乌鲁木齐10号泉流量变化的多尺度小波分析与异常识别

结合10号泉开始观测以来的震例，利用基于小波分析的多尺度分析原理对乌鲁木齐10泉流量的日均值观测序列进行了分析，总结了10号泉流量多尺度分析结果的映震特征。10号泉流量的地震前兆异常绝大多数是高值异常且具有较好的重复性，即一般都是在高值异常结束后几个月之内发生地震，个别情况下地震会发生在高值异常持续时段内。初步可以认定，利用基于小波变换的多尺度分析原理将流量数据分解成若干频段后，能够较好地识别和提取流量原始测值中不易分辨的震前异常。

MULTI-SCALE WAVELET ANALYSIS AND ANOMALY IDENTIFICATIN OF OUTFLOW VARIANCE IN NO.10 SPRING OF URUMQI

The multi-scale analysis method based on BWT was used to analyse the daily mean of outflow of observational data in No. 10 spring of Urumqi, combining with some cases since No. 10 spring started to be observed. And we summarized the characteristics of reflecting earthquakes of multi-scale analysis results about the outflow of No. 10 spring. Most anomalies of seismic prognostic in No. 10 spring are some anomalies with high value, and have some relatively good repeatability. That is, the earthquake often occurred within several months after anomalies with high value ended. However, individual earthquakes occurred in the process of high value anomaly. According to the preliminary analysis, we thought, after we separated the data of outflow into several parts with different frequency, the multi-scale analysis method based on BWT is able to identify and extract very well some anomalies of not easily to identify in the raw data of outflow before the earthquake.