|Abstract: ||本計劃將基於正在進行中的97 年國科會研究計劃之重要成果，利用已發展成功之 電磁波散射理論模式，以嚴謹的數學方法，發展先進的合成孔徑雷達反演技術來估算與 分析地表土壤溼度。合成孔徑雷達影像能提供被觀測地表的形態以及土壤的材質，因此 近年來需求頻繁的應用即是土壤濕度的反演，尤其是在大區域範圍的地質遙測應用中這 項需求越顯得重要。本計劃將建立先進合成孔徑雷達影像反演技術，並將其應用於地表 土壤濕度自動化估算與分析系統，且將於未來應用在台灣的合成孔徑雷達影像上，反演 並分析特定區域土壤的含水量。 在反演的問題上除了使用查詢表之外，我們將使用一些先進的反演演算法，如the Tikhonov-Phillips Method, The Truncated Singular Value Decomposition, Iterative Methods, Regularization by Discretization, and Maximum Entropy 等等方法。土壤含水量反演的成果 將可用來產生各影像點的介電常數，進而得到SAR 影像的土壤含水量的分佈。因此本 計劃將可發展出架構在理論模式上，專為土壤含水量反演與分析的SAR 影像處理軟體， 而且具備有去雜訊以及含水量分佈切割等影像處理技術能力。 在儘量降低計算時間且保有高度正確性的情況下，本系統預期將可成為改善SAR 影像解讀的重要工具，並且可用來重建地表土壤含水量的特徵。在整合遙測資料與現場 量測資料之後，我們將致力於確認並降低本系統預測值與現場實際量測值之間的誤差。|
Based on the achievements of the proceeding NSC research project in the combination of the wave scattering model and Synthetic Aperture Radar (SAR) techniques, where the Integral Equation Model has been chosen for the modeling of the surface scattering mechanism, this project is proposed to develop, in mathematically rigorous manner, the advanced SAR inversion techniques in order to estimate and analyze the near-surface soil moisture. Synthetic Aperture Radar (SAR) images contain information about morphology and ground conductivity of the observed terrain. The ability of SAR techniques to obtain information about physical properties of the surface has led to many innovative applications in geoscientific research and concerning environmental issues. One of these applications in greatly demand recently is the inversion of the soil moisture, which has been very important for the application of remote sensing to geology, particularly over large areas. Previous studies were commonly based on empirical, semiempirical, and image difference techniques for soil moisture inversion. There has been less research carried out from physical and theoretical EM scattering models because of complex mathematics manipulation and calculation. In this project, an automatic soil moisture inversion and analysis system based on the electromagnetic scattering model will be developed and applied to the soil moisture contents analyses of the SAR images acquired in Taiwan. Instead of look up table method, we summarize some advanced regularization methods for resolving the inverse problems, including the Tikhonov-Phillips Method, The Truncated Singular Value Decomposition, Iterative Methods, Regularization by Discretization, and Maximum Entropy. The results can be used to produce maps detailing the values of the dielectric constant index within the imaged sample and develop algorithms to obtain the soil contents distribution in the SAR image. The contribution to the radar signal from multiple scattering effects is also essentially taken into account for understanding and in turn optimization of the inversion system. Based on the developing theoretical model, a sophisticated image processing software dedicated to the proposed soil moisture inversion and analysis system will be developed, where the image processing techniques for noise removal and segmentation of moisture distribution are included. By decreasing computation time while preserving high accuracy this system is anticipated to serve as important tools for improving interpretation of SAR technologies, moreover, for reconstructing moisture characteristics of soil surface from SAR images. With the integration of the remote-sensing data and field observations, we will try to validate the prediction from our developed system with field works.