Higher order spectral analysis of weather signals

Abstract

Higher Order Spectral Analysis (HOSA) is very useful technic for analysing the statistical relationships between spectral components and their distribution. HOSA can be defined in terms of bispectrum as a third order spectrum. One of the main reasons to consider HOSA on weather signals is that to characterise the Doppler spectrum obtained through the backscattered echoes which are generally skewed and having multiple peaks. The analysis also brings out any deviations from gaussianity and the performance under noisy environment. This will help in identifying and detecting the signals and their by better estimation of moments for further processing The bispectrum estimation technique has been applied to a complex time series data derived from a Polarimetric X-band Doppler Weather Radar (DWR). The results of this analysis are compared with the conventional techniques like Fourier method and pulse pair technique. It is observed that through bispectrum approach the gaussian noise components are significantly removed from the back scattered signals and improves the detectability of the weather signals under noisy condition.