中文版 | English
IOCAS-IR  > 海洋环流与波动重点实验室
陆地RFI源对星载L波段微波辐射计影响机制及检测算法研究
王新新
学位类型博士
导师魏恩泊
2024-05
学位授予单位中国科学院大学
学位授予地点中国科学院海洋研究所
学位名称理学博士
关键词海表盐度 L波段微波辐射计 射频干扰 机制分析 检测定位算法
摘要

    星载L波段微波辐射计可用于观测全球范围内的海表盐度（Sea Surface Salinity, SSS）。然而，由于L波段海表亮温对SSS的敏感性相对较弱，平均只有0.5K/psu，这对L波段微波辐射计的灵敏度要求很高，导致其极易受到无线射频干扰（Radio Frequency Interference, RFI）的影响。RFI是一个全球性的问题，已经持续多年，沿海分布的RFI源导致SSS卫星遥感数据质量下降，甚至完全缺失。由于陆地RFI源类型众多，强度各异，且混叠信号随时空变化，导致RFI对L波段微波辐射计的影响极其复杂，难以有效抑制。为此，本文充分发挥了L波段微波辐射计第三和第四Stokes参数对RFI敏感的特性，通过合成参数检测算法（Synthesis Parameters Detection Algorithm，SPDA）构建了极化特征参量WSPDA，评估了WSPDA对RFI特征的表征能力，并系统性研究了L波段微波辐射计对RFI的响应机制和时空变化特征；采用机器学习的方法开发了一种有约束迭代自适应的RFI检测、聚类、识别和定位算法；利用SMAP卫星L波段微波辐射计数据对不同类型RFI源开展了定位分析及验证研究；以西太平洋为研究区域，进一步探究了L波段微波辐射计对沿岸RFI源的响应特征及其对SSS反演的影响反馈。具体的工作和结论如下：

  （1）基于简化的弗林斯（Friis）传输方程，构建了单一RFI源等效全向辐射功率（Equivalent Isotropic Radiated Power, EIRP）和WSPDA之间的多元函数关系，识别了影响陆地RFI源干扰强度的特征因素。在理论分析方面，WSPDA能够近似反映RFI源对L波段微波辐射计的干扰强度。影响WSPDA的因素主要有RFI源的发射功率、发射天线与接收天线的增益，以及天线之间的距离和空间方位指向。对于陆地RFI源的相关参数未知，且发射功率可能随时间动态变化的问题，可以利用WSPDA的连续瞬时观测，精细化描述陆地RFI源干扰强度的时空变化特征。理论上，在RFI源发射功率恒定的条件下，当RFI源发射天线和L波段微波辐射计接收天线的主瓣对准并且距离最近时，WSPDA将达到最大值。在实验验证方面，基于上述特征设计了三个实验：距离、方位及发射功率因子影响特征实验。对比实验结果发现，发射功率是主要的影响特征因子；当RFI源发射功率越大、距离越近，以及天线对准角度越小时，RFI源的干扰强度越大，进一步验证了理论分析。

  （2）构建了基于RFI检测样本WSPDA强度和分布密度的多重迭代聚类算法（Multiple Iterative Clustering Algorithm based on Emission Intensity and Density，MICA-BEID），旨在更有效地识别和分类不同强度等级的RFI源。该算法利用机器学习方法，基于SMAP卫星WSPDA的概率密度函数（Probability Density Function， PDF）和累积分布函数（Cumulative Distribution Function, CDF），构建了一套有约束迭代自适应的RFI检测、聚类及识别定位算法。在RFI检测方面，该算法将基于WSPDA动态阈值的RFI检测样本和SMAP卫星原始RFI标记样本融合，构建了一个涵盖不同强度RFI源的检测样本互补数据集。在RFI检测样本聚类方面，该算法在基于密度的聚类算法（Density-Based Spatial Clustering for Applications with Noise, DBSCAN）的基础上，综合考虑了RFI检测样本WSPDA强度和密度的空间分布特征，并通过PDF和CDF等统计学方法，设置RFI源最大作用半径Rmax等边界条件，并通过多重迭代，形成了多个有效可分的聚类簇。在RFI源识别和定位方面，该算法基于WSPDA样本强度随着与最大强度（Wmax）位置的距离增大而减小的特征识别RFI源，并标记了聚类簇内Wmax位置为单次定位结果；通过对长时间尺度单次定位结果的二次迭代聚类，最终以聚类簇的质心位置表征RFI源的定位结果。经与卫星影像交叉对比分析，初步验证了该算法的定位精度，以在撒哈拉沙漠中识别的RFI源为例，定位精度能达到1km以内。

