Periodic Reporting for period 3 - CloudCT (Climate CT- Cloud Tomography by Satellites for Better Climate Prediction)
Período documentado: 2022-08-01 hasta 2024-01-31
The public awareness of the anthropogenic effect on climate and of climate change has grown significantly in recent years. The mean global temperature is rising, with 2023 marking the warmest year recorded in modern records. The world experiences powerful tropical storms and intense rain events, leading to widespread and severe floods on the one hand and frequent heat waves on the other. Consequently, addressing the challenges posed by climate change has become a critical societal concern and a key focus for the scientific community. Future climate prediction stands out as a major area of investigation.
In these predictive efforts, it is widely acknowledged that clouds, especially warm shallow clouds (the central focus of this project), play a major role. Understanding their properties and how they respond to climate change, is crucial for reducing uncertainties in climate predictions. Warm clouds are abundant globally and significantly influence the radiative budget, as well as the transport of heat, humidity, pollutants, and aerosols within the boundary layer and into the free troposphere.
Clouds forming within cloud fields create complex systems characterized by interactions between dynamic, thermodynamic, microphysical, and radiation processes. These processes are coupled and modulated by positive and negative feedbacks such that cloud properties dictate, as well as being affected by the field's organization. Deciphering these dynamic systems and their evolution in changing environmental conditions remains one of the most challenging aspects of climate research.
The primary objective of this project is to investigate the properties and trends of these warm cloud systems. Achieving this goal requires an observation system capable of comprehensively measuring both macrophysical (coverage) and microphysical properties (optical properties, droplet size distribution, liquid water content, and precipitation). This comprehensive data will be used to improve the parameterization of shallow warm cloud properties in climate models.
Despite their significance, a significant subset of shallow clouds is inadequately observed. These clouds are often sparse and small, making them easily overlooked by most current Earth observing systems. Furthermore, even with satellites offering high spatial resolutions, current methods often fall short in retrieving their properties adequately.
This project aims to address a crucial knowledge gap in cloud and climate models by introducing a novel tomography and AI-based approach utilizing small satellite technology. This approach involves capturing images of the same small clouds from multiple angles by a few satellites and then utilizing this data to estimate the 3D properties of key microphysical variables.
This collaborative project involves three groups with complementary expertise. The Weizmann Institute group focuses on cloud physics through theoretical approaches, observations, and numerical modeling. Their responsibility includes developing detailed, high-resolution models of cloud fields that closely resemble real clouds. These simulations serve as training and validation tools to the algorithms developed by the Technion group.
To achieve these objectives, the Weizmann group improves their high-resolution numerical models of single clouds and cloud fields, leading to more realistic simulations of shallow warm cloud fields. The methodology involves combining theoretical approaches, continually tested and refined through observations from both ground-based and space-based sources.
The Technion group derives imaging technologies for CloudCT, in tight collaboration with the the Weizmann and ZfT groups. The project posed new questions that would apply to future proejsts beyond CloudCT. These include: how to best sense microphysics in 3D from space? How to account for object evolution while multiview images are taken for scattering-based CT? How to calibrate polarimetric obsevational cameras, while in orbit, without being sensitive to unknown winds and aerosols? Can self-calibration be defined for polarimetric imaging and for CT? How can machine learning be utilized for scattering-based CT? The project not only posed these questions, but also enables us to create the solutions.
The ZfT (Center for Telematics) group focuses on the realization of formation flying technologies and satellite systems. The contribution of ZfT to this project will enable generation of the cloud observations from multiple views at the same time. The most challenging aspect of this project is attitude and orbit control coupled with the miniaturization constraints. Additionally, ZfT is responsible for implementation of the other subsystems of the satellite like communication, on board computer, power systems where design is dictacted by the requirements of the formation flying topology. All these challenges are addressed by innovative design concepts which utilizes highly integrated systems in a small unit. The formation technologies developed in this project will pave the path for the future formation flying missions which will enable scientific data collection methods which were not feasible with a single satellite missions.
In these predictive efforts, it is widely acknowledged that clouds, especially warm shallow clouds (the central focus of this project), play a major role. Understanding their properties and how they respond to climate change, is crucial for reducing uncertainties in climate predictions. Warm clouds are abundant globally and significantly influence the radiative budget, as well as the transport of heat, humidity, pollutants, and aerosols within the boundary layer and into the free troposphere.
Clouds forming within cloud fields create complex systems characterized by interactions between dynamic, thermodynamic, microphysical, and radiation processes. These processes are coupled and modulated by positive and negative feedbacks such that cloud properties dictate, as well as being affected by the field's organization. Deciphering these dynamic systems and their evolution in changing environmental conditions remains one of the most challenging aspects of climate research.
