Acquisition, Pointing, and Tracking (APT)

• Acquisition, Pointing, and Tracking (APT) is a comprehensive workflow that acquires targets, precisely points instruments, and tracks observations to ensure complete and reliable survey data.
• It employs automated command file generation, dynamic scheduling with quantitative scoring, and real-time feedback to optimize exposure and observation quality.
• APT’s integrated approach, featuring adaptive feedback and a unified database, minimizes data gaps while maximizing scientific throughput and operational resilience.

Acquisition, Pointing, and Tracking (APT) encompasses the full operational cycle required for advanced astronomical surveys, precision guidance systems, and autonomous observatories to (1) acquire new targets or survey fields, (2) precisely point the instrument’s optical or sensor axis, and (3) track the evolution of those targets across time for both data completeness and quality control. APT integrates deterministic scheduling, adaptive feedback, and real-time state awareness to ensure both survey efficiency and scientific reliability. In the context of J-PLUS at the Observatorio Astrofísico de Javalambre (OAJ), the APT paradigm is implemented via an end-to-end workflow featuring automated sequencing, dynamic rescheduling, and robust survey-state feedback through a dedicated tracking tool and feedback software system (Civera, 2022).

1. Acquisition Subsystem: Command Generation and Exposure Calculation

The acquisition subsystem is responsible for translating science priorities and field selections into actionable telescope command files. The workflow proceeds as follows:

• Input Parameters: Each observation is defined by a set of target coordinates (Right Ascension, Declination), a set of required filters $f \in F$, and a user choice between full or partial filter sets.
• Command File Production: The system generates a telescope-ingestible CSV file; each entry encodes:
  • Filter identifier $f$
  • Calculated exposure time $t_{exp}(f)$
  • Telescope RA/Dec
  • Detector parameters (e.g., binning, readout speed)
  • Observation group identifier
• Exposure Time Model: Exposure time per filter is analytically determined to achieve uniform survey depth, compensating for lunar phase ($\phi$), moon-target separation ($d_{Moon}$), and ambient sky brightness. The exposure model:

$$t_{exp}(f) = t_0(f) \times 10^{0.4 [m_{lim}(f) - m_{sky}(f)]} \times \left[ 1 + \alpha_f \frac{1+\cos\phi}{2} \right] \times \left( \frac{d_0}{d_{Moon}} \right)^2$$

This ensures systematic control over varying background and lunar conditions.

• Interface and Execution: Once in CSV form, command files are uploaded to the JAST80 Instrument Control System (ICS) REST API, which parses, optimizes execution, configures hardware (filter wheel, detector), and streams images to the survey’s data pipeline (Civera, 2022).

2. Pointing Selection, Scoring, and Scheduling Optimization

Efficient field selection and prioritization are achieved by a quantitative scoring framework and a modular scheduling architecture:

• Scoring Function: Each candidate field $i$ is assigned

$$S_i = w_P \left(\frac{3 - Pri_i}{3}\right) + w_V \left(\frac{\Delta t_{vis,i}}{T_{night}}\right) - w_M P_{M,i} - w_C \left(\frac{X_i - 1}{X_{max} - 1}\right)$$

• $Pri_i$: science priority (0=highest)
• $\Delta t_{vis,i}$: field's above-horizon window
• $f$0: moon-avoidance penalty
• $f$1: field median airmass
  • Constraints: A field is scheduled only if $f$2 and $f$3 (example default), imposing astronomical and operational constraints.
  • Horizon and Visibility: Given field declination $f$4 and site latitude $f$5, the observable window is computed, solving

$f$6

and window duration via the hour angle $f$7.

• Scheduling Core: A REST-based scheduler generates nightly observing blocks from the ranked list (greedy or integer-programming maximization over $f$8 under total exposure/time budget), streams the resulting sequence as command files to the telescope via a unified architecture: Python Flask for orchestration, NumPy for scoring, open-source solvers for block optimization, and a C# command generator (Civera, 2022).

3. Tracking, Feedback, and Survey-State Logic

Automated, robust survey execution is underpinned by a feedback software stack systematically evaluating exposure and survey completeness:

• Dataflow: As raw FITS images are ingested, the pipeline computes quality metrics and status flags, which are ingested into a central PostgreSQL database.
• Filter-Level Status: Each exposure/filter combination is assigned a discrete status:
  • NEVER OBSERVED
  • RAW NO OK (insufficient good raw frames)
  • PROC NO OK (reduction failure)
  • PROC OK (pending further checks)
  • PRJ REQ NO OK (fails depth/seeing/SNR)
  • NOT NUM FILTERS (incomplete filter set)
  • OBS VALIDATED
• Pointing-Level Aggregation: For a given field, if any filter is RAW NO OK, PROC NO OK, or PRJ REQ NO OK, the field status is set TO_BE_REOBS (re-observation required); if any filter is NEVER OBSERVED, the status is INCOMPLETE; all filters OBS_VALIDATED implies field OBS_VALIDATED; else, status is OBS_PENDING.
• Feedback Loop: The feedback system updates the survey’s StatusTable in real time. Fields requiring re-observation or marked incomplete are automatically reincorporated into candidate lists for subsequent nights, closing the loop between data quality assessment and scheduling (Civera, 2022).

4. System Integration and Real-Time Dataflow Architecture

APT in the J-PLUS context is architected for full automation, integrated data provenance, and real-time user awareness:

• Unified Database: All system state (PointingCatalog, StatusTable, command files, raw/reduced frame metadata, quality metrics) is maintained in a single PostgreSQL instance, ensuring consistency across acquisition, pointing, and tracking logic.
• RESTful Services: Scheduler, feedback agent, user portal/UI, and the telescope ICS communicate via REST APIs.
• Dataflow Sequence:

1. Scheduler loads current field priorities and statuses.
2. Scheduling logic generates optimized observation blocks.
3. Command files trigger telescope operations.
4. ICS/telescope write real-time exposure metadata.
5. Reduction pipeline computes quality metrics.
6. Feedback software updates statuses for both user and scheduler consumption.

• Operational Parameters: Typical defaults include $f$9 (airmass $t_{exp}(f)$0 1.5), moon separation $t_{exp}(f)$1, scoring weights $t_{exp}(f)$2, and a minimum of 3 frames per filter (Civera, 2022).

5. Automation and Survey Robustness

The APT cycle is engineered for resilience and survey-completeness through rigorous automation:

• End-to-End Loop: Fields needing observation/repeat (from quality assessment) are seamlessly fed back into the next night's queue, minimizing operator overhead and survey gaps.
• Dynamic Scheduling: The scheduler is agnostic to specific failures or interruptions; only the quantitative status flags modulate candidate lists and execution sequences.
• Immediate State Visibility: Status updates propagate to both operators and the status-aware portions of the scheduling software, ensuring that current survey completeness, remaining fields, and real-time execution problems are visible through the Web application (Civera, 2022).

6. Significance and Broader Implications

The J-PLUS implementation of APT exemplifies a state-of-the-art, feedback-driven orchestration for large-scale surveys:

• Scientific Throughput: By incorporating prioritization, adaptive exposure calculation, strict visibility/moon constraints, and tight integration of observational quality, throughput is maximized without sacrificing data uniformity.
• Quality Assurance: The strict feedback process—down to per-filter validation—prevents survey holes and systematically eliminates sub-par data without manual intervention.
• Generalizability: The architectural principles (automated command-file generation, feedback-driven scheduling, integrated status tracking, modular scoring and scheduling) are applicable to any survey or observatory requiring real-time dynamic resource allocation and autonomous quality control within APT cycles (Civera, 2022).