The Construction of a Wavelet Filter and Its Application in Environmental Research
Huang Juan,Qian Xin,Wang Cheng-Lin,Gu Liang,Qian Yu
2007-01-01
Abstract:In the environmental research field, almost all the continuously observed data (temporal sequence) are measured signs which have characteristics of multiple timescales and multiple layers. The reason why observed data have such characteristics is not only that these data have self-generated characteristics of multiple timescales and multiple layers but also that they are usually affected by all kinds of other possible causes, which have different characteristics of multiple timescales and multiple layers as well. In order to analyze and get the true temporal changes and patterns about these environmentally observed data, researchers often filter out noise (including interferences, errors, mistakes, etc) from them, and then save the real, true and primary signs. So an effective wavelet band-pass filter is constructed by using the wavelet analysis method in this paper and applied to analyze the continuously observed data in water environmental research field. Wavelet analysis is a new technique which has developed quickly in time-frequency analysis. Since it has characteristics of multiple timescales, multiple layers and multiple resolutions, wavelet analysis has been applied widely in signal analysis, information analysis and earth sciences, etc. In this paper, the power spectrums about several useful mother wavelets are analyzed at first. According to the analysis, the conclusion is that the wavelet filter constructed with dmey (Discrete Meyer Wavelet) wavelet is a better filter than others. Because its decomposed high-pass filter and decomposed low-pass filter are much more approximate to perfect filter. The wavelet band-pass filter is constructed by applying dmey wavelet. After using NINO3 index to test such filter, the good result is shown that this wavelet band-pass filter is very useful and effective. Furthermore, this filter is applied to filter the observed data of water depth and rainfall in Taihu Lake. Then, after analyzing the filtered data, it is shown that there are recurrent water depth changes in half day and one day timescale in Taihu lake. In addition, the rainfall impacts the water depth in more than a week time-scale.