北京一号小卫星NDVI时间序列重建方法研究

Study on NDVI Time Series Reconstruction Algorithm of Beijing-1 Micro-satellite

利用目前使用较广泛的Savitzky-Golay滤波拟合法和傅立叶谐波分析法对NDVI时间序列进行滤波处理,分析植被指数时间序列滤波重建方法对北京一号小卫星数据的适用性。试验结果表明,较之Savitzky-Golay方法,傅立叶谐波改进算法的重建结果更能体现地物的物候变化规律,且对原始数据的扰动较小,更有利于基于植被指数时间序列的土地覆盖分类及作物估产等定量应用,比较适合于北京一号小卫星NDVI时间序列的重建。

Due to the cloud effect and atmosphere conditions and sensors,there is serious noise in the NDVI time series.It is necessary to filter effectively to remove the residual noise and reconstruct high-quality NDVI time series.This study compares the widely-used Savitzky-Golay fitting algorithm and Fourier harmonic analysis algorithm for Beijing-1 micro-satellite,based on NDVI time series reconstruction.It turns out that the result of improved Fourier harmonic analysis algorithm can reflect better the phenological rules of all the land cover types.In addition,it remains less disturbance to the original data values.Therefore,the NDVI time series reconstructed by the improved Fourier harmonic analysis algorithm is more useful for land cover classification and crop yield prediction and other quantitative applications.The improved Fourier harmonic analysis is found more applicable for Beijing-1 micro-satellite derived NDVI time series reconstruction.