Integrated spatio-temporal DINEOF reconstruction under extreme cloud-induced data loss: application to MODIS chlorophyll-a in the eastern Indian Ocean

Persistent cloud cover severely limits the effective use of satellite-derived chlorophyll-a (Chl-a) products for marine surveillance in tropical oceans due to extensive spatial and temporal data gaps. These sampling limitations hinder robust quantification of phytoplankton variability and long-term environmental change. This study develops an integrated spatio-temporal reconstruction framework based on Data Interpolating Empirical Orthogonal Functions (DINEOF), implemented within a native three-dimensional raster time-series environment that leverages the sinkr and rtsa packages to reconstruct cloud-affected satellite Chl-a observations. Reconstruction performance was systematically evaluated under extreme missing-data scenarios (up to 80%) across temporal (1D), spatial (2D), and full spatio-temporal (3D) analytical dimensions by multiple statistical metrics, including the correlation coefficient (r) and root mean square error (RMSE). Results indicate that dominant Chl-a signals remain statistically recoverable despite severe data loss, yielding temporal correlations of r = 0.706–0.792, very strong spatial agreement (r = 0.944–0.955), and robust spatio-temporal reconstruction performance (r = 0.964). These results demonstrate that integrated spatio-temporal DINEOF reconstruction can substantially mitigate cloud-induced sampling bias, thereby enhancing the continuity and reliability of satellite-derived Chl-a time series for marine environmental monitoring and quantitative ecosystem analysis of phytoplankton dynamics associated with monsoon variability and large-scale oceanographic processes in tropical regions.