高光谱遥感模型对亚洲小车蝗危害程度研究

Study on Hyper-Spectral Remote Sensing Models for Monitoring Damage of Oedaleus Asiaticus(Orthoptera: Acrididae)

对内蒙古自治区锡林郭勒盟亚洲小车蝗危害进行了高光谱遥感监测，利用地面高光谱数据，分析和比较了正常羊草和亚洲小车蝗危害羊草冠层反射光谱和高光谱特征参数的差异，建立了高光谱特征参数与羊草叶面积指数的关系模型。结果表明：LAI和3个高光谱参数有极显著相关性，其中以红边峰值区（620～760nm）一阶微分总和与蓝边峰值区（430-470nm）一阶微分总和的比值为变量的虫害光谱指数（DI）模型精度最高，最适于反映研究区蝗虫的危害程度。从模型可以看出，DI介于51．57-79．83，蝗虫轻度发生；DI小于51．57，蝗虫严重发生。模型的预测值与实际值相关系数为0．948，平均相对误差为3．928％，表明模型的预测效果较好。

In the present paper, the ASD Field Spec Pro FRTM speetroradiometer was used for measuring damages of O. asiaticus in Xilin ColLeague, Inner Mongolia. First, the hyper-spectral data were analyzed and the canopy reflectance spectral data were compared between the healthy leaves and the infected leaves of Leymus chinensis. Second, the regression models between the leaf area index （LAI） and the hyper-speetral parameters were built, and different varieties of the O. asiaticus damage were used to test its precision. The results showed that there was a high correlation of three hyper-spectral data between the LAI and the first derivative data. Moreover the disease index （DI） model, which is mostly suitable for use in indicating the intensity of grasshoppers damage in the study area, containing the ratio of the sum of first derivative within red peak regions （620-760 nm, SDr） to the sum of first derivative within blue peak regions （430-470 nm SDb） was the best one. From the model, there were no damage if DI was over 72. 43, slight damage if DI was between 51.57 and 79.83, and serious damage if DI was less than 51.57. The model had the highest prediction precision with the correlation coefficient of prediction of 0. 948, and the mean relative error of 3. 928%. These results showed a good prediction of the model and indicated that the grasshopper damage could be estimated at the canopy level using hyper-spectral reflectance.