| Title | Blending of Landsat / Sentinel-2 and MODIS images for production of synthetic image time series - the method, result and software |
| |
| Author | Zhong, D ; Zhou,
F |
| Source | Program, 41st Canadian Symposium on Remote Sensing/Programme, 41e Symposium canadien de télédétection; 2020 p. 79  Open Access |
| Links | Online - En ligne (complete
volume - volume complet, PDF, 17.5 MB)
|
| Links | Software (free) - Logiciel (gratuit)
|
| Image |  |
| Year | 2020 |
| Alt Series | Natural Resources Canada, Contribution Series 20210415 |
| Publisher | Canadian Remote Sensing Society |
| Meeting | 41st Canadian Symposium on Remote Sensing / 41e Symposium canadien de télédétection; July 13-16, 2020 |
| Document | book |
| Lang. | English |
| Media | digital; on-line |
| File format | pdf |
| Subjects | geophysics; Science and Technology; remote sensing; satellite imagery; software; models; Landsat; Sentinel-2; MODIS; Prediction Smooth Reflectance Fusion Model; Kalman Filter Reflectance Fusion Model;
Methodology |
| Program | Canada Centre
for Remote Sensing (CCRS) |
| Released | 2020 07 01 |
| Abstract | Although Earth observations, such as Landsat, and Sentinel-2 images, provide significant information for land surface studies, the time-series datasets that they have produced are short for certain land
surface monitoring and dynamic studies due to clouds contamination and/or satellite's low revisit frequency. At Canada Centre for Remote Sensing, we have developed two image fusion models to blend medium spatial resolution images (Landsat and
Sentinel-2) with coarse resolution MODIS images to generate dense cloud-free time-series synthetic Landsat or Sentinel-2 images. The first is Prediction Smooth Reflectance Fusion Model (PSRFM) (Zhong and Zhou 2018, 2019). The second is Kalman Filter
Reflectance Fusion Model (KFRFM) (Zhou and Zhong 2020). Evaluations of both models' outputs against observations and comparisons to some well-known image fusion models, such as Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM) (Gao et
al. 2006), Enhanced version of STARFM (ESTARFM) (Zhu et al. 2010) and the Flexible Spatiotemporal Data Fusion (FSDAF) model (Zhu et al. 2016), indicated that the newly developed cutting-edge models are among the best ones. Both models are also
implemented in C++ in one software package which is multi-core enabled for high computation efficiency and can be run in a batch mode with a simple parameter control file.
In this presentation, we will discuss some key challenges in image fusion
development, and our innovative solutions. We will also introduce the models' capabilities and some of their promising applications, as well as the software package, which is open and free downloadable from https://github.com/dzhong-hub/PSRFM.
|
| Summary | (Plain Language Summary, not published)
Although Earth observations, such as Landsat, and Sentinel-2 images, provide significant information for land surface studies, the time-series datasets
that they have produced are short for certain land surface monitoring and dynamic studies due to clouds contamination and/or satellites' low revisit frequency. The Canada Centre for Remote Sensing has developed two image fusion models to blend medium
spatial resolution images (Landsat and Sentinel-2) with coarse resolution MODIS images to generate dense cloud-free time-series synthetic Landsat or Sentinel-2 images. |
| GEOSCAN ID | 329260 |
|
|