The primary objective of this project is to investigate the properties and trends of these warm cloud systems. Achieving this goal requires an observation system capable of comprehensively measuring both macrophysical (coverage) and microphysical properties (optical properties, droplet size distribution, liquid water content, and precipitation). This comprehensive data will be used to improve the parameterization of shallow warm cloud properties in climate models.
Despite their significance, a significant subset of shallow clouds is inadequately observed. These clouds are often sparse and small, making them easily overlooked by most current Earth observing systems. Furthermore, even with satellites offering high spatial resolutions, current methods often fall short in retrieving their properties adequately.
This project aims to address a crucial knowledge gap in cloud and climate models by introducing a novel tomography and AI-based approach utilizing small satellite technology. This approach involves capturing images of the same small clouds from multiple angles by a few satellites and then utilizing this data to estimate the 3D properties of key microphysical variables.
This collaborative project involves three groups with complementary expertise. The Weizmann Institute group focuses on cloud physics through theoretical approaches, observations, and numerical modeling. Their responsibility includes developing detailed, high-resolution models of cloud fields that closely resemble real clouds. These simulations serve as training and validation tools to the algorithms developed by the Technion group.
To achieve these objectives, the Weizmann group improves their high-resolution numerical models of single clouds and cloud fields, leading to more realistic simulations of shallow warm cloud fields. The methodology involves combining theoretical approaches, continually tested and refined through observations from both ground-based and space-based sources.
The Technion group derives imaging technologies for CloudCT, in tight collaboration with the the Weizmann and ZfT groups. The project posed new questions that would apply to future proejsts beyond CloudCT. These include: how to best sense microphysics in 3D from space? How to account for object evolution while multiview images are taken for scattering-based CT? How to calibrate polarimetric obsevational cameras, while in orbit, without being sensitive to unknown winds and aerosols? Can self-calibration be defined for polarimetric imaging and for CT? How can machine learning be utilized for scattering-based CT? The project not only posed these questions, but also enables us to create the solutions.
The ZfT (Center for Telematics) group focuses on the realization of formation flying technologies and satellite systems. The contribution of ZfT to this project will enable generation of the cloud observations from multiple views at the same time. The most challenging aspect of this project is attitude and orbit control coupled with the miniaturization constraints. Additionally, ZfT is responsible for implementation of the other subsystems of the satellite like communication, on board computer, power systems where design is dictacted by the requirements of the formation flying topology. All these challenges are addressed by innovative design concepts which utilizes highly integrated systems in a small unit. The formation technologies developed in this project will pave the path for the future formation flying missions which will enable scientific data collection methods which were not feasible with a single satellite missions.
In the initial two periods of this project, we improved our Large Eddy Simulation (LES) models to better simulate warm clouds with increased accuracy. A significant aspect of this involved implementing a new microphysical scheme that accounts for the regeneration of aerosols after the evaporation of drops, detailing the aerosols' return to the atmosphere post-cloud interaction. This addition significantly enhances the accuracy of aerosol budget treatment, both in the environment and within clouds, leading to a more realistic description of cloud microphysics. In parallel, we conducted super high-resolution simulations of individual clouds to examine the mixing processes between warm clouds and their surroundings. We simulated various warm cloud fields as well. In parallel, we continued exploring continental convective clouds (defined as green Cumulus), a primary cloud type within this project. We constructed a new global long-term database for these clouds, utilizing it to define the meteorological conditions that support the formation of such cloud fields.
In the field of theoretical radiation transfer, our work revealed that true color glories, an optical phenomenon resulting from sunlight interacting with water droplets in clouds, are observable in raw, unpolarized satellite images on a daily basis. This observation provides a substantial and untapped cloud dataset, establishing a straightforward link between cloud droplet size and the structure of the glory through a diffraction-like approximation.
On a global scale, we investigated climate trends in temperatures and the corresponding response of clouds. Using an innovative method, we demonstrated the warming of sea surface temperatures in most oceans, with notable cooling in the North Atlantic and Southern Ocean. The response of clouds to global warming was evident in decreased cloud coverage over most continents and an increasing trend over tropical and subtropical oceans.