  （3）应用MICA-BEID算法深入开展了长时间尺度的RFI源定位和验证研究，着重分析了该算法对不同类型和强度RFI源的分级迭代聚类和识别定位的适应性，系统评估了该算法在处理RFI源时空变化方面的性能。在适应RFI源强度方面，在长时间尺度下，该算法表现出较强适应性，能够根据WSPDA强度等级由高至低进行分级迭代聚类，并定位不同强度等级的RFI源，尤其是能够有效识别和定位弱RFI源。在适应RFI源类型方面，通过与卫星遥感影像和已知RFI源等信息的协同对比分析，该算法能够识别通信信号基站、视频监控设备、广播电视发射台、雷达、无线电中继站及广播卫星系统电视接收机等当前普遍存在的多种类型RFI源，具有较好的适应性和通用性，具备检测新类型RFI源的潜力。在应对RFI源时空变化方面，该算法能够快速响应不同强度RFI源的时空变化，并保持稳定的定位性能。综合评估结果表明，MICA-BEID算法具备识别和定位不同类型和强度RFI源的能力，并能够适应其时空变化。在长时间尺度下，与其他算法相比，两种算法的定位具有较好的空间一致性；与已知RFI源相比，MICA-BEID算法各月的定位结果存在一定差异，但基本分布在L波段微波辐射计的有效视场（Effective Field of View , EFOV）内，呈现明显的方位指向性特征，最优定位精度可达0.2km左右。

  （4）聚焦受沿岸陆地RFI影响较大的西太平洋区域，探究了L波段微波辐射计对沿岸RFI源的响应特征及其对SSS反演的影响反馈。L波段微波辐射计海面直接观测数据对沿岸RFI源响应特征方面，利用MICA-BEID算法定位了西太平洋西北部沿岸地区的RFI源，并对其进行了合理分类。通过计算SMAP卫星过滤RFI前后的天线温度差的时空平均，发现沿岸RFI源对海面的影响约在1K以内；由于沿岸RFI源信号主要通过接收天线旁瓣进入L波段微波辐射计，不受RFI源位置的限制，沿岸多个RFI源对海面造成很大范围的叠加影响，呈现自西向东延伸的特征，距离沿岸越近，影响越大。沿岸RFI源对SSS反演数据质量影响方面，评估发现，在受RFI影响的区域，SMOS和SMAP卫星的SSS数据均存在不同程度的缺失；SMAP卫星的SSS数据质量相对更高，均方根误差（Root Mean Square Error, RMSE）约在1psu以内，而SMOS卫星约在2psu以内。这表明SMAP卫星在抑制RFI影响方面效果更好，但仍有待改进。此外，与沿岸RFI源对天线温度影响的空间分布特征类似，SSS反演数据的RMSE也呈现出随着与RFI源的距离增加而降低的空间分布特征。综合分析表明，相比于SMOS卫星，SMAP卫星采用的L波段微波辐射计具有充足的时频采样数据，能更有效地抑制沿岸RFI源的影响。

