Spectro Capture: A Software System for Automated Small-Observatory Spectroscopy
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Astrophysics. The day in astrophysics: observations, models and what the data forces.
Vera: We're looking at the paper: "Spectro Capture: A Software System for Automated Small-Observatory Spectroscopy".
Jocelyn: Here's the gist of the paper. Spectro Capture is a Python-based software system developed to automate small-observatory fibre-fed spectroscopy, demonstrating that reliable unattended spectroscopy is practical at a small observatory when target acquisition, guiding, scheduling and calibration control are treated as an integrated software problem The system successfully completed 339 of 345 attempted science target blocks in primary unattended batch runs, achieving a completion rate of 98.3% System Capabilities and Design The software integrates target selection, telescope slewing, guide star acquisition, fibre position restoration, calibration and science exposure sequencing within a single observing workflow The framework is built around a modular Python application with core functionality including direct camera control via ASCOM and integration of PHD2 for guiding Key components include the Batch Runner script, which handles sequential target processing, and the Scheduler component, which manages target eligibility based on observing constraints like hour angle The system uses a shared application context to coordinate state and device connections across various scripts Operational Workflow Automation The automated workflow involves several distinct steps coordinated by the software, as illustrated in Figure 2 The guiding logic comprises four main steps: confining star detection to a defined region of interest, using a controlled exposure sequence to find the target, restoring the detected star to a stored fibre reference position, and using adaptive exposure control once guiding is established Target finding is handled by an ROI acquisition function in batch runner.py, which uses an exposure ladder to find a star within a configured rectangular region of interest (ROI) Successful centring is verified rather than assumed through checks on the star's proximity to the fibre position Data Handling and Calibration Management Spectro Capture controls the full path from camera exposure to reduction-ready FITS file, meaning it does not simply trigger exposures but manages the entire process For each exposure, the routine sets configured binning, resets the camera to a full-frame readout, starts the ASCOM exposure, and polls the camera’s ImageReady state until completion Calibration frame acquisition is handled by run calibration, which acquires tungsten flat field frames and thorium wavelength calibration frames using the same exposure and FITS generation path as science images File handling is deterministic, creating target-specific folders, date-coded ‘night’ folders, and run folders to ensure a predictable data layout Performance Metrics and Limitations The operational results show that the full automated acquisition overhead from PHD2 looping to the start of the first science exposure averaged 62 seconds, with 90% of cases completed within 92 seconds The system's success depends on repeatability, suggesting it works best when the telescope, mount, guide camera and spectrograph behave as a stable and calibrated system Limitations include the dependency on local SIMBAD lookup support for pre-slew coordinate resolution and weather handling that currently stops operations rather than treating interruptions as recoverable events Future improvements are scoped to address boundary conditions such as longer faint-target patience and more flexible settle criteria The development of Spectro Capture suggests that unattended multi-target spectroscopy is moving from an aspiration to a practical operating mode for a variety of suitably stable instruments The software has evolved into an end-to-end system for automated target selection, acquisition and data capture This shift indicates that the assumption that fully automated small-observatory spectroscopy is difficult enough to be out of reach is beginning to change The results presented here suggest that unattended spectroscopy is moving from an aspiration to a practical operating mode for a variety of suitably stable instruments The software does not replace the need for a stable installation, but rather makes use of one The more realistic path to automation is a software layer built around a stable and calibrated instrument APPENDIX A Code availability The Spectro Capture source code is publicly available on GitHub, released under a non-commercial source license permitting personal, educational and research use The repository can be accessed at https://github.com/pluckas/spectro-capture The software is provided as source code rather than as a packaged installer, requiring a working Python environment and sufficient familiarity with Python to install dependencies The required Python packages and other dependencies are listed in the repository documentation The Spectro Capture interface is organized into task-based tabs, including a Batch Main unattended observing interface and a Guide PHD2 connection The Status tab provides live status information which is saved to a nightly disk log The Batch Runner tab serves as the main observing interface for timed and scheduled targets The Guide tab provides access to PHD2 controls, configuration, and live guide information The Tools tab allows manual bias/dark utilities and manual lamp control for the eShel tungsten and thorium lamps The Setup Hardware connection and configuration for camera and telescope, ROI centre/size fields are managed in the Setup tab The Dome tab allows manual control through gui dome.py and dome backend.
