基于遥感反演的中国森林冠层叶氮浓度空间格局初步研究

Preliminary study on the spatial pattern of forest canopy foliar nitrogen concentration in China based on satellite remote sensing

森林冠层叶氮浓度(CNC)是森林生态系统生产力模拟的重要参数,因此获取并揭示我国森林CNC的空间分布格局对于准确评估我国森林生态系统生产力和碳通量都具有重要意义。本研究通过实测我国典型森林冠层叶片氮浓度并使用EO-1 Hyperion高光谱数据,建立CNC与近红外反射率(NIR)的统计模型,然后结合中分辨率成像光谱仪(MODIS)的反射率数据,实现生长季中国森林CNC的反演和分析,并进行了初步验证。结果表明:1)基于多种森林类型的采样数据,CNC与Hyperion NIR之间存在显著正相关关系(R^2=0.75,P<0.0001);2)反演的我国森林CNC分布格局大致呈现东南高、西北低的空间特征,其中海南省森林CNC均值最高,而天津市的最低;3)反演的中国森林CNC介于0.49~3.63 g/100 g之间,平均为2.24±0.28 g/100 g,处于全球森林植被叶氮浓度范围之内;4)反演的CNC与实测值的相关系数为0.52,RMSE为0.43 g/100 g,说明反演结果可以基本反映出我国森林CNC的空间格局和统计特征,但部分区域存在一定高估。森林CNC空间格局主要受森林植被功能型空间变异的影响,未来应对单一森林类型CNC的反演进行详细评估。本研究对于优化全国尺度森林生产力模拟具有重要意义。

Vegetation canopy foliar nitrogen is a key element for plants photosynthesis,and forest canopy foliar nitrogen concentration(CNC)is an important parameter for the simulation of ecosystem productivity.So it is very significant to obtain the spatial pattern of CNC for accurate estimation of ecosystem productivity and carbon fluxes at the regional scale.In this study,by measuring the CNCand extracting near infrared reflectance(NIR)from EO-1 Hyperion imagery,the statistical model between CNC and NIR was established.Then we applied the model on the BRDF reflectance of Moderate Resolution Imaging Spectroradiometer(MODIS)data,and implemented forest CNC inversion for China in growing season.Then the preliminary verification to the inversion result was carried out using the measured data.The results showed that:1)based on the sample data in different forest types,there is a significant correlation between the measured CNC and NIR(R2=0.75,P<0.0001);2)there is an obvious spatial distribution characteristic forest CNC in China shows,which is divided byHu’s line(Aihui-Tengchong Line),with high CNC at southeast of the line and low CNC at the northwest of the line;the highest CNC was in Hainan province and the lowest CNC was in Tianjin.This spatial pattern characteristicsis concerned with the forest type distribution in China;3)the inversion results show the Chinese forest CNC is about at 0.49-3.63 g/100 g,and the mean is 2.24 g/100 g with the standard deviation of 0.28 g/100 g.The values of CNC fall within the numerical range of the world leaf nitrogen concentration;4)the correlation coefficient of measured values and the inversion value reached 0.52,with RMSE of 0.43 g/100 g.Inversion result can show up the major characteristics of CNC pattern in China,but there may be a little overestimates in some parts of China.In addition,we found that the spatial distribution characteristics of forest CNC was mainly affected by the spatial variation of forest plant functional type.The results of this study can support for optimizing the simulation of forest productivity at the national scale.