改良式干涉相位濾波器應用於地表變形偵測

A Refined Interferometric Phase Filter with Applications to Detection of Surface Deformation

趙進福(Chin-Fu Chao)

指導教授 ： 陳錕山

國立中央大學/地球科學學院/太空科學研究所/博士(2012年)

摘要

干涉合成孔徑雷達(Interferometric synthetic aperture radar, InSAR)已被廣泛應用於地表高度與地表變動的測量。地表有關的高度資訊可以從雷達干涉的相位推導出來。然而，相位雜訊降低了信息的準確性和可靠度。因此，若由干涉相位反演地表高程或變動，則降低相位雜訊的最小化是至關重要的。
基於李氏干涉雷達濾波器(Lee INSAR filter)和Sigma濾波器，本文提出一個改良式干涉相位濾波器。改良式演算法的基本概念是根據像元週圍區域的雜訊強弱水平，適時調整相位雜訊的最小化，特別在干涉條紋破碎不連續的極端情況下，同時達到(1)保持干涉條紋應有的邊緣形狀、(2)減少殘差像元的數量、(3)降低相位的誤差/偏移、(4)降低相位的變異量等四個目標。經過模擬影像和實際干涉雷達應用影像的驗證顯示，改良式濾波器之濾波性能優於其他常用的濾波器。
此外，本研究將改良式濾波器實際應用於茂密植被所覆蓋的區域(越南北部)，低同調性(low coherence)干涉影像，證明有良好的濾波性能。藉由改善干涉雷達影像的品質，進一步探討和分析兩個實際觀測地表變形情況的案例，其結果如下:
（1）對哀牢山 - 紅河不連續的地質斷裂帶（A-RRSZ）的觀測：A - RRSZ是中南半島和中國南部的交界處。這地區的地質結構具有持續地層活動的顯著特性，是現今地球科學界的關注焦點。本研究旨在應用差分干涉雷達技術（DInSAR），對於A-RRSZ在越南河內市附近區域的地表變形觀測。利用L波段的ALOS/PALSAR和C波段的ENVISAT衛星影像來觀測地變形。初步結果顯示，在2007～2009年期間，地表活動是沿著A-RRSZ斷裂帶由西北向東南的方向變化，這區域地殼表面因為地層相互擠壓結果，有明顯每年約1.3公分隆起縐摺的現象。此外，也觀察到這10公里寬A-RRSZ斷裂帶的地表變形與鄰近地區不同，相對比較穩定。
（2）對越南河內市地層下陷的觀測：有學者研究發現在越南河內市抽取地下水源的人類行為也是造成這市區地層下陷的影響因素之一。由雷達干涉的觀測結果，與5個監測水井的地下水位變化相互比較，同時也與其他方法量測的結果相互驗證，顯示在1996～2010年間，河內市地層下陷的變化，早期是發生在北邊區域，後來北邊區域地層變化趨於穩定，而逐漸轉移到南邊。河內市區繼續發生約每年3公分的地層下陷。

關鍵字

干涉雷達 ； 干涉相位雜訊 ； 改良式干涉相位濾波器

並列摘要

Radar interferometry has been widely applied in measuring the surface height and deformation which may be derived from phase interferogram. However, the inherent phase noise reduces the accuracy and reliability of that information. Hence, the minimization of phase noise is essential to the retrieval of surface height information at acceptable accuracy.
This work proposes a refined filter that is based on the revised Lee InSAR filter, the revised sigma filter and the Minimun Mean Square Error estimator. The basic idea is to use 16 directional windows to counter the difficulty of preserving the fringe patterns, to select proper pixels within a reasonable range, and to filter adaptively the interferometric phase according to the local noise level, especially in such extreme cases as involve broken fringes. The goals are to preserve the fringe pattern, to reduce the number of residues, and to minimize the phase error and deviation at a same time. The proposed filter was validated using both simulated data and real interferometer data. Results confirm that the filter performance is better than that of commonly used filters.
Furthermore, we also show a practical application of the proposed filter in area of dense vegetation of low coherence to demonstrate its performance. By improving the interferometric quality, two actual sets of observations of surface deformation in the area of low coherence are further analyzed in what follows.
(1) Observations of the Ailao-Red River Shear Zone (A-RRSZ) with discontinuous geologies: the A-RRSZ is at the junction between Indochina and south China. The geological structure that exhibits remarkable continuous activity is a focus of concern for the scientific community. This investigation applies Differential Interferometric Synthetic Aperture Radar (DInSAR) to observe surface deformation in the vicinity of near Hanoi in the A-RRSZ. Both C-band data of ENVISAT and L-band data of ALOS/PALSAR were used to analyze the mechanisms of deformation. The preliminary results reveal that the tectonic activity varied from the northwest to the southeast along the A-RRSZ near Hanoi. The crustal surface clearly underwent an uplift of 1.3 cm/year, caused by squeezing movement in the period 2007~2009. Surface deformation of the 10 km-wide A-RRSZ is observed not only to differ from that in adjacent areas but also to be relatively stable.
(2) Observations of land subsidence in Hanoi, northern Vietnam: human activity of groundwater extraction is one of the factors that cause surface deformation in the region. The interferometric results, which compared the groundwater levels at five monitoring wells, suggest a land subsidence in the northern part of Hanoi city in the period 1996~1999. Later, land subsidence stopped changing in the northern region and occurred at the rate of around 3 cm/year in the southern region in the period 2004~2010.

並列關鍵字

refined filter ； interferometry ； phase noise filtering ； INSAR filter

參考文獻

[23] H. V. Tran, and T. Q. Nguyen, ”The Alteration of Geological Environment in the Hanoi Area in the Process of Urbanization (in Vietnamese). Centre for Applied Geophysics, National Centre for Scientific Research, Hanoi, Vietnam,1990.

[1] J. S. Lee, K. W. Hoppel, S. A. Mango and A. R. Miller, “Intensity and Phase Statistics of Multilook Polarimetric and Interferometric SAR imagery,” IEEE Transactions on Geoscience and Remote sensing, vol. 32, no. 5, pp. 1017– 1028, September 1994.

[2] L. M. Carlos and F. Xavier, “Modeling and Reduction of SAR InterferometricPhase Noise in the Wavelet Domain,” IEEE Transactions on Geoscience and Remote sensing, vol. 40, no. 12, December 2002

[3] R. M. Goldstein and C. L. Werner, “Radar Interferogram Filtering for Geophysical Applications,” Geophys. Res. Letter, vol. 25, no. 21, pp. 4035–4038, 1998.

[4] A. L. B. Candeias, J. C. Mura, L. V. Dutra, J. R. Moreira and P. P. Santos, “Interferogram Phase Noise Reduction Using Morphological and Modified Median Filters,” in Proc. IEEE IGARSS’95, vol. 1, pp. 166–168.