摘要
以渭北地区刺槐林为研究对象,通过计算高分辨率影像不同窗口的纹理参数,分别运用一元一次方程、一元二次方程、乘幂模型和指数模型,研究刺槐林野外实测叶面积指数(LAI)与快鸟影像纹理参数间的相关关系,遴选黄土高原地区估算刺槐林LAI最适合的纹理参数以及计算窗口大小.结果表明:纹理指数对LAI反演精度影响显著,其中角二阶矩阵和熵的估算精度明显高于其他纹理指数;计算窗口的大小显著影响LAI反演精度,非相似度和对比度的反演精度在窗口9×9达到最大,其余纹理指数对于LAI的估算精度随着计算窗口的变大不断降低,在窗口3×3时达到最大;乘幂模型的估算精度低于其他3种方法.
The black locust plantations located in Weibei area were chosen as research objects and the texture parameters of different window sizes from high resolution imagery were measured. Four different techniques, including simple linear regression model, quadratic regression model, power model and exponential model, were developed to describe the relationship between the texture pa rameters and field measurements of LAI and to select the most effective texture parameters and win dow size. The results showed that the texture parameters influenced the accuracy of LAI estimation. Angular second moment and entropy index yielded better adjust r2 than the other parameters. The r2 changed with the window size. Dissimilarity and contrast index gained the largest r2 when the win dow size was 9x9. The r2 of the other texture parameters reduced as the window size increased and a window size of 3x3 was more successful than any of the others. Power equation performed poorest than the other three techniques for estimation of LAI.
出处
《应用生态学报》
CAS
CSCD
北大核心
2014年第5期1266-1274,共9页
Chinese Journal of Applied Ecology
基金
"十二五"农村领域国家科技支撑计划项目(2012BAD22B0302)
森林培育学教学团队项目(Z105021003)资助
关键词
叶面积指数
空间特征
高分辨率影像
刺槐林
leaf area index
spatial characteristics
high resolution imagery
Robinia pseudoacaciaplantation.
