花生冠层APAR和FAPAR高光谱遥感估测模型研究

Hyperspectral remote sensing estimation model of peanut canopy apar and fapar

为了快速、无损监测花生生长发育,建立整个生育期内花生冠层吸收性光合有效辐射(APAR)和光合有效辐射吸收系数(FAPAR)的高光谱遥感估测模型,本试验利用高光谱遥感技术,测定沈阳地区5种不同生态类型的花生冠层光谱数据,同期获取APAR、FAPAR;并对原始光谱数据进行logρ、1/ρ、ρ′变换,构建6种植被指数,分别与APAR和FAPAR进行Pearson相关分析,并建立估测模型,对模型进行检验与评价。研究结果表明:4种变换形式的光谱数据中最优波段与APAR和FAPAR均达极显著相关(r≥0.3969,P<0.01),以ρ′在759 nm波段处与APAR(r=0.7574)和FAPAR(r=0.6276)的相关性最好,ρ′759nm处的高光谱参数与APAR、FAPAR建立的估测方程y=797.3846 e271.4883x(R=0.5512,P<0.01;RE=0.1213)和y=0.756e85.21x(R=0.4204,P<0.01,RE=0.0788)拟合系数最高、预测精度较好,估测效果很好。比值植被指数(RVI)、差值植被指数(DVI)、归一化植被指数(NDVI)、复归一化差值植被指数(RDVI)、垂直植被指数(PVI)和修改土壤调整植被指数(MSAVI)这6种植被指数的最优波段与APAR的相关性优于与FAPAR的相关性,MSAVI[723,761]与APAR所建立的对数函数y=1554ln(x)+1631(R=0.7566,P<0.01;RE=0.0870)和RDVI[731,764]与FAPAR建立的多项式函数y=1.027x2+0.713x+0.729(R=0.6194,P<0.01;RE=0.0699)的模拟值和实测值均达到了极显著、预测精度较高,MSAVI对APAR和RDVI对FAPAR估测效果很好。一阶微分光谱和植被指数可以较好地估测花生冠层APAR和FAPAR。

In order to monitor the growth and development of peanut quickly and nondestructively,hyperspectral data of peanut canopy including five cultivars of different ecological types in Shenyang,were measured with hyperspectral detection technology,absorbed photosynthetically active radiation(APAR)and fraction of absorbed photosynthetically active radiation(FAPAR)were acquired simultaneously;logρ、1/ρ、ρ'were performed on the original spectral data and six kinds of vegetation indexs were established,then pearson correlation analysis between peanut canopy APAR,FAPAR and hyperspectral data were discussed,and the hyperspectral remote sensing estimation models were constructed,and simulation models were tested for some chosen models in order to determine the best hyperspectral remote sensing estimation models of peanut canopy APAR and FAPAR in the whole growing period.The results were as follows:The correlations between the optimal bands of hyperspectral reflectance,its transformation forms and APAR,FAPAR were highly significant(r≥0.3969,P<0.01),the correlation between the first derivative spectralρ′when the spectral band was 759 nm and the APAR(r=0.7574),the FAPAR of peanut canopy(r=0.6276)was the best.APAR,FAPAR could be best simulated by estimation equations of hyperspectral parameters atρ′759nm,y=797.3846 e271.4883 x(R=0.5512,P<0.01;RE=0.1213)and y=0.756e85.21x(R=0.4204,P<0.01;RE=0.0788)with the highest fitting coefficient and higher value of average relative error(RE=0.0870).The correlations between the optimal bands of six kinds of vegetation indexs including ratio vegetation index(RVI),difference vegetation index(DVI),normalized vegetation index(NDVI)、renormalized difference vegetation index(RDVI),vertical vegetation index(PVI),modify soil adjusted vegetation index(MSAVI),and APAR is better than that with FAPAR,APAR,FAPAR could be best simulated by MSAVI[723,761]with logarithmic function y=1554ln(x)+1631(R=0.7566,P<0.01;RE=0.0870)and RDVI[731,764]with polynomial function y=1.027x2+0.713x+0.729(R=0.6194,P<0.01;RE=0.0699),both with higher value of average relative error,the correlation coefficient between simulating data and testing data was highly significant.Peanut canopy APAR and FAPAR could be better estimated byρ′and vegetation indexs.