Inversion of Rice Height Using Multitemporal TanDEM-X Polarimetric Interferometry SAR Data
GUO Xian-yu1, 2, 4, LI Kun2, 3, 4*, SHAO Yun2, 3, 4, Juan M. Lopez-Sanchez5, WANG Zhi-yong1
1. College of Geomatics, Shandong University of Science and Technology, Qingdao 266590, China
2. Laboratory of Radar Remote Sensing Application Technology, Institute of Remote Sensing and Digital Earth (RADI), Chinese Academy of Sciences, Beijing 100101, China
3. College of Resources and Environment, University of Chinese Academy of Sciences, Beijing 100049, China
4. Laboratory of Target Microwave Properties (LAMP), Zhongke Academy of Satellite Application in Deqing (DASA), Deqing 313200, China
5. University of Alicante, Alicante, 99, Spain
摘要： 水稻株高是水稻本身以及土壤、水文、气象等因素的综合反映，是水稻长势监测的重要指标。准确、高效、大范围的株高反演为水稻品种识别、物候监测、病虫害评估和产量预测等提供了可靠的依据。合成孔径雷达（SAR），具有全天时、全天候、穿透性的优势，成为水稻株高反演的重要手段之一。基于极化干涉测量（PolInSAR）的散射模型的反演算法具有严密的物理模型的支撑及较高的反演精度等特点，成为植被高度反演研究的热点。结合极化干涉SAR技术，构建了一种基于RVoG（Random Volume over Ground）模型的水稻株高反演算法，并利用2015年水稻生长季内9个时相的TanDEM-X极化干涉SAR数据，进行了水稻株高反演试验。首先基于每个时相下的极化干涉SAR数据分别得到8个复相干系数，利用这8个复相干系数在考虑卫星双站模式等情况下进行去相干处理，然后建立适用于水稻田特性的RVoG模型，接着构建基于该模型的水稻株高反演迭代算法，最后对9个时相下的TanDEM-X数据进行研究区的水稻株高反演及精度评定。结果表明，当水稻株高高于0.4m时，该方法的反演结果较好，决定系数（R2）为0.86，均方根误差RMSE为6.79 cm; 当水稻株高较低时（水稻株高小于0.4 m），反演误差在0.1～0.8 m之间，反演结果较差，被明显高估。通过分析认为，基于极化干涉理论，TanDEM-X数据在较好地反映出水稻植株的较大体散射量的前提下，利用所构建的基于RVOG模型的水稻株高反演算法，能够较好地反演株高在0.33～1.2 m的水稻株高。
关键词：水稻；株高；TanDEM-X；极化干涉SAR；RVoG模型；模式搜索法
Abstract：Rice height, an important index of rice growth monitoring, is a comprehensive reflection of rice itself, soil, hydrology and meteorology. So accurate, efficient, and large-scale inversion of rice crop height can provide reliable basis for rice identification, phenological monitoring, pest and yield estimation. Synthetic Aperture Radar (SAR), because of its all-weather day-night imaging capability, has been proven to be one of the important means for inversion of rice height. Based on polarimetric SARinterferometry (PolInSAR), the inversion algorithm of scattering model has the characteristics of support of rigorous physical model and high inversion accuracy, which has become a hot spot of inversion of vegetation height. In this paper, based on PolInSAR technology, a new method based on Random Volume over Ground (RVoG) model for rice height inversion was proposed. The inversion experiment of rice height was carried out using the TanDEM-X PolInSAR data of 9 time phases in the rice growing season of 2015. First of all, 8 complex coherence coefficients were obtained based on PolInSAR data in each phase. and these were used for a product of decorrelation under the consideration of satellite dual-station mode. Then, the RVoG model was established for the characteristics of paddy fields. Moreover, using this model, an iterative algorithm of rice height inversion was constructed. Finally, the rice height inversion and precision evaluation using TanDEM-X data of 9 time phases were carried out. The results showedthat when rice height was higher than 0.4 m, a coefficient of determination (R2) of was 0.86 and RMSE was 6.69 cm. When rice height was low (rice height was less than 40 cm), inversion resultswith inversion error of 0.1～0.8 m were significantly overestimated. Through analysis, on the premise that TanDEM X data reflect volume scattering of rice plants well, the inversion algorithm of rice height based on RVOG model can invert the rice height between 0.33～1.2 m with high precision.