其他摘要
  The satellite-borne L-band microwave radiometers can be used to observe global sea surface salinity (SSS). However, due to the relatively weak sensitivity of L band brightness temperature to SSS, averaging only 0.5K/psu, there is a high sensitivity requirement for L-band microwave radiometers, making them susceptible to terrestrial radio frequency interference (RFI). The RFI is a global issue that has persisted for many years, and the RFI sources distributed along the coast have led to a decline in the quality of SSS satellite remote sensing data, and even complete data loss. Due to the diverse types and varying intensities of RFI sources widely distributed on land, coupled with the spatiotemporal variability of overlapping RFI signals, the impact of RFI on L-band microwave radiometers is extremely complex, making it difficult to effectively suppress. Therefore, fully utilizing the sensitivity of the third and fourth Stokes parameters of L-band microwave radiometer to RFI, this study constructs a polarimetric characteristic parameter called WSPDA through the Synthesis Parameters Detection Algorithm (SPDA), evaluates the capability of WSPDA to characterize RFI features, and systematically studies the response mechanism and spatiotemporal variation characteristics of L-band microwave radiometer to RFI. The study develops a constrained iterative adaptive algorithm for RFI detection, clustering, identification, and localization using machine learning methods, which is applied to analyze and validate the localization of various types of RFI sources using SMAP satellite L-band microwave radiometer data. Focusing on the Western Pacific as the study area, the study further explores the response characteristics of L-band microwave radiometers to coastal RFI sources and their impact on SSS retrieval. The specific findings and conclusions are outlined as follows:
  (1) Based on the simplified Friis transmission equation, a multivariate functional relationship was established between the Equivalent Isotropic Radiated Power (EIRP) of a single RFI source and WSPDA, which identifies the characteristic factors influencing the interference strength of terrestrial RFI sources. In terms of theoretical analysis, WSPDA can approximately reflect the interference strength of RFI sources on L-band microwave radiometers. The factors influencing WSPDA primarily include the transmission power of the RFI source, the gains of the transmitting and receiving antennas, as well as the distance and spatial orientation between the antennas. For situations where the relevant parameters of terrestrial RFI sources are unknown and the transmission power may vary dynamically over time, continuous instantaneous observations of WSPDA can be utilized to finely describe the spatiotemporal variation characteristics of the interference strength from terrestrial RFI sources. Theoretically, under conditions of constant transmission power from the RFI source, when the main lobe of the RFI source's transmitting antenna is aligned with the receiving antenna of the L-band microwave radiometer and they are at the closest distance, WSPDA will reach its maximum value. In terms of experimental verification, three experiments were designed based on the aforementioned characteristics, namely distance, orientation, and transmission power factor influence experiments. Comparative analysis of the experimental results reveals that transmission power is the primary influencing factor; the greater the transmission power of the RFI source is, the closer the distance is, and the smaller the antenna alignment angle is, the stronger the interference strength of the RFI source will be, further validating the theoretical analysis.
  (2) A novel algorithm, named the Multiple Iterative Clustering Algorithm based on Emission Intensity and Density (MICA-BEID), was constructed. It utilizes the WSPDA intensity and distribution density of RFI detection samples to enhance the identification and classification of RFI sources across various intensity levels. The algorithm includes a set of constrained iterative adaptive sub-algorithms for RFI detection, clustering, identification and localization constructed based on the Probability Density Function (PDF) and Cumulative Distribution Function (CDF) of WSPDA from the SMAP satellite using machine learning methods. In terms of RFI detection, the algorithm integrates RFI detection samples based on dynamic thresholds of WSPDA and original RFI flagged samples from the SMAP satellite, constructing a complementary dataset covering RFI sources of different intensities. In terms of clustering of RFI detection samples, the algorithm takes the spatial distribution characteristics of WSPDA intensity and density in RFI detection samples into comprehensive consideration on the basis of the Density-Based Spatial Clustering for Applications with Noise (DBSCAN) algorithm. It sets boundary conditions, such as the maximum effective radius (Rmax) of RFI sources, using statistical methods like PDF and CDF. Through multiple iterations, the algorithm forms multiple effectively separable clusters. In terms of RFI source identification and localization, the algorithm identifies RFI sources based on the characteristic that the intensity of WSPDA samples diminishes as the distance increases from the location of maximum intensity (Wmax), and flags the Wmax position within a cluster as the single localization result. Through secondary iterative clustering of single localization results over an extended time scale, the algorithm ultimately represents the localization results of RFI sources by the centroid position of the clusters. Through cross-comparison analysis with satellite imagery, the algorithm's localization accuracy was validated preliminarily. With the identification of RFI sources in the Sahara Desert as an example, the localization accuracy can reach within 1 km.
