Abstract: Observation by satellite-based remote sensing is an important way of study the variation of marine environment. However， raw remote sensing data from some observatories is often missed due to the impacts of limited observation period or weather etc.， which defects in the study of consecutive variations of marine environment. To figure it out， Data Interpolation Empirical Orthogonal Functions （DINEOF） is used to reconstruct the missed remote sensing data. Firstly， based on the combined and pixel-averaged level 3 chlorophyll-a production of SeaWiFS （1998.01～2010.12）， MODISAqua （2002.07～2014.12） and MODIS-Terra （2000.02～2014.12）， the missed monthly remote sensing chlorophyll-a concentration data on Taiwan Strait are reconstructed and examined in tempo-spatial deviation （error） and variability. The results show that the reconstructed chlorophyll-a concentration data could basically reveal its tempo-spatial variation features on Taiwan Strait. The study also indicates that the reconstruction approach could be easily operated without prior knowledge in reconstructing the high accuracy massive visual data， thus providing a sound basis for the exploration of long-term variations of marine environment and ecology．