Our efforts also extended to the interface of cloud physics, nonlinear dynamics, and data science. In a paper introducing a new route to complexity via phase-dependent stochastic parameterizations, we presented a general framework to enhance the realism of solutions for nonharmonic oscillations by breaking down their nonlinearity. By superimposing stochastic parameterizations on these structures, we achieved stochastic chaotic solutions, uncovering a new pathway to complexity with realistic cloud oscillations that exhibit enhanced time-variability. These stochastic parameterizations aim to replace missing physics in simpler models, particularly those related to rain effects.
We greatly advanced methods for computed tomography (CT) in non-trivial setings, and what we learned in clouds is helping us affect other domains. For clouds, we derived efficient scattering-based CT. This has been demonstrated by several studies corresponding to different principles: machine learning, stoachstic differential rendering, recycling monte-carlo paths for efficient use of graphical processing units, multi-scale (coarse to fine) recovery, and use of monotnonicity priors. Moreover, we generalized scattering-based CT to polarimetric imaging, thereby retrieving microphysics of cloud droplets.
The algorithmic methods helped us affect additional domains of application. One of then includes X-ray CT exploiting scattering: our methods enable this to be achieved. Moreover, we showed that this approach can speed up X-ray CT acquisition, by source multiplexing. Another domain is of atmospheric turbulence fields: we showed, for the first time, that scintillation of light sources in the open air can be a data source for CT of of the turbulence strength, when using an array of cameras spread outdoors in a county-scale. Moreover, we devised an efficient self-calibration method using CT projections, and showed that it can significantly improve in-situ CT of populations of plankton.
The issues of calibration and self-calibration became increasingly significant as our project advanced towards a real-world system. Therefore, we drived a method for calibrating atmospheric aerosol lidars, and invented methods for polarimetric calibration of spaceborne cameras - in orbit. One approach uses solar-farms as a calibration target. Another method uses zodiacal light (scattered by inter-planetary dust) as the target. In addition, we created a calibration setup – including polarization – for the spaceborne payloads.
Work on the payloads went from scratch to working systems through the project. Initially, we were unsure what is required of the payload. We worked intensively to define this. As a result, we defined a detailed tender. A company (Dragonfly) won the tender and is building for CloudCT the payloads. We already received two of the payloads and they undergo testing. This is part of our empirical lab-based research in the project. Additional lab work involved tomography in controlled settings. One branch of this work involved building a special chamber that created (for a few seconds) indoor clouds not bounded by walls. Another branch looks into 3D recovery of a scattering medium in a microscopic setting.
In the field of theoretical radiation transfer, our work revealed that true color glories, an optical phenomenon resulting from sunlight interacting with water droplets in clouds, are observable in raw, unpolarized satellite images on a daily basis. This observation provides a substantial and untapped cloud dataset, establishing a straightforward link between cloud droplet size and the structure of the glory through a diffraction-like approximation.
On a global scale, we investigated climate trends in temperatures and the corresponding response of clouds. Using an innovative method, we demonstrated the warming of sea surface temperatures in most oceans, with notable cooling in the North Atlantic and Southern Ocean. The response of clouds to global warming was evident in decreased cloud coverage over most continents and an increasing trend over tropical and subtropical oceans.
Our efforts also extended to the interface of cloud physics, nonlinear dynamics, and data science. In a paper introducing a new route to complexity via phase-dependent stochastic parameterizations, we presented a general framework to enhance the realism of solutions for nonharmonic oscillations by breaking down their nonlinearity. By superimposing stochastic parameterizations on these structures, we achieved stochastic chaotic solutions, uncovering a new pathway to complexity with realistic cloud oscillations that exhibit enhanced time-variability. These stochastic parameterizations aim to replace missing physics in simpler models, particularly those related to rain effects.
We greatly advanced methods for computed tomography (CT) in non-trivial setings, and what we learned in clouds is helping us affect other domains. For clouds, we derived efficient scattering-based CT. This has been demonstrated by several studies corresponding to different principles: machine learning, stoachstic differential rendering, recycling monte-carlo paths for efficient use of graphical processing units, multi-scale (coarse to fine) recovery, and use of monotnonicity priors. Moreover, we generalized scattering-based CT to polarimetric imaging, thereby retrieving microphysics of cloud droplets.
The algorithmic methods helped us affect additional domains of application. One of then includes X-ray CT exploiting scattering: our methods enable this to be achieved. Moreover, we showed that this approach can speed up X-ray CT acquisition, by source multiplexing. Another domain is of atmospheric turbulence fields: we showed, for the first time, that scintillation of light sources in the open air can be a data source for CT of of the turbulence strength, when using an array of cameras spread outdoors in a county-scale. Moreover, we devised an efficient self-calibration method using CT projections, and showed that it can significantly improve in-situ CT of populations of plankton.
