星载扫描微波散射计方位向高分辨率处理方法研究

发布时间：2018-05-24 08:34

本文选题：散射计 + 方位向高分辨率　；参考：《中国科学院国家空间科学中心》2017年博士论文

【摘要】：微波散射计具有观测刈幅宽、全球覆盖周期短、测量精度高等优点,正逐渐成为陆地定量遥感的最佳选择。由于散射计最初主要针对空间分辨率要求不高的海面风场测量,因此对于空间分辨率要求较高的陆地参数观测存在一定的局限性。针对这一问题,论文开展了基于高密度采样的散射计高分辨率处理方法的研究。虽然前人已经对星载微波散射计高分辨率处理开展了初步的研究,但尚缺少系统、完整性的研究工作,特别是针对散射计分辨率提高以及辐射精度保持的相关研究还很少。因此,论文采用由理论建模、模型求解,以及解的定量评价层层递进的研究路线开展了星载扫描微波散射计方位向高分辨率处理方法的系统研究,具体内容如下:首先,以微波散射计后向散射模型为基础,参考孔径函数采样与重建理论,建立基于高密度采样的散射计后向散射系数(σ~0)重建模型。在此基础上,综合考虑散射计实际处理过程,提出基于条带σ~0测量的方位向一维高分辨率处理模型。该模型将重建模型的二维问题简化为一维问题,一方面提高了运算效率,另一方面为散射计高分辨率处理算法的研究奠定了基础。其次,对高分辨率处理算法进行了研究,指出散射计方位向高分辨率处理的本质是消除空间响应函数对地面σ~0的加权作用,属于反卷积高分辨率重建问题。重点分析线性逆问题求解的病态性,并引入规整化方法来缓解该病态性,同时在该方法框架下,提出一种改进的适应性规整化算法。另外,搭建了DPS散射计0s高分辨率重建仿真模型,对比分析多种算法性能,结果显示改进的适应性规整化算法具有同L1算法相似的分辨率提高特性,且重建准确度明显优于L1算法。再次,对已有的散射计图像重建算法(AART、MART和SIR)开展定量分析,提出利用局部脉冲响应函数和归一化标准偏差计算分辨率和重建精度的定量评价方法。借此讨论了多个参数(测量误差Kp、SRF波束宽度、SRF函数类型以及地面σ~0)对散射计图像重建算法空间分辨率和重建精度的影响规律,得到了散射计图像重建算法的分辨率与重建精度折中关系曲线。研究结果表明:三种散射计图像重建算法均不能实现DPS散射计2~5 km的重建分辨率要求;AART算法勉强可实现5 km的分辨率,但重建精度过低;SIR算法具有较高的重建精度,但所能达到的最优分辨率远低于5 km。最后,开展规整化反卷积高分辨率处理算法的分辨率和σ~0重建误差特性定量研究。基于散射计测量特点,通过引入与算法重建分辨率相同的系统无噪测量结果作为重建σ~0的真值,提出一种新的σ~0重建准确度与精度评价模型。通过分析规整化反卷积算法的空间分辨率与重建误差的定量关系,得到了算法空间分辨率分别与σ~0重建准确度/精度的定量关系。研究结果表明:在同时满足σ~0重建分辨率2~5 km及重建精度优于0.5 d B的条件下,Tikhonov规整化反卷积高分辨率重建算法在测量误差Kp=7%,Kp=10%和Kp=12%时所能获得的最佳分辨率分别为3 km、4 km和5 km。通过以上研究,对散射计高分辨率处理模型的建立、处理算法的对比分析、算法分辨率和处理精度的定量评价等进行了深入、系统的研究,建立了一套完整的散射计方位向高分辨率处理研究体系,推进了现有星载微波散射计高分辨率处理技术的发展,且为DPS散射计系统参数设计和数据处理提供了理论支撑。
[Abstract]:The microwave scatterometer has the advantages of observation and mowing width, short global coverage cycle and high measurement precision, which is becoming the best choice for land quantitative remote sensing. Because the scatterometer is mainly focused on the low spatial resolution of the sea surface wind field measurement at first, there are some limitations for the high land parameter observation for spatial resolution. In order to solve this problem, a high resolution processing method based on high density sampling is carried out. Although the predecessors have studied the high resolution processing of the spaceborne microwave scatterometer, there is no system and integrity research, especially for the enhancement of the resolution of the scatterometer and the preservation of the accuracy of the radiometer. Therefore, a systematic study of the azimuthal high-resolution processing method of spaceborne scanning microwave scattermeter is carried out by using theoretical modeling, model solving and quantitative evaluation of the solution. The specific contents are as follows: first, the reference aperture function is based on the backscatter model of the microwave scatterometer. The backscatter coefficient (sigma ~0) reconstruction model of the scatterometer based on high density sampling is established. On this basis, considering the actual processing process of the scatterometer, the azimuthal high resolution processing model based on the strip Sigma ~0 measurement is proposed. The model simplifies the two dimensional problem of the heavy modeling to one dimension problem, and improves one aspect. On the other hand, it lays the foundation for the study of the high resolution processing algorithm of the scatterometer. Secondly, the high resolution processing algorithm is studied. It is pointed out that the essence of the azimuth direction to high resolution processing is to eliminate the weighting of the spatial response function to the ground Sigma ~0, and it is the problem of the high