【摘要】 在环境研究领域,连续观测的数据(时间序列)几乎全都是具有多尺度性质的测量信号.观测数据之所以具有多尺度性质,一方面是因为其自身存在多尺度性质,另一方面往往受到外界各种因素的影响,而这些影响因子又存在不同的时间和空间尺度.为了解析生态环境过程的动态变化及其规律性,往往对连续观测数据首先进行滤波去"噪"(如干扰、误差、错误等),保留真实、客观、主要的信号.为此,本文利用小波分析的方法构造了带通滤波器,并用其分析得出了水环境中连续观测数据的变化规律.小波分析是时间——频率分析领域近年来迅速发展的一种新技术,具有多时间尺度,多层次和多分辨的特性,已被广泛地应用在信号分析、信息分析和地球科学研究上.本文首先对常见的几种小波基的尺度函数和小波函数作了功率谱分析,发现dmey(discrete Meyer wavelet)小波的高通和低通较其他小波函数更近似理想滤波器;为此利用dmey小波设置了带通滤波器,并以NINO3指数为例,检验了此滤波器,研究表明此滤波器有着较好的带通滤波效果.进而,用此滤波器对太湖现场实测的水深和降水数据进行了滤波,分析发现太湖水深变化在周以上尺度主要受降水影响;此外,太湖水深变化还存在1d和0.5d的显著周期变化.更多还原

【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.更多还原