Differentiable simulation of the Earth's atmosphere
Event details
| Date | 25.08.2022 |
| Hour | 14:00 › 16:00 |
| Speaker | Miguel Crespo Castano |
| Location | |
| Category | Conferences - Seminars |
EDIC candidacy exam
Exam president: Prof. Mark Pauly
Thesis advisor: Prof. Wenzel Jakob
Co-examiner: Prof. Sabine Süsstrunk
Abstract
The retrieval of the physical properties that constitute the atmosphere is a challenging task. Previous work uses 1D models for this purpose, but they can be inaccurate due to the assumptions required. This thesis aims to improve the field by introducing 3D differentiable models that allow a higher accuracy in the retrieved parameters. At the same time, we will design sampling strategies targeted to this specific scenario, improving both the simulation's quality and the computed gradients.
Finally, we will validate our method and try to demonstrate the benefits of the 3D information versus the 1D approximation, potentially targeting future satellite design toward being used for 3D retrievals of the Earth's atmosphere.
Background papers
Exam president: Prof. Mark Pauly
Thesis advisor: Prof. Wenzel Jakob
Co-examiner: Prof. Sabine Süsstrunk
Abstract
The retrieval of the physical properties that constitute the atmosphere is a challenging task. Previous work uses 1D models for this purpose, but they can be inaccurate due to the assumptions required. This thesis aims to improve the field by introducing 3D differentiable models that allow a higher accuracy in the retrieved parameters. At the same time, we will design sampling strategies targeted to this specific scenario, improving both the simulation's quality and the computed gradients.
Finally, we will validate our method and try to demonstrate the benefits of the 3D information versus the 1D approximation, potentially targeting future satellite design toward being used for 3D retrievals of the Earth's atmosphere.
Background papers
- Mayer, B. "Radiative transfer in the cloudy atmosphere." (https://www.epj-conferences.org/articles/epjconf/abs/2009/01/epjconf1008/epjconf1008.html)
- Buras, Robert, and Bernhard Mayer. "Efficient unbiased variance reduction techniques for Monte Carlo simulations of radiative transfer in cloudy atmospheres: The solution." (https://www.sciencedirect.com/science/article/pii/S0022407310003791)
- Müller, Thomas, Markus Gross, and Jan Novák. "Practical path guiding for efficient light‐transport simulation." (https://onlinelibrary.wiley.com/doi/full/10.1111/cgf.13227)
Practical information
- General public
- Free