The issues of calibration and self-calibration became increasingly significant as our project advanced towards a real-world system. Therefore, we drived a method for calibrating atmospheric aerosol lidars, and invented methods for polarimetric calibration of spaceborne cameras - in orbit. One approach uses solar-farms as a calibration target. Another method uses zodiacal light (scattered by inter-planetary dust) as the target. In addition, we created a calibration setup – including polarization – for the spaceborne payloads.
Work on the payloads went from scratch to working systems through the project. Initially, we were unsure what is required of the payload. We worked intensively to define this. As a result, we defined a detailed tender. A company (Dragonfly) won the tender and is building for CloudCT the payloads. We already received two of the payloads and they undergo testing. This is part of our empirical lab-based research in the project. Additional lab work involved tomography in controlled settings. One branch of this work involved building a special chamber that created (for a few seconds) indoor clouds not bounded by walls. Another branch looks into 3D recovery of a scattering medium in a microscopic setting.
In the field of cloud physics, our objective is to formulate a more accurate physical depiction of the formation and life cycle of warm convective clouds and cloud fields. This involves providing a realistic account of the vertical development of clouds, considering the role of entrainment and mixing, and examining how these factors impact cloud microphysical properties. Additionally, our research aims to enhance our comprehension of the organization of warm cloud fields. This will enable us to address key questions such as identifying prevalent organization patterns, understanding their correlation with thermodynamic conditions and aerosol properties, and uncovering the driving mechanisms behind these patterns. The integration of updates to the Large Eddy Simulation (LES) model, theoretical studies, and new CloudCT measurements, is anticipated to enhance our understanding of shallow warm clouds. This encompasses gaining insights into their coverage, lifespan, organization patterns, and optical properties, all of which will elucidate their role in the climate system. Based on that and furthermore, we anticipate that our combined measurements and theoretical models will offer a more realistic and physical basis for parameterizing such clouds in climate models, thereby reducing uncertainties. Simultaneously, the new measurements will facilitate the development of alternative simulation approaches with rules more closely aligned with data-driven trends. Ultimately, by combining a deeper understanding of cloud observations, reanalysis data, LES simulations, and the new CloudCT data, we aim to better capture climate trends that are challenging to discern due to the intricate and highly variable nature of cloud systems.
In terms of imaging technologies and algorithms, all that is written above in this context is beyond the state of the art. The methodologies should have long term implications, because they are not specific to clouds in the CloudCT project. Mostly, the results we derive are applicable to various scattering media, including tissue and humans in medical diagnosis, or underwater specimens. We believe the architectures of machine learning for CT of clouds will be also applicable to scattering-CT in biomedical context. In-orbit calibration of polatization response in cameras should be applicable to any future spaceborne projects that would opt to use polarimetric imaging. Now, as the project progresses, we reach convergence of our understanding of many problems in this cross-disciplinary challenge. Consequently, results of different aspects of the project converge. So, by the ned of the project will form a comprehensive toolkit for 3D tomography including 3D retreivals using scattering, that recovers microphysics in 3D, as well as uncertainty measures in 3D. This is expected – by design – to be achieved despite various real-world implefections of sensing, and doing so at fast runtime, to meet communication rates from space.
That being said, the most impressive outcome of imaging technologies is yet to come in this project: real image data from space taken by the formation, analyzed by the toolkit of methods we derived, demonstrating cloud tomgography from space and thus addressing the major climate questions.
In terms of imaging technologies and algorithms, all that is written above in this context is beyond the state of the art. The methodologies should have long term implications, because they are not specific to clouds in the CloudCT project. Mostly, the results we derive are applicable to various scattering media, including tissue and humans in medical diagnosis, or underwater specimens. We believe the architectures of machine learning for CT of clouds will be also applicable to scattering-CT in biomedical context. In-orbit calibration of polatization response in cameras should be applicable to any future spaceborne projects that would opt to use polarimetric imaging. Now, as the project progresses, we reach convergence of our understanding of many problems in this cross-disciplinary challenge. Consequently, results of different aspects of the project converge. So, by the ned of the project will form a comprehensive toolkit for 3D tomography including 3D retreivals using scattering, that recovers microphysics in 3D, as well as uncertainty measures in 3D. This is expected – by design – to be achieved despite various real-world implefections of sensing, and doing so at fast runtime, to meet communication rates from space.
That being said, the most impressive outcome of imaging technologies is yet to come in this project: real image data from space taken by the formation, analyzed by the toolkit of methods we derived, demonstrating cloud tomgography from space and thus addressing the major climate questions.