基于DIEOF方法重构海表叶绿素a遥感缺失数据

Reconstruction of missing remote sensing data of sea surface chlorophyll-a using DIEOF

以2007年1月到2010年12月的MODIS Aqua CHL-a Level 2海表水色产品为基础数据,获得南海北部海域海表叶绿素a浓度的月平均影像集,基于影像集数据的时空相关性利用DIEOF(Data Interpolating Empirical Orthogonal Functions)方法重构其缺失数据。通过分析重构前后数据变化、验证重构结果的时空特征、计算模型精度指标等对重构结果进行评价。研究结果表明:DIEOF方法重构的MODIS海表叶绿素a影像,能够体现研究区海表叶绿素a的时空变化特征,重构结果的复相关系数R2可达到0.98,平均绝对误差MAE小于0.01;该方法重构过程中无需先验信息,易操作,能够有效重构大面积成片缺失或缺失比例较高的影像。

The monthly mean images of surface chlorophyll-a concentration in the Northern South China Sea（NSCS）were derived from the data preprocessing of the MODIS chlorophyll a（CHL-a） concentration Level 2 products（from January 2007 to December 2010）. And then not only does the study implement the reconstruction of missing data in the monthly mean CHL-a image using Data Interpolating Empirical Orthogonal Functions（DIEOF）, based on space-time correlation of data, but also evaluate the reconstructed result by analyzing the difference between data before and after the reconstruction, verifying the temporal and spatial variability, and calculating the precision index. The study shows that the complete MODIS CHL-a images, which are reconstructed using DIEOF, characterize the temporal and spatial variability of CHL-a in NSCS. The multiple correlation coefficient of the reconstruction result is 0.98, and the mean absolute error is less than 0.01. As it allows the calculation of missing data without requiring a prior knowledge about the error covariance structure, the DIEOF can be also successfully applied to the reconstruction of missing data in a large area or in a high proportion.