TL;DR
The Stingray sensor system is a novel, fully autonomous 15-camera optical array designed for persistent, high-cadence astrometric and photometric surveys of the geosynchronous Earth orbit (GEO) belt. This system addresses the critical need for detailed characterization of GEO objects by providing a continuous stream of data, enabling new avenues for research in space situational awareness and satellite behavior analysis.
What Problem Does This Paper Address?
Space situational awareness (SSA) relies on understanding the vast number of objects orbiting Earth. However, characterizing these objects, especially those in the geosynchronous Earth orbit (GEO) – a region critical for many applications – is challenging. Traditional methods involve either rapidly "hopping" between targets with limited observation time per object, or focusing on a few objects for extended periods, leading to significant gaps in data and knowledge. This lack of persistent, high-cadence data hinders detailed characterization, identification, and the development of advanced analysis techniques, particularly for machine learning models that require large, real-world datasets. The Stingray system aims to bridge this gap by providing continuous, all-night observations of a large number of GEO satellites.
Methods
The Stingray system employs a multi-aperture approach using 15 ZWO ASI1600MM Pro cameras, each paired with a Sigma 135mm f/1.8 lens. This configuration provides a wide field of regard, covering a significant portion of the GEO belt visible from Tucson, Arizona. A key innovation is the system's complete autonomy in data acquisition and processing, managed by four Intel NUC computers. This automation is crucial for handling the immense data volume and ensuring continuous operation without human intervention.
The data processing pipeline involves several stages:
- Nonuniformity Correction: Using calibration images (bias, dark, and flat frames) collected periodically.
- Astrometric Reduction: Plate solving using the GAIA DR2 star catalog to determine precise celestial coordinates. This involves an iterated isophotal first-order moment method for centroid computation, followed by cross-correlation with star catalogs using Scamp.
- Photometric Calibration: Converting instrumental magnitudes to GAIA G magnitudes using solar-type stars and a custom local star catalog.
The system is designed to operate autonomously, with a central node managing weather monitoring, roof operation, and image collection commands to the camera control nodes.
Datasets
- GAIA DR2 Star Catalog: Used for astrometric plate solutions and photometric calibration. A custom local catalog derived from GAIA DR2, containing over 75 million stars, was created to facilitate efficient in-situ processing.
- Landolt Standard Star Fields: Used during the hardware trade study for evaluating lens performance.
- WAAS Satellite Ephemerides: Publicly available high-accuracy ephemerides for three Wide Area Augmentation System (WAAS) satellites (SES-15, GALAXY-30, and Eutelsat 117W B) were used as calibration targets for astrometric accuracy.
No new benchmark datasets were introduced; rather, the system generates a large dataset of real-world observations for existing and future SSA research.
Key Insights and Ideas
- Autonomous Operation is Essential: The sheer volume of data and the need for persistent observation make full automation a necessity, not a luxury, for effective GEO belt surveys.
- Multi-Aperture Design Balances Coverage and Cost: The choice of 15 cameras with specific lenses strikes a balance between covering a wide field of regard and managing cost and logistical complexity, avoiding the pitfalls of very wide-field lenses or an excessive number of telescopes.
- Real-World Data is Crucial for ML: The system's output is intended to provide much-needed real-world data to train and validate machine learning models for satellite characterization, which have historically relied heavily on simulated data.
- Persistent Observation Unlocks New Analysis: Continuous, all-night custody of GEO objects opens doors for novel characterization analyses focusing on instantaneous, short-term, and long-term effects, and behavior patterns that were previously inaccessible.
- Tumbling Satellites as a Unique Data Source: The observation of a tumbling satellite (Raduga 1M-1) and the successful extraction of its rotation period highlights the system's capability to characterize non-stabilized objects and debris, providing valuable insights into their dynamics.
Key Results
- Photometric Uncertainty: An aggregate photometric uncertainty of 0.062 ± 0.008 magnitudes was achieved during an initial one-month survey.
- Astrometric Accuracy: The system demonstrates astrometric accuracy consistent with theoretical sub-pixel centroid limits. For WAAS satellites, astrometric errors were found to be statistically consistent with zero, with an average exposure latency correction improving results.
- Limiting Magnitude: The system achieves a limiting magnitude of just brighter than 16 GAIA G magnitude from the Biosphere 2 site, which is approximately 1 magnitude fainter than at a less dark testing site, indicating good performance in a darker environment.
- Data Volume: The chosen data collection cadence (10s exposure, 2x2 binning, every minute) produces approximately 50 GB of data per night, a manageable amount for automated processing and transfer.
- Satellite Characterization: The system successfully collected data for approximately 200 near-GEO satellites nightly, with an average of 233 observations per satellite during the initial survey. The light curve analysis of a tumbling satellite yielded a rotation period of 908.1135 ± 0.8207 seconds.
Limitations and Future Work
- Calibration Image Frequency: Due to the manual process of changing filters, calibration images (bias, dark, flat) are collected monthly rather than nightly. The impact of calibration data age on results is reserved for future analysis.
- Star Trail Contamination: Star trails in images, though managed by stochastic filtering, can reduce the total number of observations for each satellite.
- Exposure Latency: While addressed with an average correction, there is still some random additional latency in image capture and readout that requires further characterization for improved precision.
- Future Work: The authors envision using the collected data to validate and improve physics-based and machine learning-based characterization techniques. Further analysis of the data is planned to explore various techniques for understanding satellite behavior and dynamics. The system's ability to provide near-real-time data also opens possibilities for adapting the output for real-time analysis streams.
Summary updated