一种基于无人机高光谱数据的植被盖度估算新方法

发布时间：2018-05-01 11:31

本文选题：植被覆盖度 + 无人机　；参考：《光谱学与光谱分析》2017年11期

【摘要】：从分析对植被覆盖度(FVC)敏感的光谱特性入手,使用Avafield-3光谱仪(测量范围300~2 500nm),利用人工草坪控制植被覆盖度的方式研究混合光谱与植被覆盖度的关系,通过实验发现红边区间(680~760nm)对植被覆盖度最为敏感,而红边区间光谱的一阶导数与植被覆盖度的相关性最高(0.98),且有较强的稳定性,因此选择红边斜率k作为估算植被盖度的参数。参考混合光谱分解法反演植被覆盖度的经典模型——即以NDVI(normalized difference vegetation index)为参数的植被覆盖度反演模型,以红边斜率代替NDVI构建了2个反演植被覆盖度FVC的新的红边斜率模型,该模型是对经典模型的进一步改进。为验证模型精度,以研究区内无人机(UVA)的高光谱数据和研究区实际测量的植被覆盖度数据进行验证:对高光谱数据计算每个像元680~760nm之间的斜率,利用PPI(pixel purity index)提取纯像元,计算纯植被像元光谱斜率的最大值和纯土壤像元光谱斜率最小值,利用新的红边斜率FVC模型求取植被覆盖度;实测数据采用照相方法,经过几何校正、监督分类后统计植被覆盖度,结果表明:通过实测数据与无人机高光谱数据获取的植被覆盖数据进行验证,新构建的基于红边斜率的两个植被覆盖度模型的精度(R2分别达0.893 3和0.892 7)都略高于以NDVI为参数的模型(R2分别达0.839 9和0.829 9)。提出使用红边斜率计算植被覆盖度的模型,具有较明确的生物物理意义,具有较高的应用潜力和推广价值。
[Abstract]:Based on the analysis of the spectral characteristics sensitive to vegetation coverage, the relationship between mixed spectrum and vegetation coverage was studied by using Avafield-3 spectrometer (measuring range 300 ~ 2 500 nm) and using artificial lawn to control vegetation coverage. It is found that the red edge interval is most sensitive to vegetation coverage, and the correlation between the first derivative of the red edge spectrum and the vegetation coverage is the highest (0.98) and has strong stability. Therefore, the red edge slope k is chosen as the parameter to estimate the vegetation coverage. Referring to the classical model of vegetation coverage inversion using mixed spectral decomposition method, that is, the vegetation coverage inversion model with NDVI(normalized difference vegetation index) as the parameter, two new red edge slope models for retrieving vegetation coverage FVC are constructed by using red edge slope instead of NDVI. The model is a further improvement on the classical model. In order to verify the accuracy of the model, the hyperspectral data of UAV in the study area and the vegetation coverage data measured in the study area were verified. The slope between each pixel 680~760nm was calculated from the hyperspectral data, and the pure pixel was extracted by using PPI(pixel purity indexs. The maximum value of spectral slope of pure vegetation pixel and the minimum value of spectral slope of pure soil pixel are calculated, and the vegetation coverage is obtained by using a new red edge slope FVC model. The results show that the vegetation coverage is verified by the measured data and the data obtained from the UAV hyperspectral data. The accuracy of the two vegetation coverage models based on the red slope is 0.893 3 and 0.892 7 respectively), which is slightly higher than that of the NDVI model (0.839 9 and 0.829 9). A model for calculating vegetation coverage with red edge slope is presented, which has clear biophysical significance and has high application potential and popularization value.
【作者单位】：北京林业大学精准林业北京市重点实验室;山东交通学院;
【基金】：国家自然科学基金项目(41371001) 国家高技术研究发展计划(863)重大项目(2008AA121806)资助
【分类号】：Q948;TP751

【相似文献】