Remote sensing data imputation using deep learning for multispectral imagery
Researchers have developed deep learning models to fill in missing data in multispectral satellite imagery, a common issue caused by cloud cover. These models, including CNN and CNN-LSTM architectures, significantly outperformed traditional linear interpolation in reconstructing spectral bands for four lakes with historical algal bloom data. The study found that CNN models were particularly effective, enabling more reliable water monitoring applications by improving the completeness of satellite datasets. AI
IMPACT Enhances the reliability of satellite data for environmental monitoring, potentially improving early detection of events like algal blooms.