python数据处理高斯滤波_论文推荐｜郭飞霄：GRACE时变重力场各向异性高斯组合滤波方法...

《测绘学报》

构建与学术的桥梁 拉近与权威的距离

GRACE时变重力场各向异性高斯组合滤波方法

郭飞霄^1,2,3, 孙中苗^2,3, 任飞龙⁴, 汪菲菲⁵

1. 信息工程大学, 河南 郑州 450001;

2. 地理信息工程国家重点实验室, 陕西 西安 710054;

3. 西安测绘研究所, 陕西 西安 710054;

4. 西安航天天绘数据技术有限公司, 陕西 西安 710054;

5. 31002部队, 北京 100094

收稿日期：2018-08-22；修回日期：2019-01-25

基金项目：国家自然科学基金(41674082；41774018)；大地测量与地球动力学国家重点实验室开放基金(SKLGED2017-3-2-E)

第一作者简介：郭飞霄(1988-), 男, 博士生, 工程师, 研究方向为卫星重力数据处理与应用。E-mail：gravity_gfx@126.com

摘要：受测量误差等因素影响，直接使用GRACE时变重力场模型的地表质量变化反演结果呈现严重条带噪声，必须采用滤波消除。本文对不同滤波方法进行了试验分析，以信噪比最大为准则，确定了不同滤波方法的最优滤波参数，并在此基础上提出了一种各向异性组合滤波方法。该方法根据时变重力场模型球谐系数误差特性，结合各向异性高斯滤波和均方根滤波特点，对精度较高的低次项系数采用较大权重以保留更多有效信号，而对精度较差的高次项系数采用较小权重以压制噪声。不同于传统的两步法组合滤波，该方法仅需进行一步滤波处理。试验结果表明，本文提出的各向异性组合滤波方法计算步骤简单，能够有效消除条带噪声；与单一滤波和传统两步法组合滤波方法相比，提高了反演结果信噪比，保留了更多真实信号。

关键词：组合滤波 条带噪声 各向异性高斯滤波 均方根滤波 信噪比

Non-isotropic Gaussian combination filtering method of GRACE time-variable gravity

GUO Feixiao^1,2,3, SUN Zhongmiao^2,3, REN Feilong⁴, WANG Feifei⁵

1. Information and Engineering University, Zhengzhou 450001, China;

2. State Key Laboratory of Geo-Information Engineering, Xi'an 710054, China;

3. Xi'an Research Institute of Surveying and Mapping, Xi'an 710054, China;

4. Xi'an Aerospace Data Technology Co., Ltd, Xi'an 710054, China;

5. Troops 31002, Beijing 100094, China

Foundation support: The National Natural Science Foundation of China (Nos. 41674082;41774018); The Open Foundation of State Key Laboratory of Geodesy and Earth's Dynamics (No. SKLGED2017-3-2-E)

First author: GUO Feixiao(1988—), male, PhD candidate, engineer, majors in satellite gravimetry data processing and application.E-mail: gravity_gfx@126.com.

Abstract: Affected by factors of measurement errors and so on, the surface mass variation inversion results would be dominated by serious north-to-south stripe noise with using time-variable gravity models directly. Filtering methods must be adopted. Optimal parameters of different filtering methods are determined by taking signal-to-noise ratio maximum as the criterion. On this basis, a new non-isotropic combination filtering method is proposed based on the features of time-variable gravity model spherical harmonic serrors. The new non-isotropic combination filtering method is different from traditional two-steps combination filtering methods. It just performs in one step with combining non-isotropic Gaussian filtering and RMS filtering. The basic idea of the new combination filtering method is to apply larger weights to low-order spherical harmonic coefficients while smaller weights to h