Doppler Weather Radar Clutter Time Domain And Frequency Domain Inhibition Study

The Doppler weather radar data are often contaminated by ground clutter. It has a significant effect on the accuracy of the base data estimates (reflectivity, velocity and spectrum width). Because all the products and algorithms are based on the base data, if the ground clutter is not at least partially removed, clutter-induced bias of base data not only brings into question on the reliability of data presented on the base products, but also has a detrimental effect on all ensuing processing. Therefore, filtering techniques that attempt to ameliorate these ground clutter signals are essential for high data quality in all Doppler weather radar systems.Firstly, this dissertation introduces the research background and significance of ground clutter suppression, analyzes the characteristics of the ground clutter and weather signals in the Doppler weather radars and simulates Doppler radar echo signals (It includes ground clutter, weather echo signals and the mixture of them). The simulated signals are used later to study the time and frequency domain ground clutter suppression.Secondly, this dissertation talks about the time domain filtering, gives the basic theory of time domain filtering and describes the design method of the usually used fifth-order elliptic infinite impulse response (IIR) ground clutter filter. In the time domain, the work focuses on the regression filter. From the orthogonal polynomials fit, this dissertation gives the basic theory of the regression filter for ground clutter suppression and the filtering process using a regression filter. Through the study of the computational complexity of regression filter, the orthogonal polynomials which make the regression filter has the minimum computational complexity are found. We analyze the frequency response characteristics of the regression filter and compare the computational complexity of regression filter with that of fifth-order elliptic IIR ground clutter filter. Using the simulated radar signals, we analyze the ground clutter suppression performance of regression filter. In practice, regression filter is a high-pass filter that removes frequency components at either side of zero Doppler velocity. This removes not only the ground clutter with low frequency but also the overlapped weather echo signals in the stop band. Fixing the spectrum width of the weather echo signals, we change theirs mean Doppler frequency and filter them using regression filter. After filtering, we analyze the attenuation of them. Using an actual collected radar signal, we change the orders of the regression filter from first-order to tenth-order and obtain the corresponding ground clutter suppression ratios.Thirdly, this dissertation talks about the frequency domain filtering. We give the basic theory of frequency filtering. In the frequency domain, the work focuses on the adaptive Gaussian frequency filter. In the study of adaptive Gaussian frequency filter, we introduce its characteristics and application assumptions. Based on the simulated Doppler weather radar signals, we describe the filtering process of the adaptive Gaussian frequency filter. Using the actual collected radar echo signals, we analyze the ground clutter suppression performance of the adaptive Gaussian frequency filter.In the end, this dissertation analyzes and compares the ground clutter suppression performance of the time and frequency domain filtering. In the three filtering algorithms, the ground clutter suppression performance of the adaptive Gaussian frequency filter is better than that of the two others, and the ground clutter suppression performance of regression filter is better than that of the fifth-order elliptic IIR filter. However, the computation required to perform the adaptive Gaussian frequency filter is substantially greater than that required of the regression filter and fifth-order elliptic IIR filter because of the use of DFT's, the optional rank noise calculation and the fact that many times the analysis needs to be recalculated to select the optimal window. The time domain filtering algorithms require small storage and computation and can be performed in real-time. When the resources and processing velocity are limited, time domain filtering is a good choice.