Exploring the retrieval potential of the Ice Cloud Imager
Licentiate thesis, 2024
The upcoming Ice Cloud Imager (ICI) and the recently launched Arctic Weather Satellite (AWS) mark a new era in cloud ice observations. For the first time, continuous and global observations of the troposphere will be made at sub-millimetre wavelengths. Sub-millimetre observations are highly sensitive to larger ice crystals. These crystals contain a significant fraction of the ice mass in clouds which, despite their influence on Earth's climate, remain poorly understood. As a result, ICI and AWS will offer unparalleled data on atmospheric ice.
This thesis poses the question: What information on cloud ice can ICI observations provide? The primary objective of ICI is to provide ice cloud variables covering the entire atmospheric column. However, vertical information on ice has never before been derived from a passive microwave instrument. The question is therefore explored in two contexts: Firstly, how reliably will ICI fulfil its primary objective? Secondly, can we determine the vertical distribution of ice from ICI observations?
Since ICI is not yet launched, high-quality radiative transfer simulations of ICI are required to train the inversion model. Since there will be no co-locations of ICI with a radar providing cloud ice data, empirical retrievals will not be feasible after the launch. Consequently, the retrieval model used during ICI's operational phase must rely on the simulations. In this thesis, state-of-the-art simulations are presented, and a quantile regression neural network (QRNN) is used to produce probabilistic retrieval estimates.
The findings in this thesis indicate that ICI will produce reliable retrievals of the column-integrated variables: ice water path, mean mass height, and mean mass diameter, with a sensitivity to ice water paths ranging from 10 g m-2 to 10 kg m−2. The simulations pertaining to the column variables lay the foundation for the EUMETSAT level-2 ICI product. Vertical profiles of cloud ice are retrieved from ICI observations, achieving a resolution of ~2.5 km.
Together, the observations from AWS and ICI will provide benefits to numerical weather prediction and deepen our understanding of ice clouds. The long-term cloud ice dataset from ICI will also support climate monitoring and validation of climate models. Furthermore, ICI could provide a truly novel dataset of vertical cloud ice, offering insights throughout the entire depth of an ice cloud.
This thesis poses the question: What information on cloud ice can ICI observations provide? The primary objective of ICI is to provide ice cloud variables covering the entire atmospheric column. However, vertical information on ice has never before been derived from a passive microwave instrument. The question is therefore explored in two contexts: Firstly, how reliably will ICI fulfil its primary objective? Secondly, can we determine the vertical distribution of ice from ICI observations?
Since ICI is not yet launched, high-quality radiative transfer simulations of ICI are required to train the inversion model. Since there will be no co-locations of ICI with a radar providing cloud ice data, empirical retrievals will not be feasible after the launch. Consequently, the retrieval model used during ICI's operational phase must rely on the simulations. In this thesis, state-of-the-art simulations are presented, and a quantile regression neural network (QRNN) is used to produce probabilistic retrieval estimates.
The findings in this thesis indicate that ICI will produce reliable retrievals of the column-integrated variables: ice water path, mean mass height, and mean mass diameter, with a sensitivity to ice water paths ranging from 10 g m-2 to 10 kg m−2. The simulations pertaining to the column variables lay the foundation for the EUMETSAT level-2 ICI product. Vertical profiles of cloud ice are retrieved from ICI observations, achieving a resolution of ~2.5 km.
Together, the observations from AWS and ICI will provide benefits to numerical weather prediction and deepen our understanding of ice clouds. The long-term cloud ice dataset from ICI will also support climate monitoring and validation of climate models. Furthermore, ICI could provide a truly novel dataset of vertical cloud ice, offering insights throughout the entire depth of an ice cloud.
Remote sensing
clouds
sub-millimetre
retrieval
EA, EDIT, Campus Johanneberg, Hörsalsvägen 11
Opponent: Jie Gong, NASA Goddard Space Flight Center, USA
Author
Eleanor May
Chalmers, Space, Earth and Environment, Geoscience and Remote Sensing
The Ice Cloud Imager: retrieval of frozen water column properties
Atmospheric Measurement Techniques,;Vol. 17(2024)p. 5957-5987
Journal article
May, E. and Eriksson, P. The Ice Cloud Imager: retrieval of frozen water mass profiles.
Subject Categories
Meteorology and Atmospheric Sciences
Publisher
Chalmers
EA, EDIT, Campus Johanneberg, Hörsalsvägen 11
Opponent: Jie Gong, NASA Goddard Space Flight Center, USA