guest ::
| Home > Publications database > Leveraging a large-scale radiative transfer simulation for an emulator based retrieval scheme of sun-induced fluorescence in HyPlant imagery |
| Home > Publications database > Leveraging a large-scale radiative transfer simulation for an emulator based retrieval scheme of sun-induced fluorescence in HyPlant imagery |
| Conference Presentation (After Call) | FZJ-2024-06511 |
; ; ; ; ; ; ; ; ;
2024
This record in other databases:
Please use a persistent id in citations: doi:10.34734/FZJ-2024-06511
Abstract: The prediction of sun-induced fluorescence (SIF) from hyperspectral radiance has been identified as a corner stone to assess plants’ photosynthetic efficiency remotely. It is widely accepted that remotely sensed SIF offers great potential for a variety of applications. To provide such estimates, top-of-canopy SIF products derived from passively sensed radiance measurements of various airborne and spaceborne sensors have been developed over the last decades. To date, however, physically based SIF retrieval schemes require a prohibitive use of computationally costly radiative transfer simulations especially when used in complex observational conditions such as in hilly terrain. In this contribution we report on our on-going work to develop a lightweight self-supervised neural network to retrieve SIF in the O$_2$-A absorption band of HyPlant acquisitions. We aim at a tight integration of a physical radiative transfer model with the network to ensure physically sound predictions by leveraging large scale simulation and emulation of HyPlant at-sensor radiance observations. We report on first results that we achieve on a dedicated data set.
; 
|
The record appears in these collections: |
PDF