Trial of Real Rainfall Detection by Digital Signal Processing from Turbulence of Electric Field Around Leaky Coaxial Cable

利用漏同轴电缆（Leaky Coaxial Cables， LCX）信号进行了室外暴雨探测实验。接收到的信号受高频、高强度噪声的污染，利用离散小波变换去除高频分量，提取暴雨中不连续点的低频分量。由于同时提取了由外部噪声引起的不连续点，我们建立了一种消除不必要的不连续点的算法。该算法（不必要不连续点排除算法）分为三个步骤；1)“用信号波动的阈值判断法检测大脉冲噪声”，2)“用Lipschitz-Holder指数检测连续噪声的边缘”，3)“用百分比值检验离群值检测小脉冲噪声”。将该算法应用于信号波动，结果表明，1 min降雨强度大于30mm/h的暴雨是可检测的。

Outdoor experiments were conducted to detect the torrential rain from the signal of LCX(Leaky Coaxial Cables). The received signal was contaminated by high-frequency and high-intensity noise, thus, we removed the high frequency components using discrete wavelet transform, and could extract the low frequency components of the discontinuous points due to the heavy rain. Because the discontinuous points originating in external noise were also extracted simultaneously, we built an algorism which eliminatesunnecessary discontinuous points. The algorithm(Unnecessary Discontinuous Point Exclusion Algorithm) consists of three steps ; 1)"Detection of the large impulsive noise by the threshold-judge of the signal-fluctuation", 2)"Detection of the edges of the continuous noise by the Lipschitz-Holder index", and 3)"Detection of the small impulsive noise by the test of outliers using percentile". As a result of applying the algorithm to the signal-fluctuation, it was shown that heavy rain with 1 min rainfall intensity greater than 30mm/h is detectable.