Paper summary: Vera: So, to get into the summary of this paper, "Spectro Capture: A Software System for Automated Small-Observatory Spectroscopy," it's fundamentally about creating a Python-based software system that automates small observatory fiber spectroscopy. The central claim is that you can integrate target selection, telescope slewing, guide star acquisition, fiber position restoration, calibration management, and science exposure sequencing all within a single observing workflow.
Jocelyn: They test this integration at Shenton Park Observatory in Australia over a period from January to May two thousand twenty-six <ref:2605.19459#pg1,from January to May 2026>. The system’s demonstration involves running primary unattended batch runs where they achieved a completion rate of ninety-eight point three percent for their science target blocks <ref:2605.19459#pg1,a completion rate of 98>.
Subrahmanyan: That result, achieving that ninety-eight point three percent completion rate, is what the paper uses to demonstrate its thesis: reliable unattended spectroscopy is practical at a small observatory when target acquisition, guiding, scheduling and calibration control are treated as an integrated software problem.
Vera: The system’s design itself is modular; it’s built around a Python application that has core functions like direct camera control through ASCOM and integrating PHD2 for guiding. They also have specific components like a batch runner script that manages the sequential target processing and a scheduler component that handles target eligibility based on constraints like hour angle.
Jocelyn: And they describe the operational workflow as having four distinct steps in their guiding logic: confining star detection to a defined region of interest, using a controlled exposure sequence to find the target, restoring the detected star back to its stored fiber reference position, and then using adaptive exposure control once guiding is established.
Subrahmanyan: That sequence sounds like they’re building a feedback loop that moves from initial acquisition through verification and finally into active control. It’s an attempt to mirror a conventional imaging workflow but tailored specifically for spectroscopy needs, which makes sense because it grounds the automation in something familiar.
Vera: Beyond the guiding, data handling is where they manage the whole path from camera exposure to a reduction-ready FITS file. This means for every exposure, the routine sets configured binning, resets the camera to a full frame readout, starts that ASCOM exposure, and then polls until it’s complete.
Jocelyn: And they also handle calibration frames using a function called run calibration, which acquires tungsten flat field frames and thorium wavelength calibration frames using that same exact exposure and FITS generation path as the science images. That consistency is vital for reliable data.
Subrahmanyan: The deterministic file handling—creating target-specific folders, date-coded night folders, and run folders—is important because it guarantees a predictable data layout. That structure is essential when you’re dealing with a large volume of observations from many different sources.
Vera: And they also detail the performance metrics they gathered on the system itself. They found that the full automated acquisition overhead, from PHD2 looping to the start of the first science exposure, averaged sixty-two seconds, with ninety percent of those cases completing within ninety-two seconds.
Jocelyn: That sixty-two second average for that entire sequence is quite tight, but it shows they managed to keep that overhead manageable during their batch runs. It speaks to the efficiency of the integration they designed into the software framework.
Subrahmanyan: From a theoretical side, that timing gives us a concrete measure of the latency introduced by this automation layer. If you can keep that overhead low, it means you’re not adding significant downtime between targets, which is what makes multi-target observing feasible.
Vera: The paper also clearly outlines where they are still
Conclusion: Vera: So, Spectro Capture is a Python program that automates fiber spectroscopy at small observatories.
Jocelyn: It takes everything from picking the right target to sequencing all the exposures in one workflow.
Subrahmanyan: The authors are working on making this kind of reliable unattended spectroscopy practical for instruments that aren't huge facilities.
Vera: They show results from Shenton Park Observatory where they ran three hundred forty-five science blocks and got ninety-eight point three percent completion.
Jocelyn: That means they managed to get most of the work done without constant human intervention during those batch runs.
Subrahmanyan: What this suggests is that if you treat target finding, guiding, scheduling, and calibration control as one integrated software problem it becomes a viable method.
Vera: The title itself is pretty straightforward—it's about capturing data automatically with software for small setups.
