论文部分内容阅读
利用土壤有机质(SOM)高光谱数据和模拟GF-1多光谱影像的波段响应函数生成的宽波段多光谱模拟数据,对比高光谱预处理和构建土壤植被指数,探索模拟GF-1光谱预测SOM的潜力。研究表明,SOM的一阶微分高光谱和模拟GF-1光谱数据构建的土壤指数与SOM的相关性最好。PLSR建模分析表明采用一阶微分高光谱数据可以很好的对SOM进行预,而且模型稳健(R2=0.962,RPD=4.87);模拟GF-1光谱也可以较好的进行SOM的预测,但是模型的稳定性相对较差R2=0.557,RPD=1.43。同时,SOM制图的空间分布表明,采用一阶微分光谱数据和模拟GF-1数据预测得到的SOM含量与实测的SOM表现出相似的空间分布特征。这为采用多光谱数据进行大尺度、大范围的SOM预测提供了基础。
Using the spectral data of soil organic matter (SOM) and the spectral response of broad-spectrum multi-spectral data of simulated GF-1 multispectral imagery, we compared the hyperspectral pretreatment and the construction of soil vegetation index to explore the simulated SOM potential. The results show that the correlation between soil index and SOM is the best among the SOM first-order differential hyperspectral and simulated GF-1 spectral data. PLSR modeling analysis shows that the first-order differential hyperspectral data can be used to predict the SOM well, and the model is robust (R2 = 0.962, RPD = 4.87). Simulating the GF-1 spectrum can also predict the SOM better The stability of the model is relatively poor R2 = 0.557, RPD = 1.43. At the same time, the spatial distribution of SOM mapping shows that the SOM content predicted by the first-order differential spectral data and simulated GF-1 data shows a similar spatial distribution with the measured SOM. This provides the basis for large-scale, large-scale SOM prediction with multispectral data.