resolution reconstruction of the deconvolution. The ill conditioned linear inverse problem is solved, and the normalization method is introduced to alleviate the morbid condition. At the same time, an improved adaptive regularization algorithm is proposed under the framework of the method. In addition, the DPS scatterometer 0s high resolution reconstruction simulation model is built, and the performance of a variety of algorithms is compared and analyzed. The results show that the improved adaptive regularization algorithm is available. Similar to the L1 algorithm, the resolution is improved, and the accuracy of the reconstruction is obviously better than that of the L1 algorithm. Thirdly, the quantitative analysis of the existing image reconstruction algorithm (AART, MART and SIR) is carried out, and a quantitative evaluation method is proposed to calculate the resolution and the reconstruction precision by using the local impulse response function and the normalized standard deviation. The influence of the number (measurement error Kp, SRF beam width, SRF function type and ground Sigma ~0) on the spatial resolution and reconstruction precision of the reconstructing algorithm of the scatterometer is obtained. The relation curve between the resolution and the reconstruction accuracy of the scatterometer image reconstruction algorithm is obtained. The results show that all the three scatterometer image reconstruction algorithms can not achieve the DPS dispersion. The resolution of the reconstruction of 2~5 km is required; the AART algorithm can barely achieve the resolution of 5 km, but the reconstruction precision is too low; the SIR algorithm has a high reconstruction precision, but the optimal resolution can be far lower than 5 km.. The resolution of the normalized deconvolution high resolution processing algorithm and the Sigma ~0 reconstruction error characteristics are quantitatively studied. By introducing the non noise measurement results of the system with the same resolution as the algorithm to reconstruct the true value of the sigma ~0, a new evaluation model for the accuracy and accuracy of the sigma ~0 reconstruction is proposed. By analyzing the quantitative relation between the spatial resolution and the reconstruction error of the regularized deconvolution algorithm, the spatial resolution and the sigma of the algorithm are obtained. The quantitative relationship between the accuracy / accuracy of ~0 reconstruction shows that, under the condition that the resolution 2~5 km and the reconstruction precision are better than 0.5 D B at the same time, the best resolution of the Tikhonov regularized deconvolution high resolution reconstruction algorithm is 3 km, 4 km and 5, respectively, when the measurement error Kp=7% is Kp=7%, and the Kp=10% and Kp=12% respectively. Research on the establishment of the high resolution processing model of the scatterometer, the contrast analysis of the processing algorithm, the resolution of the algorithm and the quantitative evaluation of the processing precision and so on. A complete set of research system for the azimuth high resolution processing of the scatterometer has been established, and the high resolution processing technology of the existing spaceborne microwave scatterometer has been promoted. It provides theoretical support for parameter design and data processing of DPS scatterometer.
【学位授予单位】：中国科学院国家空间科学中心
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P407.7

【相似文献】