Jocelyn: I think the implication for us right now is that we can get more science out of smaller telescopes without needing a massive team on site every single night.
Subrahmanyan: It moves the goalposts away from needing a giant observatory to getting high-quality results, provided you have stable hardware and good software.
Vera: The authors aren't just saying it works; they're building the actual end-to-end system for data capture right there in Python.
Jocelyn: It’s less about a single trick and more about building this whole automated pipeline, which is pretty impressive engineering work.
Subrahmanyan: But we have to remember the limitations they mentioned. The system needs a stable setup, and it still stops working if there are weather issues or if they can't do that local determination of coordinates quickly enough.
Vera: Exactly. It’s not magic; it’s just a very well-engineered tool that works best when the telescope and mount are behaving predictably.
Jocelyn: So, the real question for us is how much more flexible this software can get before we start needing a human to step in and fix those weather interruptions.
Shenton Park Observatory · International Centre for Radio Astronomy Research · University of Western Australia
astro-ph.IM, astro-ph.SR
Submitted: 2026-05-19
Updated: 2026-11-08
Comments: 16 pages, 13 figures, 3 tables. Final published version in Astronomy and Computing
Journal ref: Astronomy and Computing 57 (2026) 101162
DOI: 10.1016/j.ascom.2026.101162
Code: https://github.com/pluckas/spectro-capture
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Importance score: 89/100
The gist: Spectro Capture is a Python-based software system developed to automate small-observatory fibre-fed spectroscopy…
Key concepts
- Integrated Workflow
- The software combines multiple complex tasks—like finding a target, guiding the telescope to it, and calibrating the data—into a single sequence. This means instead of running separate programs for each step, Spectro Capture manages them together in one continuous process. This integration is key to making unattended observing work smoothly.
- Guiding Logic
- This is the automated process that keeps the telescope pointed at a star while taking pictures. It involves detecting a star within a specific area, using controlled exposures to find it, moving the fiber to that location, and then using adaptive control once guiding is stable. This sequence ensures accurate data collection.
- Data Handling Path
- The system manages the entire process from capturing an image to creating a final, usable FITS file. It doesn't just trigger exposures; it controls settings like binning and camera resets before starting the exposure. This deterministic path ensures that every science image is processed consistently for reliable analysis.
- Batch Runner Script
- This script acts as the main engine for unattended observing, handling the sequential processing of multiple targets. It manages target eligibility based on constraints like hour angle and directs the system through all necessary steps, from initial target finding to final data saving.
Terminology
Summary
Spectro Capture is a Python-based software system developed to automate small-observatory fibre-fed spectroscopy, demonstrating that reliable unattended spectroscopy is practical at a small observatory when target acquisition, guiding, scheduling and calibration control are treated as an integrated software problem The system successfully completed 339 of 345 attempted science target blocks in primary unattended batch runs, achieving a completion rate of 98. p. 03%
System Capabilities and Design
The software integrates target selection, telescope slewing, guide star acquisition, fibre position restoration, calibration and science exposure sequencing within a single observing workflow The framework is built around a modular Python application with core functionality including direct camera control via ASCOM and integration of PHD2 for guiding Key components include the Batch Runner script, which handles sequential target processing, and the Scheduler component, which manages target eligibility based on observing constraints like hour angle The system uses a shared application context to coordinate state and device connections across various scripts
Operational Workflow Automation
The automated workflow involves several distinct steps coordinated by the software, as illustrated in Figure 2 The guiding logic comprises four main steps: confining star detection to a defined region of interest, using a controlled exposure sequence to find the target, restoring the detected star to a stored fibre reference position, and using adaptive exposure control once guiding is established Target finding is handled by an ROI acquisition function in batch runner.py, which uses an exposure ladder to find a star within a configured rectangular region of interest (ROI) Successful centring is verified rather than assumed through checks on the star's proximity to the fibre position
Data Handling and Calibration Management
