利用经验模态分解的近地面植被指数时序数据重构研究

Analysis of NDVI Time Series Data Reconstruction Based on Empirical Mode Decomposition Method

NDVI遥感产品可以较好地反映植被的生长情况,已被广泛应用于植被物候的遥感分析。本文利用PhenoCam物候相机观测网络林地站点的近地面多光谱数码相机照片,通过近红外和红光波段组合提取近似NDVI的时序数据,并在景观尺度上采用经验模态分解方法(EMD,Empirical Mode Decomposition)实现一定容差范围内NDVI时序数据的重构。结果表明:EMD方法能够有效地减少NDVI时序数据噪声干扰,经验证,趋势项残差的趋势较为明确。实验基于重构结果进行植被物候分析,得到的物候分布特征与植被实际生长反映的变化特征基本一致。

The NDVI (Normalized Difference Vegetation Index) has practical significance for reflecting vegetation status. It has been widely used in the field of remote sensing based phonological analysis. In this paper, the NDVI -like time series data is extracted through the combination of near-infrared and red light bands based on woodland type near-ground multi-spectral digital camera photos. And the NDVI time series data was reconstructed on the landscape scale using EMD method within a certain tolerance to effectively reduce the noise vibration. Disturbance, the trend of confirmed trend item is relatively clear. Vegetation phenology analysis is performed based on its reconstruction results. Its phenological distribution characteristics are basically the same as those reflected by the actual growth of vegetation.