Back to top
As more countries develop and accelerate space programs, maintaining a position as a global leader in space is a key economic and national security priority for the United States. Each year, thousands of new satellites—2,664 in 2023—are launched into Earth orbits, increasing the risk of collision and damage. Satellites are critical to national infrastructure and provide global positioning services data, enable global communications, and support important logistics capabilities that ensure national economic and financial systems function.
Beyond commercial impacts, the national security stakes are crucial. “Our adversaries are putting satellites into space to threaten U.S. security, most acutely by generating intelligence and reconnaissance on the position of U.S. military forces,” says Ben Bahney, the leader of Livermore’s Space Program. “We need to protect our security interests against these new and evolving threats. Livermore’s survey missions and satellite tracking technologies mark a significant step in achieving this goal.” Livermore supports the nation’s space domain awareness (SDA) efforts to track and characterize satellites, debris, and naturally occurring objects such as asteroids. For more than a decade, Livermore has been collecting data from both ground- and space-based observatories, developing computational tools to analyze that data, and working with the U.S. Space Force to find new partners for collaboration.
Sensing, Tracking, and Monitoring
Livermore has a successful track record developing and deploying optical imaging payloads for small satellites. The Laboratory partnered with Terran Orbital to develop the GEOStare spacecraft, built on the early Space-Based Telescope for Actionable Refinement of Ephemeris (STARE) optical imaging concept. (See S&TR April 2012, "Launching Traffic Cameras into Space," and S&TR April 2019, "Space Program Innovation, One Small Satellite at a Time.") Wim de Vries, Livermore’s associate program leader for mission support, recalls, “Many people were skeptical that such a small, inexpensive platform could yield useful data and that the technology would function for very long.” The GEOStare spacecraft continues to operate in excellent condition three years after its 2021 launch with no deterioration of data or hardware quality despite originally being a six-month mission. de Vries and his team intend to conduct survey missions and target tracking for as long as the satellite continues to operate.
Survey astronomy missions capture images of targets, such as stars and galaxies, over an expanded field of view and enable researchers to identify the location of different objects—both expected and unexpected—at specific moments in time. As a small nanosatellite payload, GEOStare cannot capture panoramic images in a single photograph, but the technology can take multiple images across a designated area and stitch those tiled images together. “When we started tiling the images and thereby surveying a much larger part of the sky, we sometimes captured moving objects well away from their predicted orbital positions,” says de Vries. Satellites try to stay stable in orientation, but GEOStare can slew, or move to reorient itself to a specific object such as a star, to capture a satellite’s trail (or streak) in the images. Using multiple images, GEOStare tracks the streaks and calculates the satellite’s motion over time. “If an object shows up later or earlier than expected in a series of images, that indicates a degree of orbital uncertainty or possibly a recent maneuver,” says de Vries. “If we don’t recognize a streak at all, this either means a satellite has maneuvered quite a bit or a new piece of orbiting debris drifted by.” The team now also uses GEOStare to track asteroids nearest to Earth, in conjunction with observations from ground-based observatories.
Building on the Laboratory’s experience with GEOStare, Livermore’s space hardware team developed Deep Purple, an optical payload using ultraviolet and shortwave infrared technology, in less than a year, delivering it in January 2024 for NASA’s Pathfinder Technology Demonstrator-R mission. NASA launched the optical payload, and it achieved first light in August 2024.
The effort to track satellites requires facilities to monitor space missions in real time. However, mission operations centers (MOCs), which relay commands to ground stations and then to spacecraft in orbit, are in high demand but short supply. Livermore’s lead guidance navigation and control engineer Phillip Rittmuller explains, “Space missions outpace the demand for ground stations and MOCs, so smaller experimental spacecraft with low budgets can rarely reserve them.” Recognizing this challenge, Rittmuller and his team have been developing the Satellite, Telescope, Aerial drone, and Remote sensing Mission Operations Center (STARMOC) at Lawrence Livermore. Consisting of a mission operations room, engineering development unit area, analysis room, telescope operations room, and other office space, STARMOC will enable space missions to work in cooperation with NASA and other partners to relay communications through ground stations around the world to satellites in space. STARMOC also enables researchers to test sending commands to hardware before launching the payloads and satellites into space and analyzing data from observatories such as the one located at the Laboratory’s Livermore Valley Open Campus.
STARMOC is intended to serve lower budget space missions, such as NASA’s forthcoming Pandora Pioneer-class mission. With a relatively low budget of $20 million—compared to multibillion-dollar space missions—Pandora will use a Livermore-designed telescope and a spare sensor from the James Webb Space Telescope to characterize the brightness of stars and determine how much nearby exoplanets affect changes in the stars’ brightness. The University of Arizona will provide operations support for Pandora’s 13-month primary mission, but STARMOC could be used for continuing operations afterwards. In the future, Livermore researchers at STARMOC will conduct mission planning and will send commands to adjust satellites’ attitudes (orientation in relation to another objects) to capture the best data. STARMOC will further determine what time to point satellites at the ground station to most effectively send commands and receive telemetry in return.
