基于 HJ-1A 卫星数据的高寒草地氮素评估－以青海省贵南县及玛沁县高寒草地为例

摘要

基于 HJ-1A HSI 卫星遥感资料，结合2014年青海省玛沁县、贵南县研究区高寒草地野外实测数据，建立了高寒草地氮素含量的估测模型，并对模型进行了精度评价，筛选出最优反演模型，分析了研究区高寒草地氮素的空间分布。结果表明，1）高寒草地 HSI 的原始光谱反射率、一阶导数光谱反射率及去包络线光谱反射率与地面实测氮素含量有一定的相关性，特别是吸收特征波段1（750．95～791．95 nm）和吸收特征波段3（889．03～921．30 nm）与氮素含量具有显著的相关性；2）基于吸收特征波段1构建的波段深度指数 BD767．99可以较好的估测研究区高寒草地氮素含量，利用此变量拟合的线性回归模型可以解释研究区高寒草地氮素含量变化的44％，模型估测精度可达81．6％；3）贵南研究区氮素含量的空间变异较大，整体而言，西北部和西南部地区氮素含量较高，东北部地区氮素含量较低；玛沁研究区氮素含量的空间变异较小，草地氮素含量水平整体较低。%Using data sourced from a hyper-spectrum imager (HSI)from an environmental mitigation satellite (HJ-1A)and ground observation during 2014 in Maqin and Guinan Counties,Qinghai Province,inversion mod-els were established for estimating the nitrogen (N)content of alpine grassland.The optimal inversion model was selected and the spatial distribution map of grassland N was analyzed in the study area.The results showed that there was a correlation between N and the original spectral reflectance,the first-order differential spectral reflectance and the continuum removed spectral reflectance of the original hyper-spectral image,and significant correlations between grassland N content and absorption feature band one (750.95 -791.95 nm)and absorp-tion feature band three (889.03-921.3 nm)were also found.The band depth index BD767.99 was able to es-timate N content with the greatest accuracy;the linear model was able to explain over 44% of the variation in N content.The spatial variation of N content in Guinan County was large;generally N content was higher in northwest and southwest than in the northeast.The spatial variation of grassland N content in Maqin County was low and grassland N content was also low.