2014,50(22)
数据融合技术在提高 NPP 估算精度中的应用
黄登成
1,2
,张 丽
2
,尹晓利
2,3
,王 昆
2,3
HU ANG Dengcheng
1,2
, ZHANG Li
2
, YIN Xiaoli
2,3
, WAN G Kun
2,3
1.辽宁工程技术大学 测绘与地理科学学院,辽宁 阜新 123000
2.中国科学院 数字地球重点实验室,中国科学院遥感与数字地球研究所,北京 100094
3.山东科技大学 测绘科学与工程学院,山东 青岛 266590
1.Institute of Surveying and Mapping, and Geographic Science, Liaoning Technical University, Fuxin, Liaoning 12 3000, China
2.Key Labor atory of Di gital Earth, I nstitute of Remote Sensing and Digital Earth, Chinese Academy of Scien ces, Beijing
100094, China
3.College of Geoma tics, Shandong University of Science and Technology, Qingdao, Shandong 266590, China
HUANG Dengcheng, ZHANG Li, YIN Xiaoli, et al. Appli cation o f image fusion in improving NPP estimation accu-
racy. Computer Engineering and Applications, 2014, 50(22):193-198.
Abstract:The current remote sensing data can not simultaneously satisfy th e precise monit oring of vegeta tion productivity
changes in both high temporal and spatial reso lutions. In this s tudy, application of an image fusio n method to an ecosystem
model for improving the accuracy of NPP evaluations is proposed. Fir stly, the Spatial and Te mporal Adaptive Reflectance
Fusion Model(STARFM)is applied to get higher temporal and spatial resolution NDVI d ata(30 m)from the MODIS-NDVI
and TM-NDVI images and then multi-scale Net Primary Productivity(NPP)of Xi linhot grasslands are estimated b ased on
the CAS A model using different scales of MODIS-NDVI data and the 3 0 m fusion data . The results indicate that the corre-
lation between the model-estimated NPP and the measured aboveground biomass i s gradually increased with t he improve-
ment of the resolution of the input NDVI data. The max correlation coefficient(r)reached 0.915. Additionally, the coe ffi-
cient between the NPP estimations derived from fusion NDVI data and the observed biomass is higher t han the coefficient
of non-fusion image. The results also indicate that the accuracy of NPP estimations from fusion NDVI data is bet ter t han
non-fusion NDVI data and t he fusion NDVI image as the model input da ta can improve the accuracy of NPP estimati ons.
Key words:data fusion; Spatial and Temporal Adaptive Reflectan ce Fusio n Model(STARFM); CAS A model; Net Primary
Productivity(NPP)
摘 要:针对现有遥感数据不能同时满足在时间和空间上精确监测植被动态变化的问题,提出利用时空适应性反射率
融合模型(STARFM)的方法对 MODIS-NDVI和 TM-NDVI影像数据进行融合处理获得 30 m较高时空分辨率的融合
NDVI影像,进而将多种尺度的 MOD IS-NDVI和融合 NDVI数据分别输入到 CASA 模型,对锡林浩特地区进行植被净
初级生产力(NPP)的多尺度估算。将不同尺度的 NPP估算结果与地上生物量地面实测值进行验证比较,结果表明:随
着输入 NDVI空间分辨率的提高,NPP估算值与实测地上生物量之间的相关性也逐渐增大,
r
最大值达到了 0.915。此
外以融合 NDVI影像作为输入数据之一的 NPP估算值与实测地上生物量的相关性均比未融合 NDVI 的相关性高,说
明融合 NDVI估算 NPP 的效果较未融合 NDVI好,并且以融合 NDVI 影像作为模型输入数据可提高 NPP估算精度。
关键词:数据融合;时空适应性反射率融合模型;CASA 模型;净初级生产力
文献标志码:A 中图分类号:TP391 doi:10.3778/j.issn.1002-8331.1301-0034
基金项目:国家科技支撑计划课题(No.2012BAH27B05);中国科学院对地观测与数字地球科学中心主任创新基金(No.Y2ZZ19101B)。
作者简介:黄登成(1987—),男,硕士,研究领域为植被遥感;张丽(1975—),通讯作者,女,博士,副研究员,研究领域为植被遥感。
E-mail:lizhang@ceode.ac.cn
收稿日期:2013-0 1-06 修回日期:2013-0 3-15 文章编号:1002-8 331(2014)22-0193-06
CNKI网络优先出版:2013-04-18, http://www.cnki.net/kcms/detail/11.2127.TP.201304 18.1618 .016.html
C omputer Engineering a nd Applications 计算机工程与应用
193