Earth observation sensor fusion is the combination of observations from different sensors or missions so that they can be used together as a single, more consistent source of information about the surface.12
An ESA article from 2012 says that combining data of different resolutions, acquisition times and conditions is difficult. The project it describes, EO-LDAS, simulates a virtual sensor from many sensors and uses assimilation to harmonise the data, so that datasets from different instruments become comparable, and the system is also to be used for cross-calibration of sensors.1
NASA’s Harmonized Landsat and Sentinel-2 project is a second example. It ingests data from Landsat 8 and 9 and from the Sentinel-2 satellites and applies radiometric and geometric adjustments, with the stated aim of removing differences in retrieved surface reflectance that arise only from differences in instrumentation.2 The project reports revisit intervals of 1.6 days in 2022 and 1.4 days in 2025 for the combined constellation.2
Both sources concern optical land observation and harmonisation or assimilation. They do not establish methods for combining optical with radar or with other sensor types, so those uses are outside what is supported here.
Footnotes
-
ESA, “Integrating Earth observation data”, 21 November 2012. https://www.esa.int/Applications/Observing_the_Earth/Integrating_Earth_observation_data ↩ ↩2
-
NASA Earthdata, “Harmonized Landsat and Sentinel-2”. https://www.earthdata.nasa.gov/data/projects/hls ↩ ↩2 ↩3