Ocean color remote sensing provides continuous observations, with the Geostationary Ocean Color Imager-II (GOCI-II) offering high temporal resolution data for the Korean Peninsula. However, cloud cover frequently creates data gaps, compromising the re...
Ocean color remote sensing provides continuous observations, with the Geostationary Ocean Color Imager-II (GOCI-II) offering high temporal resolution data for the Korean Peninsula. However, cloud cover frequently creates data gaps, compromising the reliability of chlorophyll-a measurements— a critical marine ecosystem indicator. This study proposes the Spatiotemporal Attention Partial Convolution model to reconstruct missing values in multi-day satellite imagery. The model extends partial convolution to three dimensions, enabling simultaneous exploration of spatiotemporal dynamics. A 3D attention mechanism guides the network to focus on informative features. To ensure robust training, we developed a cloud mask generation strategy using Gaussian Random Fields with parameterized indices, creating diverse scenarios exceeding observational variability. Evaluation using 2024 GOCI-II masks demonstrated superior performance, achieving mean absolute error of 0.036 mg/m³, root mean squared error of 0.068 mg/m³, and R² of 0.865. Consistency analysis across ten regions showed substantial improvements, with valid pixels increasing from 54% to 97%. Qualitative assessments confirmed that reconstructed images preserve natural spatial continuity without artifacts, significantly enhancing the utility of GOCI-II data for marine environmental monitoring.