应用光谱反射测定的植物、土壤及混合光谱特性

以野外采集的土壤与活体植物叶片为研究对象,采用美国生产的SVC HR-1024便携式地物光谱仪,测定了纯土壤、纯植物叶片以及不同植物叶片盖度下的土壤光谱反射率,并对混合光谱特性进行了分析研究,建立了植物叶片盖度光谱估算模型。结果表明:(1)随着植物叶片盖度的增加,土壤光谱曲线逐渐呈现"五谷四峰"的特征,但始终不会高于植被光谱曲线;(2)植物叶片盖度与原始高光谱组成的差值、比值和归一化值3种形式的指数存在着良好的相关性,差值指数的显著性最为明显,组合较好的波段为410~710 nm与700~1400 nm,能反映相关系数最大的波段为690 nm和450 nm;(3)通过一元线性回归法、多元线性逐步回归法和偏最小二乘回归法预测植物叶片盖度可知:光谱特征参数与植物叶片盖度之间具有相关性,其中光化学指数与植被衰减指数的相关系数较大,模型拟合度较高,R^2均达到0.8以上;3种光谱特征参数(植被指数、红边参数、光谱吸收特征参数)中,植被指数与植物叶片盖度建立的模型最稳定、精度最高,可以有效地估算植物叶片盖度3种方法中,多元线性逐步回归法最适宜建立植物叶片盖度预测模型。

Spectral Characteristics of Plant-soil Mixture with Spectral Reflection Measurement

With the soil and living plant leaves collected in the field as the research object,we used the SVC HR-1024 portable ground spectrometer produced in the United States to measure the spectral reflectance of soil under soil,pure plant leaves and different plant leaf coverage.The mixed spectral characteristics were analyzed and studied,and the plant leaf coverage spectrum estimation model was established.The results showed that:(1)With the increase of plant blade coverage,the soil spectral curve showed the characteristics of"five concaves and four peaks"gradually,but it was not higher than the plant spectral curve.(2)The difference,ratio and normalized value of plant blade coverage were well correlated with the original hyperspectral components,and the difference index was the most significant.The best bands were 410-710 and 700-1400 nm,and the highest correlation coefficients were 690 and 450 nm,respectively.(3)It can be known from predicting plant blade coverage by unary linear regression,multiple linear stepwise regression and partial least squares regression that there is a correlation between spectral characteristic parameters and plant blade coverage,among which the correlation coefficient between photochemical index and plant attenuation index is larger,and the model fit is higher,R^2 is above 0.8;among the three kinds of vegetation index,red edge parameter,and spectral absorption characteristic parameter,vegetation index established by the model and plant leaf coverage are the most stable and accurate,and can effectively estimate plant leaf coverage.Among the three methods,multivariate linear stepwise regression method is the most suitable one to establish a prediction model for plant leaf coverage.