  (3) In-depth research on the localization and validation of RFI sources over a long time scale was conducted using the MICA-BEID algorithm. The research focused on analyzing the adaptability of the algorithm for hierarchical iterative clustering and identification, localization of RFI sources of various types and intensities and systematically evaluated the performance of the algorithm in handling the spatiotemporal variations of RFI sources. In terms of adapting to the intensity of RFI sources, the algorithm exhibits strong adaptability over a long time scale, and it can conduct hierarchical iterative clustering based on WSPDA intensity levels from high to low and localize RFI sources of different intensity levels, in particular, it is effective in identifying and localizing weak RFI sources. In terms of adapting to different types of RFI sources, through collaborative comparative analysis with such information as satellite remote sensing imagery, and known RFI sources, the algorithm is capable of identifying a variety of commonly present RFI sources, including communication signal base stations, video surveillance equipment, broadcasting and television transmission stations, radar, radio relay stations, and broadcast satellite television receivers. The algorithm demonstrates good adaptability and versatility, with the potential to detect new types of RFI sources. In terms of addressing the spatiotemporal variations of RFI sources, the algorithm can rapidly respond to the spatiotemporal changes of RFI sources of different intensities while maintaining stable localization performance. Comprehensive assessment results indicate that the MICA-BEID algorithm has the capability to identify and localize RFI sources of various types and intensities and can adapt to their spatiotemporal changes. The algorithm exhibits good spatial consistency with other algorithms over a long time scale. Despite certain differences in the monthly localization results of the MICA-BEID algorithm compared to known RFI sources, the results are essentially distributed within the Effective Field of View (EFOV) of the L-band microwave radiometer, exhibiting distinct directional characteristics. The optimal localization accuracy can reach approximately 0.2 km.
  (4) Focusing on the Western Pacific region, which is significantly affected by RFI from coastal land areas, the study explores the response characteristics of L-band microwave radiometers to coastal RFI sources and their feedback on SSS retrieval. In terms of the response characteristics of direct sea surface observation data from L-band microwave radiometers to coastal RFI sources, the RFI sources along the northwestern coast of the Western Pacific Ocean were located and appropriately classified using the MICA-BEID algorithm. By calculating the spatiotemporal average of the difference in antenna temperature before and after filtering RFI by the SMAP satellite L-band microwave radiometer, it was found that the influence of coastal RFI sources on the sea surface is approximately within 1K. Due to the fact that signals from coastal RFI sources primarily enter the L-band microwave radiometer through the sidelobes of the receiving antenna, without limitation by the location of the RFI sources, multiple RFI sources along the coast contribute to a significant overlapping effect on the sea surface over a large area, exhibiting a characteristic of extending from west to east. The closer to the coast is, the greater the impact will be. In terms of the impact of coastal RFI sources on the quality of SSS retrieval data, assessments show that in areas affected by RFI, there is a varying degree of data loss in the SSS data from both SMOS and SMAP satellites. The quality of SSS data from the SMAP satellite is relatively higher, with a root mean square error (RMSE) of approximately within 1 psu, while for the SMOS satellite, it is approximately within 2 psu. This indicates that the SMAP satellite is more effective in suppressing the interference of RFI, but there is still room for improvement. Additionally, similar to the spatial distribution characteristics of the impact of coastal RFI sources on the antenna temperature, the RMSE of SSS retrieval data also exhibits a spatial distribution characteristic of decreasing with increasing distance from RFI sources. Comprehensive analysis indicates that, compared to the SMOS satellite, the L-band microwave radiometer used by the SMAP satellite has ample time-frequency sampling data, which can more effectively suppress the impact of coastal RFI sources.
语种中文
目录

