基于RVoG模型的双极化SAR水稻株高反演

Inversion of Rice Plant Height from Dual-Polarization SAR Data Based on RVoG Scattering Model

水稻株高被广泛应用于物候监测、水稻健康评估及产量预测等领域,极化合成孔径雷达(Polarimetric Synthetic Aperture Radar,PolSAR)信号能穿透水稻冠层记录水稻垂直结构信息,有助于实现高分辨率、大范围水稻株高提取。该文提出一种适用于非干涉条件下PolSAR数据水稻株高反演方法:利用极化SAR分解技术分离冠层及地表散射信号,引入随机地体二层散射模型(Random Volume over Ground,RVoG)对分解得到的水稻冠层散射能量进行建模,从而建立水稻株高反演模型;最后,联合邻域同质像素并利用NSGA-Ⅱ遗传算法对模型进行解算。利用2019年13景Sentinel-1影像在西班牙地区进行试验,水稻株高反演精度达到0.1 m,R^(2)达0.96以上,证明新方法能较好地适用于非干涉条件下PolSAR数据水稻株高反演。

Rice plant height is widely used in the fields of climate monitoring,health assessment and yield prediction.Polarimetric Synthetic Aperture Radar(PolSAR)signals can penetrate the rice canopy and record the vertical structure information of rice,which provides the feasibility of high-resolution and large-scale rice plant height extraction.To realize rice plant height estimation,a correlation between PolSAR observations and rice biophysical parameters needs to be established.In view of this,this paper proposes a rice plant height inversion method for PolSAR data under non-interferometric conditions:first,the dual-polarization decomposition technique is used to separate the scattered signals from the canopy and ground surface;after that,the random volume over ground(RVoG)scattering model is introduced to model the scattered energy of rice canopy obtained from the decomposition,and the rice height inversion model is established;finally,the model is solved by combining neighborhood homogeneous pixels and NSGA-Ⅱgenetic algorithm.The experimental validation was carried out in the southern region of Seville in Spain using Sentinel-1 image in 2019,and the results showed that the accuracy of rice plant height inversion reached 0.1 m and R^(2) was above 0.96.It proves that the new method can be better applied to the rice plant height inversion of PolSAR data under non-interferometric conditions,and can provide strong support for realizing efficient rice plant height monitoring.