B1703
Title: Creating daily, fine spatial resolution Sentinel-2 time-series
Authors: Pete Atkinson - Lancaster University (United Kingdom)
Wang Qunming - Lancaster University (United Kingdom) [presenting]
Wenzhong Shi - HKPU (China)
Abstract: Image fusion is a common problem in the field of remote sensing and there are many different approaches for this. Space-time (ST) image fusion is potentially of novel benefit for fusing Sentinel-2 and Sentinel-3 images. These two sensors have very different spatial and temporal resolutions (Sentinel-2 MSI sensor 10 m, 20 m and 60 m, 10 days, albeit 5 days with 2 sensors; Sentinel-3 OLCI sensor 300 m, less than two days, conditional upon clear skies). The problem is that for local monitoring one either has the desired spatial (e.g. 20 m) or temporal (e.g. nearly daily) resolution, but not both. While the OLCI instrument is suitable for monitoring over large areas, it is not suitable for monitoring at the local scale. Area-to-point regression kriging (ATPRK) was, therefore, used to provide a very fine spatial resolution, frequent time-series of Sentinel-based images for the first time. The effectiveness of ATPRK-based ST fusion is demonstrated with examples. The approach can be applied globally, offering the prospect, given the appropriate infrastructure, of a global fine resolution Sentinel product for local applications.
Journal