Data Analysis
Even with advanced hardware and resources such as GEOStare and Livermore’s STARMOC, data collection and analysis in space is challenging: Insufficient signal-to-noise ratios (a parameter that indicates the performance and quality of systems that transmit signals) and long latency (the amount of time data travels between a user and a device) are hurdles in collecting high-quality data in the necessary quantities and speeds for rigorous research. If the data collected is of insufficient quality or quantity, subsequent analysis is likely to be inaccurate. Research in space can also swing from regimes in which almost no real data exists to regimes in which too much data exists to easily analyze, as in the case of large space surveys, yet both regimes are expected to be analyzed at the same time in different contexts.
Livermore has developed several tools and models to address data collection and analysis challenges, including a suite of tools under the Modeling and Analysis for Data-Starved and Ambiguous Environments (MADSTARE) program. Originally a Laboratory Directed Research and Development (LDRD) project led by Livermore’s Michael Schneider, MADSTARE identifies how machine learning can most effectively be used to solve space data collection problems, specifically through overcoming slow and expensive data-training procedures to cover all eventualities.
One tool developed on MADSTARE is MuyGPs, a Gaussian process computer estimation code that leverages statistical distributions to more accurately quantify uncertainty in space data. Livermore statistician and principal investigator for the MuyGPs exploratory research project Amanda Muyskens chose Gaussian processes as the foundation for the code because statistical models provide more robust solutions and uncertainty calculations, while neural networks are often insufficient in these areas. Muyskens explains, “When we’re tracking a space object, we focus on predicting the variation from the statistical likelihood of where the object is traveling or how the object is moving as opposed to finding a simple point estimate that neural networks might provide. Because statistical models provide both predictions and uncertainty estimates, we can develop models that we trust and make decisions accordingly.”
MuyGPs offers significant contributions to space data analysis when data is sparse by interpolating so that data points closer together have a higher correlation than data spread far apart. In other words, only nearest neighbors are needed to make predictions about the statistical likelihood of a particular event and its uncertainty with high degrees of accuracy.
When compared to other Gaussian process estimation methods, MuyGPs has provided the fastest and most accurate model. Muyskens and her team have used MuyGPs to track temperature changes across Earth despite clouds or other objects obstructing the satellites’ ability to collect data. MuyGPs also analyzes Monetgrams, data on a satellite’s distinct fluorescence, to characterize satellites’ behaviors over time. MuyGPs is also applicable to numerous other domains beyond space and is available as open source software.
Another tool developed under the MADSTARE program is Space Situational Awareness for Python (SSAPy). Python-based, open-source SSAPy uses high-performance computing (HPC) systems to provide fast, high-fidelity modeling capabilities for orbit propagation—the prediction of a satellite’s future orbital characteristics based on current characteristics—and orbital state inference. “When tracking an object in space, we want to be able to estimate where it might be going,” says Luc Peterson, associate program leader for data science in Livermore’s Space Program. “From just a few measurements, and many calculations, we can create something similar to a hurricane tracker that predicts a range of possibilities for where, exactly, that object might be in the future.” Running enough calculations to achieve uncertainty estimates can be expensive, so Livermore researchers designed SSAPy to be fast, accurate, and executable on HPC systems and released it as an open-source tool under an LDRD project led by spacecraft systems engineer Kerianne Pruett.
SSAPy accounts for forces that impact satellites, such as the drag force in Earth’s atmosphere and gravity from the moon, sun, and planets. After incorporating these realistic physics models, SSAPy accurately propagates the motion of a satellite through space and time. Where SSAPy differs from commercial models, however, is its ability to accurately calculate solutions to chaotic real-world orbits, in which small changes in initial conditions can lead to large changes in the final state. By releasing SSAPy as open-source software, the Livermore team has given the broader SDA community a free, fast, and accurate tool to customize for applications.
A third challenge of SDA is closing kill chains, such as determining whether a rocket launch poses a threat to a satellite in space. Fusing data from multiple sensors with different modalities is key to confidently and correctly assessing possible threats. To this end, Livermore collaborates with the U.S. Space Force’s Space Domain Awareness Tools, Applications, and Processing Laboratory (SDA TAP Lab) in evaluating capabilities developed by government, industry, and academia to close these kill chains. Specifically, Livermore assists the SDA TAP Lab in rigorously and objectively benchmarking applications developed by SDA TAP Lab participants, ensuring that the proposed solutions effectively address the Space Force’s targeted technical problems.
Looking to the future of SDA, Livermore and its partners are tracking the ongoing challenges and finding new approaches to solve them, whether through innovative cube satellites and optical payloads, such as GEOStare, or new computer codes to analyze the telemetry sent from satellites. Observing the trends in space, Bahney says, “Space as we knew it is a thing of the past. We now face a space domain that is congested, contested, and competitive. We must leverage all our tools and resources to find effective solutions.”
—Sheridan Hyland
For further information contact Ben Bahney (925) 423-0353 (bahney1 [at] llnl.gov (bahney1[at]llnl[dot]gov)).