第1章 绪论... 1

1.1 研究背景概况... 1

1.1.1 SSS卫星微波遥感概况... 2

1.1.2 RFI影响基本概况... 5

1.2 国内外研究现状... 7

1.2.1 陆地RFI源对L波段星载辐射计影响机制研究进展... 7

1.2.2 RFI源检测与抑制算法研究进展... 9

1.3 研究内容... 12

1.4 研究意义... 14

第2章 RFI源对L波段微波辐射计影响机理与特征分析... 16

2.1 引言... 16

2.2 RFI影响特征分析理论基础... 16

2.3 RFI影响因子特征实验... 19

2.3.1 实验平台搭建... 19

2.3.2 实验仪器参数... 19

2.3.3 实验方案设计... 20

2.3.4 实验结果分析... 21

2.4 典型RFI源电视接收机干扰L波段微波辐射计机理... 28

2.4.1 日本地面广播系统RFI源情况... 28

2.4.2 电视接收机干扰L波段微波辐射计机理... 28

2.5 小结... 30

第3章 陆地RFI源检测识别定位算法构建... 32

3.1 引言... 32

3.2 SMOS和SMAP卫星RFI检测算法... 33

3.2.1 SMOS卫星... 33

3.2.2 SMAP卫星... 33

3.3 RFI检测算法面临的潜在限制... 35

3.4 基于RFI强度和空间分布特征的检测识别定位算法构建... 36

3.4.1 算法基本流程... 36

3.4.2 RFI疑似样本检测... 37

3.4.3 RFI样本迭代聚类... 43

3.4.4 RFI源识别与定位... 57

3.5 小结... 61

第4章 基于MICA-BEID算法的不同类型RFI源识别定位研究... 63

4.1 引言... 63

4.2 MICA-BEID算法对不同强度RFI源的检测性能评估... 63

4.2.1 RFI检测样本代表性评估... 63

4.2.2 极化合成参量携带信息分析... 66

4.3 MICA-BEID算法对不同强度RFI源的分级迭代聚类研究... 68

4.4 基于长时间尺度数据的不同类型RFI源定位结果分析... 76

4.4.1 与SMAP卫星定位结果对比分析... 76

4.4.2 不同类型RFI源定位结果分析... 85

4.4.3 不同类型RFI源定位结果时空变化特征... 88

4.4.4 RFI源对不同卫星的影响差异... 93

4.4.5 日本典型RFI源的时空变化特征... 96

4.5 小结... 106

第5章 陆地RFI源对SSS卫星遥感数据质量影响研究... 109

5.1 引言... 109

5.2 陆地RFI源对西太平洋的影响特征分析... 110

5.3 西太平洋西北部沿岸地区RFI源检测与定位... 111

5.4 陆地RFI源对西太平洋SSS反演的影响评估... 113

5.4.1 评估数据情况... 113

5.4.2 数据匹配评估方法... 115

5.5 评估结果... 117

5.5.1 数据匹配结果... 117

5.5.2 SSS卫星遥感数据与实测数据差异dSSS和Bias分布特征... 119

5.5.3 SSS卫星遥感数据与实测数据RMSE分布特征... 122

5.5.4 RMSE随距离和纬度变化特征... 124

5.6 小结... 126

第6章 总结与展望... 128

6.1 结论... 128

6.2 创新点... 130

6.3 不足与展望... 131

参考文献... 133

致谢... 141

作者简历及攻读学位期间发表的学术论文与其他相关学术成果 143

文献类型学位论文
条目标识符http://ir.qdio.ac.cn/handle/337002/185217
专题海洋环流与波动重点实验室
推荐引用方式
GB/T 7714		 王新新. 陆地RFI源对星载L波段微波辐射计影响机制及检测算法研究[D]. 中国科学院海洋研究所. 中国科学院大学,2024.
条目包含的文件
文件名称/大小 文献类型 版本类型 开放类型 使用许可
陆地RFI源对星载L波段微波辐射计影响机（74960KB）学位论文 暂不开放CC BY-NC-SA
个性服务
推荐该条目
保存到收藏夹
查看访问统计
导出为Endnote文件
谷歌学术
谷歌学术中相似的文章
[王新新]的文章
百度学术
百度学术中相似的文章
[王新新]的文章
必应学术
必应学术中相似的文章
[王新新]的文章
相关权益政策
暂无数据
收藏/分享
所有评论 (0)
暂无评论
 

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。