结合Sentinel-1B和Landsat8数据的针叶林叶片含水量反演研究

Conifer Leaf Water Content Retrieval Based on Sentinel-1B and Landsat8 Data

为研究SAR影像结合光学影像反演叶片含水量的可行性,本文以吉林省长春市净月潭国家森林公园为研究区,以Sentinel-1B、Landsat8 OLI遥感影像和通过外业调查获取的叶片含水量为数据源,通过相关性分析,选取出与叶片含水量相关性较大的波段组合和植被指数,并对其进行主成分提取,建立主成分与叶片含水量之间的线性、二次多项式、三次多项式和指数模型,并利用精度最高的模型反演出净月潭国家森林公园的叶片含水量。结果表明:(1) Sentinel-1B的VV极化、VH/VV极化比和OLI传感器的短波红外1波段、短波红外2波段、归一化水分指数(NDWI)、比值植被指数(RVI)与叶片含水量相关性较大;(2) Sentinel-1B和Landsat8 OLI数据结合相较于仅使用Landsat8 OLI数据、提取出的主成分与叶片含水量相关性较高;(3)利用提取出的主成分与叶片含水量建立的反演模型中三次多项式模型的拟合精度最高(R2=0.629 9,RMSE=0.035 8)。表明Sentinel-1B结合Landsat8 OLI数据能较好得反演出针叶林的叶片含水量。

In order to study the feasibility of inverting leaf water content from SAR images combined with optical images, the paper takes the moon lake national forest park in Changchun City, Jilin Province as the study area, and uses Sentinel-IB and Landsat 8 OLI remote sensing images and the leaf water content obtained through field investigation as the data source. Through the correlation analysis, the band combinations and vegetation indices with large correlation with the leaf water content are selected and the principal components are extracted. The linear, quadratic polynomial, cubic polynomial and exponential models between the principal components and the leaf water content are established. Finally, the paper uses the model with the highest precision to reverse the leaf water content of the moon lake national forest park. The results show that： Sentinel-IB W polarization, VH / W polarization ratio and OLI sensor short wave infrared 1 band, short wave infrared 2 band, the normalized difference water index （NDWI）, the ratio vegetation index （RVI） have large correlation with leaf water content. The principal components extracted from the combination of Sentinel-IB and Landsat8 OLI data are more correlated with leaf water content than Landsat8 OLI data alone. The cubic polynomial model in the inversion model established using the extracted principal components and leaf water content has the highest fitting accuracy （/？2=0.6299, It shows that Sentinel-IBcombined with Landsat8 OLI data can better reverse the leaf water content of coniferous forest.