利用RADARSAT-2雷达数据与改进的水云模型反演冬小麦叶面积指数

利用微波遥感反演植被参数往往受到植被分布不均、稀疏植被覆盖、地表裸土等因素影响,导致微波遥感用于农业参数估计的效果不佳。为解决微波遥感反演地表植被参数的问题,本研究在原有的水云模型基础上引入植被覆盖度以及裸土对于雷达后向散射系数的直接作用信息,提出一种改进的水云模型,并充分考虑地表植被的覆盖分布情况,结合地面实测数据及RADARSAT-2雷达数据对改进模型进行验证,然后根据改进模型通过查找表法反演出植被含水量,最后利用叶面积指数与植被含水量的经验关系间接得到叶面积指数的估测值。结果表明,改进的水云模型对后向散射系数的模拟精度比原有的水云模型精度高,模拟的决定系数在HH和VV极化时分别为0.850和0.739,均方根误差分别为0.918dB和1.475dB。由此可见,改进的模型对研究区植被条件更为敏感,能够较好地分离出植被与土壤信息对雷达后向散射系数的影响,同时利用其反演得到的叶面积指数精度较高,决定系数达到0.841,均方根误差为0.233。

Leaf Area Index Inversion of Winter Wheat Using RADARSAT-2 Data and Modified Water-cloud Model

The inversion of vegetation parameters using microwave remote sensing is usually affected by the heterogeneous distribution of vegetation,sparse vegetation cover and the influence from bare soil,which leads to the unsatisfactory results in parameter estimation for agricultural applications.In this study,in order to solve the problem of surface vegetation parameters retrieval using microwave remote sensing,a modified water-cloud model was developed to retrieve leaf area index (LAI) by adding vegetation coverage and the direct effect of bare soil on the total backscatter coefficients,which gave full consideration to the distribution of vegetation coverage.The modified model was validated between the simulated backscatter coefficients and measurements based on ground observations and RADARSAT-2 data in China.Consequently,a look-up table algorithm was applied to calculate the value of vegetation water content and retrieve LAI according to the linear relationship between the vegetation water content and LAI.Results indicated that the modified model was more sensitive to the vegetation condition and the estimation accuracy was higher than that of water-cloud model.The R and RMSE were 0.850 and 0.739 dB in HH polarization,0.918 and 1.475 dB in VV polarization,respectively.Meanwhile,the modified model could effectively distinguish the scattering influences produced by the vegetation cover and bare soil component on the backscatter coefficients.The accuracy of LAI retrieval was significantly high with the R and RMSE of 0.841 and 0.233.This method provides support to estimate the LAI of winter wheat using radar data in a wide range.