Spectro Capture controls the full path from camera exposure to reduction-ready FITS file, meaning it does not simply trigger exposures but manages the entire process For each exposure, the routine sets configured binning, resets the camera to a full-frame readout, starts the ASCOM exposure, and polls the camera’s ImageReady state until completion Calibration frame acquisition is handled by run calibration, which acquires tungsten flat field frames and thorium wavelength calibration frames using the same exposure and FITS generation path as science images File handling is deterministic, creating target-specific folders, date-coded ‘night’ folders, and run folders to ensure a predictable data layout
Performance Metrics and Limitations
The operational results show that the full automated acquisition overhead from PHD2 looping to the start of the first science exposure averaged 62 seconds, with 90% of cases completed within 92 seconds The system's success depends on repeatability, suggesting it works best when the telescope, mount, guide camera and spectrograph behave as a stable and calibrated system Limitations include the dependency on local SIMBAD lookup support for pre-slew coordinate resolution and weather handling that currently stops operations rather than treating interruptions as recoverable events Future improvements are scoped to address boundary conditions such as longer faint-target patience and more flexible settle criteria
Conclusion
The development of Spectro Capture suggests that unattended multi-target spectroscopy is moving from an aspiration to a practical operating mode for a variety of suitably stable instruments The software has evolved into an end-to-end system for automated target selection, acquisition and data capture This shift indicates that the assumption that fully automated small-observatory spectroscopy is difficult enough to be out of reach is beginning to change The results presented here suggest that unattended spectroscopy is moving from an aspiration to a practical operating mode for a variety of suitably stable instruments The software does not replace the need for a stable installation, but rather makes use of one The more realistic path to automation is a software layer built around a stable and calibrated instrument
APPENDIX A Code availability
The Spectro Capture source code is publicly available on GitHub, released under a non-commercial source license permitting personal, educational and research use The repository can be accessed at https://github.com/pluckas/spectro-capture The software is provided as source code rather than as a packaged installer, requiring a working Python environment and sufficient familiarity with Python to install dependencies The required Python packages and other dependencies are listed in the repository documentation The Spectro Capture interface is organized into task-based tabs, including a Batch Main unattended observing interface and a Guide PHD2 connection The Status tab provides live status information which is saved to a nightly disk log The Batch Runner tab serves as the main observing interface for timed and scheduled targets The Guide tab provides access to PHD2 controls, configuration, and live guide information The Tools tab allows manual bias/dark utilities and manual lamp control for the eShel tungsten and thorium lamps The Setup Hardware connection and configuration for camera and telescope, ROI centre/size fields are managed in the Setup tab The Dome tab allows manual control through gui dome. p. 2py and dome backend. p. 6
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Refine target acquisition and guiding logic to allow adaptive exposure control based on target brightness, as described by
The two-stage adaptive exposure logic enables Batch Runner to handle targets within the observatory’s practical magnitude limit with remarkable reliability.
-
Implement a more robust method for handling guide star acquisition failure, as indicated by the log analysis showing that
four were associated with guide acquisition or fibre-restore failure
among non-completed targets. Future improvements should focus onlonger faint-target patience and more flexible settle criteria
to mitigate these failures. -
Enhance SIMBAD dependence to improve resilience, as the paper notes,
Live SIMBAD dependence is another operational weakness,
suggesting that developingsome level of local coordinate storage or pre-run caching would make the system less vulnerable to network or service interruptions.
-
Improve weather handling by treating interruptions as recoverable events, as suggested by the limitation that
Weather handling is intentionally conservative
and a future version couldpause, secure the observatory, monitor conditions and resume the remaining observing plan if safe conditions return.
-
Streamline file and directory handling by implementing automatic disabling of completed targets within batch lists to
streamline routine use in a survey-style workflow,
as suggested in the limitations section.
Related papers
- On-sky demonstration of a vector Zernike wavefront sensor in a cascaded adaptive optics system
- Euclid Data Release 1 (DR1): Calibration strategy and observations
- The General Antiparticle Spectrometer (GAPS) Antarctic Balloon Payload
- Detection and excision of radio-frequency interference in radio arrays
- Unified matched-filter and excess-power pipeline for gravitational-wave detector characterization
- Ushering in sub-arcsecond angular resolution observatories in the far-infrared: The case for a double-Fourier interferometer