulation and simulation experiments prove that the proposed method can effectively estimate the radar echo spectral moment under their respective applicable conditions.The main work is summarized as follows:Firstly,a fast estimation method of radar echo spectral moment based on DFT is proposed,which can be used in the case of low signal-to-noise ratio.By analyzing the characteristics of the covariance matrix of the radar echo,using matrix decompose and transform it to obtain a closed covariance matrix,the model can still maintain better when t he spectrum width value is larger,and the use of its Vandermonde-structured characteristics to obtain a FFT-based search function without search cost function,does not require frequency division when estimating the cost function,and uses all data for ca lculation,which reduces the amount of calculation and improves the estimation performance.Secondly,a parameterized spectral moment estimation method based on L1-PCA echo is proposed,which can be used in the case where the radar received echo data is a bnormal and there are many outliers.By decomposing the parametric model of radar echo covariance matrix when data is abnormal,using the orthogonality of signal subspace and noise subspace and the stability of 1 norm to construct the cost function,and then estimate the radar echo spectral moments.This method can maintain good estimation performance when the radar echo data is abnormal and there are many outliers.