期刊论文

中文

中南林业科技大学学报

1673-923X

2018

38

4

10-15

10.14067/j.cnki.1673-923x.2018.04.003

林业公益性行业科研专项（201404710）；

本校为第一机构

针对MODIS 16 d合成的N_(dvi)时间序列数据的重构问题,以厦门岛为实验区,以林地为实验类型,基于双Logistic函数拟合法、非对称高斯函数拟合法和Savitzky-Golay滤波法,进行区域时间序列N_(dvi)数据的重构措施及验证研究。结果表明,3种重建方法均可以整体提高时序N_(dvi)平均值,有效地避免异常值及噪声的干扰,A-G函数与D-L函数拟合法重建的时序N_(dvi)更能反映林地成长的曲线特征;S-G滤波法和A-G函数拟合法在保持高质量N_(dvi)真实值特征方面比D-L法效果较好;在生长季峰期内,S-G滤波法拟合效果要优于A-G函数和D-L函数拟合算法,更能反映林地成长的最佳情况。

For the reconstruction problem of MODIS 16d synthetic time-series of N_(dvi) data, we carried out the researches on reconstruction method of regional N_(dvi) time-series data and validation studies with the experimental area of Xiamen Island, which was based on double Logistic function fitting, symmetric Gaussian fitting and Savitzky-Golay filter. The results showed that the three reconstruction methods can improve the overall time-series N_(dvi) average value, effectively avoid noise interference and outliers, A-G and D-L fitting reconstruction of N_(dvi) time-series could reflect the charact...