Category: Time series and analysis
Data Fusion
Data fusion combines multiple data sources or sensors to create a more complete or consistent product.
Also known as: multi-sensor fusion
Expanded definition
Data fusion uses complementary information from different sensors or datasets. A common example is combining optical imagery with SAR so monitoring can continue during cloudy periods.
Fusion can happen at different levels: pixel level (combining measurements), feature level (combining derived metrics), or decision level (combining model outputs). Each level has different tradeoffs in interpretability and error propagation.
Fusion can improve coverage and stability, but it introduces assumptions. It is important to understand what is observed directly versus inferred, and how sensor differences were harmonized before combining.
Related terms
SAR (Synthetic Aperture Radar)
SAR is an active radar sensor that works day or night and can see through clouds, measuring surface scattering rather than reflected sunlight.
Optical Imagery
Optical imagery measures reflected sunlight, providing rich spectral information but being sensitive to clouds and illumination.
Gap Filling
Gap filling estimates missing values in time series caused by clouds, shadows, or data dropouts.
Harmonization
Harmonization reduces differences between scenes or sensors so values are more comparable across time.