摘要
利用云南省西双版纳的Hyperion高光谱图像,利用多元逐步线性回归建立了Hyperion一阶导数反射率与氮浓度和叶绿素浓度的关系,结果表明:经6S模型大气校正的Hyperion反射率与野外实测冠层反射率基本吻合;经6S校正的反射率计算的NDVI,高于用绝对亮度、表观反射率计算的NDVI,而且前者与野外实测计算的NDVI最接近;预测氮和叶绿素浓度的模型中大部分入选波长与蛋白质的吸收有关,R2分别为0.586和0.506。产生了冠层水平氮和叶绿素浓度的空间分布。结果表明:水稻的氮浓度最高,为2.5%-3.5%,其次为甘蔗、土豆、茶树,氮浓度为1.0%-2.5%,而大多数森林的氮浓度在1.0%-1.5%。对于叶绿素,水稻、马铃薯的叶绿素浓度最高,为25%-35%,其次为玉米、甘蔗,叶绿素浓度为20%-30%,而栗树的叶绿素浓度为20%-25%。证明高光谱图像是大尺度估算植被生化组分的有效方式。
Using Hyperion hyperspectral image acquired over Xishuangbanna in Yunnan Province, China, the relationships between first-derivative reflectance of Hyperion and nitrogen and chlorophyll concentration were established using multivariable stepwise linear regression. Results show that Hyperion reflectance after 6S atmospheric correction is consistent with field-measured canopy reflectance, NDVI computed from reflectance after 6S correction is higher than those from absolute radiance and apparent reflectance, and the former is closest to that from field measured. Most of the selected wavelengths in models predicting nitrogen and chlorophyll are related to absorption of protein, R2 are 0. 586 and 0. 506, respectively. Spatial distribution of nitrogen and chlorophyll concentration at canopy level is produced, the results show that nitrogen concentration of rice is the highest, and it is 2.5 %- 3.5 % ; the next are most crops with value of 1.0%- 2.5%, and nitrogen concentration of forest is 1.0%-1.50%. As for chlorophyll, rice and potato have the highest chlorophyll concentration, the value is 25%-35% the next are corn and sugar cane with value of 20%-30%, and chlorophyll concentration of chestnut is 20%-25%. This demonstrates that hyperspectral image is an effective way to estimate biochemical components of vegetation at large scale.
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2007年第4期172-178,共7页
Transactions of the Chinese Society of Agricultural Engineering
基金
国家自然科学基金(40571117)
中科院知识创新项目(KZCX3-SW-338)
