A recent program has created a simulation of several satellites that capture images of a cloud from different perspectives simultaneously, potentially enhancing our comprehension of the processes occurring within clouds.
David Stanley, motivated by his passion for climate change, devised a program that optimizes data collection for studying clouds. The program allows for multiple satellites to capture images of a cloud from various angles at once, which could help illuminate the internal dynamics of clouds.
“Typically, we can only observe the external features of a cloud,” Stanley explained. “Computed cloud tomography derives its name from computed tomography, similar to a CT scan. However, instead of X-rays, satellites take images of the cloud from numerous angles in a brief timeframe.”
Stanley pointed out that one significant uncertainty in climate models is the extent to which convective transport influences the formation of new clouds. Convection involves the movement of heat and moisture in the atmosphere, particularly vertical air movements in unstable conditions.
“By producing multiple observations of the same cloud over time, one can track how convection evolves and affects the formation of other clouds moving forward. Increased cloud formation can amplify the greenhouse effect.”
After earning his master’s degree in aerospace engineering at the University of Illinois Urbana-Champaign, Stanley applied to pursue a Ph.D. at the same institution.
“I discussed my general interests in engineering, space engineering, and the crucial need to enhance our understanding of climate change and seek solutions,” he detailed. “Robyn Woollands recognized my passion and invited me to join her research team. She connected me with Federico Rossi and Amir Rahmani from NASA’s Jet Propulsion Laboratory’s Multi-Agent Autonomy Group, who then introduced me to JPL scientists Changrak Choi and Anthony Davis, experts in cloud tomography, atmospheric clouds, and aerosols. It aligned perfectly with my interests, and Robyn was exploring a mission proposal that utilized multi-agent systems to aid Earth science missions.”
For the simulation, Stanley employed a mixed integer linear programming solver, a versatile tool used across various applications. He crafted the code to create a scheduler that would optimize the timing and camera angles for a fleet of satellites, enabling them to capture the maximum number of images of the cloud.
“What intrigued me about this project was our use of mixed integer linear programming to automatically identify the most effective pointing pattern for the satellite formation. All satellites needed to focus on the same target simultaneously. However, there could be numerous potential targets beneath each satellite, and if not pointed correctly, some might be overlooked.”
The aim was to enhance the number of targets the satellites could observe during their orbit.
“We conducted two distinct simulations. The first involved clouds generated on Earth with a defined lifespan; in the computer model, they were simply points on a sphere. The second simulation tracked the satellite formation. This could be done either using straightforward methods or more intricate, accurate models.
“By merging the data from these two simulations, the program computes where the satellites are at various points in their orbit and where the clouds are located at those times. It then determines the optimal viewing pattern for the satellites relative to the clouds below.”
Stanley encountered multiple instances during the research where he had differing ideas on the best approach for simulating and relaying data to the solver.
“You might require an array for each time increment and each satellite, or you could have arrays for different sections of Earth. Initially, I tried using distinct sections of the Earth as pointing coordinates by brute-forcing subdivisions. However, the vast area resulted in millions of indices, which were beyond the solving capacity of a standard desktop computer.”
Ultimately, Stanley drew inspiration from Woollands’ earlier work, which involved a constellation of satellites orbiting Mars designed to maximize observations of Martian dust devils. Instead of subdividing the earth as a whole, they segmented areas beneath the satellites, which minimized the number of indices needed.
“Additionally, I realized I could use the clouds as the index,” he added. “This strategy effectively reduced the pool from millions of indices to just a few hundred at a time, making it much more manageable.”
Stanley emphasized that this information is based on simulations.
“We’ve made certain assumptions regarding cloud formation and movements, offering considerable room for improvement through real-world data instead of relying solely on generated data. The key takeaway is that we’ve established a novel method that could greatly enhance the collection of 3D cloud data, leading to improved insights into cloud dynamics and their long-term climatic impacts.