国贤玉，李 坤，邵 芸，Juan M. Lopez-Sanchez，王志勇. 基于多时相TanDEM-X极化干涉SAR数据的水稻株高反演[J]. 光谱学与光谱分析, 2020, 40(03): 878-884.
GUO Xian-yu, LI Kun, SHAO Yun, Juan M. Lopez-Sanchez, WANG Zhi-yong. Inversion of Rice Height Using Multitemporal TanDEM-X Polarimetric Interferometry SAR Data. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2020, 40(03): 878-884.
[1] Pasher J, King D J. Remote Sens. Environ., 2010, 114(8): 1718.
[2] ZHOU Zhi-qiang, YUE Cai-rong, XU Tian-shu, et al（周志强, 岳彩荣, 徐天蜀，等）. Modern Agricultural Science and Technology（现代农业科技）, 2012, (2): 198.
[3] Guo H, Liu G, Liao J, et al. J. Appl. Remote. Sens., 2014, 8(1): 1.
[4] Karila K, Vastaranta M, Karjalainen M, et al. Remote Sens. Environ.，2015, 159: 259.
[5] Li Z, Guo M, Wang Z Q, et al. China Earth Sci., 2014, 57(6): 1314.
[6] Tanase M, Panciera R, Lowell K, et al. IEEE Trans. Geosci. Remote Sens., 2014, 52(8): 4671.
[7] Cloude S R, Papathanassiou K P. IEEE Trans. Geosci. Remote Sens.，1998, 36(5): 1551.
[8] Papathanassiou K P, Cloude S R. IEEE Trans. Geosci Remote Sens., 2001, 39(11)：2352.
[9] Hajnsek I, Kugler F, Lee S K, et al. IEEE Trans. Geosci. Remote Sens., 2009, 47(2)：481.
[10] ZHOU Mei, WANG Xin-hong, TANG Ling-li, et al（周 梅, 王新鸿, 唐伶俐, 等）. Science & Technology Review（科技导报）, 2008, 26(21): 90.
[11] Lee S K, Kugler F, Papathanassiou K P, et al. IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., 2013, 6(3): 1351.
[12] Treuhaft R N, Siqueira P R. Radio Sci., 2000, 35(1): 141.
[13] Treuhaft R N, Cloude S R. IEEE Trans. Geosci. Remote Sens., 1999, 37(5): 2620.
[14] Kugler F, Schulze D, Hajnsek I, et al. IEEE Trans. Geosci. Remote Sens., 2014, 52(10): 6404.
[15] Lopez-Sanchez J M, Fernando Vicente-Guijalba, Esra Erten, et al. Remote Sens. Environ., 2017, 192: 30.
[16] Erten E, Lopez-Sanchez J M, Yuzugullu O, et al. Remote Sens. Environ., 2016, 187: 130.
[17] Lopez-Sanchez J M, Ballester-Berman J D. Radio Sci., 2009, 44(2): 1.
[18] Gatelli F, Monti Guarnieri A, Parizzi F, et al. IEEE Trans. Geosci. Remote Sens., 1994, 32(4): 855.
[19] YUAN Xiao-fang, WANG Yao-nan, WU Liang-hong（袁小芳, 王耀南, 吴亮红）. Journal of Hunan University·Natural Sciences)（湖南大学学报·自然科学版）, 2007, 34(9): 30.
[20] Cloude S R. Polarisation: Applications in Remote Sensing. New York: Oxford University Press, 2009: 1.
