A new benchmark for spatiotemporal fusion of Sentinel-2 and Sentinel-3 OLCI images

Resumen

In Earth Observation, the use of multiple sensors has gained considerable attention as a strategy to overcome the limitations inherent in individual datasets. However, the operational lifespans of sensors are finite, creating an ongoing need to explore and integrate new sensors to sustain critical Earth Observation capabilities across diverse applications.

To address this imperative, there is a clear need to establish a novel benchmark dataset featuring the integration of new sensors. In response, we present a new benchmark remote sensing dataset, representing a significant contribution to the existing literature. This curated dataset leverages data from Sentinel-2 and Sentinel-3 OLCI, comprising more than 689 image pairs. It spans a wide range of temporal and spatial variations, capturing diverse landscapes, ecosystems, and weather conditions.

Importantly, the dataset is publicly accessible, facilitating research on the development of more robust data fusion methods. Furthermore, we conduct a comprehensive evaluation of widely used spatiotemporal fusion (STF) methods, providing a detailed quantitative and qualitative comparison as an application of this dataset.
The dataset is freely available at: https://doi.org/10.5281/zenodo.14860220