新阈值函数去噪方法在可降水量预报中的应用-气象研究与应用

文章信息/Info

Title:: Application on new threshold function denoising method in perceptible precipitation forecast

Author(s):: Chen Xi; Du Yanxia; Wu Yongkai; Wen Jichang; Quanzhou Municipal Meteorological Service, Quanzhou Fujian 362000

Keywords:: wavelet threshold denoising; new threshold function; signal to noise ratio; mean square error; perceptible precipitation; Kolmogorov entropy

摘要:: 大气观测数据中常含有噪声,如何减小噪声是大气观测数据应用的基础。本文在传统软、硬阈值函数去噪的基础上提出了新阈值函数去噪。首先,运用这三种方法对MATLAB中wnoise函数产生的bumps含噪信号进行去噪,并运用信噪比(SNR)和最小均方误差(MMSE)对去噪效果进行衡量。试验表明,新阈值函数去噪效果较好。最后,用新阈值函数去噪对可降水量时间序列进行处理,并用Kolmogorov熵和神经网络预测来衡量去噪效果。结果表明,新阈值函数同样适用于可降水量时间序列的去噪。

Abstract:: Atmospheric observation data often contains noise, how to reduce the noise is the basis of the atmospheric observation data applications. Based on the traditional soft and hard threshold denoising function, the new denoising threshold function is proposed. First of all, bumps noise signal from wnoise function in MATLAB was denoised by using these three methods and the denoising effect was measured by signal-to-noise ratio (SNR) and minimum mean square error (MMSE). The result shows that the denoising effect of new threshold function is better. Finally, time series of perceptible rainfall was deal with a new threshold function denoising procession, and the denoising effect was measured by neural network prediction and Kolmogorov entropy. The results show that the new threshold function applies to time series of perceptible precipitation denoising