Master's Thesis Defense: Manon Huguenin

Thu Jun 14 2018 01:00 PM to 03:00 PM
Weber Building - CoVE

You're invited to hear...

Master’s Thesis Proposal


Manon Huguenin

(Advisor: Prof. Dimitri N. Mavris)

"Development and Validation of 3-D Cloud Fields
Using Data Fusion and Machine Learning Techniques"


1 P.M., Thursday, June 14
Weber -  CoVE

Climate change predictions are currently achieved using Global Climate Models (GCMs), which are complex representations of the major climate components and their interactions. These predictions present high levels of uncertainty, which are mostly due to the way clouds are represented in climate models. Cloud-related phenomena, such as cloud-radiative forcing, are represented through physically-motivated parameterizations, which lead to uncertainties in cloud representations. Improving the parameterizations required for representing clouds in GCMs is thus a current focus of climate modeling research efforts. Integrating cloud satellite data into GCMs has been proved essential for achieving this goal. Yet, the availability of satellite data is limited, in particular for vertical data. In order for satellite cloud data to be usefully compared to global representations of clouds in GCMs, additional vertical cloud data has to be generated to provide a more global coverage. Consequently, the overall objective of this thesis is to support the validation of GCMs cloud representations through the generation of 3D cloud fields using cloud vertical data from space-borne sensors.

This has already been attempted by several studies through the implementation of physics-based and similarity-based approaches. However, such studies have a number of limitations, which motivate the need for novel approaches to the generation of 3D cloud fields. For this purpose, efforts have been initiated at ASDL to develop an approach that leverages data fusion and machine learning techniques to generate 3-D cloud field domains. In particular, these efforts have led to the development of a cloud predictive classification model that is based on decision trees and integrates atmospheric data to predict vertical cloud fraction. However, several limitations were identified in this model and the approach that led to it. First, its performance is lower when predicting lower-altitude clouds, and its overall performance could still be greatly improved. Second, the model has only been assessed at “on-track” locations, while the construction of data at “off-track” locations is necessary for generating 3D cloud fields. Last, the model has not been validated in the context of GCMs cloud representation, and no satisfactory level of model accuracy has been determined in this context.

This work aims at overcoming these limitations by taking the following approach. The model obtained from previous efforts will be improved by integrating additional, higher-accuracy data, by investigating the correlation within atmospheric predictors, and by implementing additional classification machine learning techniques, such as Kernel methods and random forests. Then, the model will be evaluated at “off-track” locations against an existing dataset obtained through a physics-based approach as part of the CERES project. This will lead to the generation of a global 3D cloud fields dataset. Last, radiative transfer codes will be implemented on the obtained global vertical profiles, so that the model can be evaluated in the context of GCM cloud-radiative forcing representation. If successful, this research is expected to demonstrate the potential of a machine learning-based approach to generate 3D cloud fields. More importantly, it is expected to contribute to NASA’s ongoing efforts towards improving GCMs and climate change predictions as a whole.


  • Prof. Dimitri Mavris, School of Aerospace Engineering, Georgia Institute of Technology
  • Dr. Olivia Pinon Fischer, School of Aerospace Engineering, Georgia Institute of Technology
  • Dr. Patrick Taylor, Climate Science Branch, NASA Langley Research Center


Weber Building - CoVE