Universität zu Köln

Burchart, Yannick ORCID: 0000-0001-6798-2596 (2025). From Simulation to Observation: A Path-Tracing Approach Using Stereo Cameras and Large-Eddy Simulations for Cloud Morphology Validation. PhD thesis, Universität zu Köln.

PDF Thesis_Printversion_without_signature.pdf - Accepted Version Bereitstellung unter der CC-Lizenz: Creative Commons Attribution. Download (70MB)

Abstract

Clouds cover a substantial portion of the Earth’s surface, playing a crucial role in regulating the planet’s climate. Clouds impact global radiative transfer and the transport of heat, moisture, and momentum across various spatial and temporal scales. Despite their importance, clouds remain a major source of uncertainty in global warming projections and play a critical role in both the hydrological cycle and photovoltaic energy production. Thus, improving cloud representation in weather and climate models is essential for advancing climate predictions and deepening our understanding of their complex atmospheric interactions. Conventional meteorological instruments face challenges in capturing the intricate spatio-temporal dynamics of cumulus clouds. Groundbased observations provide high-resolution data but are limited in spatial coverage, while satellite imagery offers broader coverage at the expense of the fine resolution needed to resolve smaller-scale, shallow cumulus clouds. High-resolution hemispheric camera networks present a promising alternative, delivering multidimensional insights into cloud dynamics and addressing these observational gaps. This thesis presents a novel approach to bridging the gap between Large-Eddy Simulations (LES) and real-world cloud observations by developing a stereo camera simulation framework. At the heart of this research is a stereo camera simulator that leverages LES-generated liquid water fields to emulate 3D cumulus cloud observations captured by stereo camera systems. Using path-tracing rendering techniques, the simulator generates synthetic cloud images that are reconstructed to replicate actual camera observations, incorporating realistic limitations and potential errors inherent to real-world data. This innovative approach provides a robust basis for direct comparison between simulated and observed cloud fields. The simulator’s performance and limitations are evaluated in three key studies. The first study introduces the simulator, assessing its ability to reconstruct cloud boundaries from virtual stereo cameras and examining its sensitivity to rendering parameters, including render samples, optical thickness, scattering properties, and sun angles. The second study compares simulator outputs with actual stereo camera observations, focusing on horizontal cloud size distributions and the characteristic powerlaw behavior. A significant finding is the underrepresentation of large clouds due to occlusion by smaller clouds, leading to lower power-lawexponents for cloud size distributions when compared to direct LES outputs in contrast to using the stereo camera simulator. The third study develops a cloud-tracking algorithm that enables temporal tracking, generating statistics on cloud evolution over time. This approach facilitates the comparison of cloud lifetime distributions, showing an overall power-law exponent close to -5/3 and exhibits longer living LES clouds compared to their observational counterparts. This research demonstrates the potential of stereo camera systems to advance cumulus cloud observational and modeling. The simulator facilitates the optimization of camera network configurations before field deployment and enhances LES accuracy by enabling more precise comparisons of real cloud dynamics over space and time. These contributions significantly advance cloud photogrammetry and atmospheric science, providing a foundation for improved LES parameterizations, especially for high-resolution models operating in the turbulent grey zone. Ultimately, deepening our understanding of cumulus cloud evolution and offering valuable insights for advancing weather and climate modeling.

Item Type:	Thesis (PhD thesis)
Creators:	Creators Email ORCID ORCID Put Code Burchart, Yannick yannick.burchart@outlook.de orcid.org/0000-0001-6798-2596 UNSPECIFIED
URN:	urn:nbn:de:hbz:38-782549
Date:	2025
Language:	English
Faculty:	Faculty of Mathematics and Natural Sciences
Divisions:	Faculty of Mathematics and Natural Sciences > Department of Geosciences > Institute for Geophysics and Meteorology
Subjects:	Generalities, Science Data processing Computer science General statistics Natural sciences and mathematics Physics Earth sciences Technology (Applied sciences)
Uncontrolled Keywords:	Keywords Language shallow cumulus clouds English virtual camera image simulator English cloud photogrammetry English large-eddy simulation English cloud modeling English ray-tracing English cloud size distributions English cumulus cloud statistics English cloud tracking English
Date of oral exam:	12 March 2025
Referee:	Name Academic Title Neggers, Roel Prof. Dr. Löhnert, Ulrich Prof. Dr.
Refereed:	Yes
URI:	http://kups.ub.uni-koeln.de/id/eprint/78254