基于线性内插的扩展卡尔曼滤波法的NDVI时间序列重构研究

发布时间：2018-06-01 20:40

本文选题：NDVI + 线性内插　；参考：《科技通报》2017年02期

【摘要】：由于归一化植被指数(NDVI)时间序列数据含有大量的噪声,对其应用带来诸多不便。为了提高NDVI数据质量,本文采用线性内插的扩展卡尔曼滤波(EKF)法对广州市森林地区的NDVI时间序列数据进行了重构,并与EKF和中值滤波方法进行比较。利用部分样点的实测数据与重构后的NDVI值进行比较,得到基于线性内插的EKF、EKF和中值滤波三种方法的相对误差分别在-1.91%~0.93%,-3.86%~5.85%和-0.28%~16.30%之间。结果表明:基于线性内插的EKF算法的时间序列重构方法重构后的NDVI时间序列能够更好的逼近高质量的数据,拟合原始曲线的波峰,在提升曲线的整体效果的同时,降低原始数据的均值偏差和数据的离散程度,对低值噪声的抑制能力更好。通过该重构方法重构后的较高质量的NDVI时间序列数据为森林监测、生态保护以及建设提供了良好的基础。
[Abstract]:Because the normalized vegetation index (NDVI) time series data contain a lot of noise, it is inconvenient for its application. In order to improve the quality of NDVI data, this paper uses linear interpolation extended Kalman filter (EKF) method to reconstruct the NDVI time series data in Guangzhou forest area, and compares them with EKF and median filtering methods. By comparing the measured data of some samples with the reconstructed NDVI values, the relative errors of the three methods based on linear interpolation EKF and median filtering are found to be between -1.91 and 0.93% -3.86%, 5.85% and -0.28%, respectively. The results show that the reconstructed NDVI time series based on linear interpolation EKF algorithm can better approximate the high quality data, fit the peak of the original curve, and improve the overall effect of the curve. It is better to reduce the mean deviation of the original data and the degree of data dispersion to suppress the low value noise. The reconstructed NDVI time series data provide a good basis for forest monitoring, ecological protection and construction.
【作者单位】：华南农业大学资源环境学院/广东省土地利用与整治重点实验室/广东省土地利用与整治重点实验室;广东省土地信息工程技术研究中心/广州市测绘地理信息行业工程中心;美国地质调查局陆地资源观测科学中心;
【基金】：国家自然科学基金项目(U1301253) 广东省科技计划项目(2014A050503060)
【分类号】：Q948;TN713

【相似文献】