Skip to content

How Skynet works

Skynet operates a highly heterogeneous fleet of optical and radio telescopes as a single, coordinated instrument. Making that work requires a common way to describe astronomical data requests, instruments, and observations — so a request can be matched to any suitable hardware in the network, no matter how differently each telescope is designed or controlled.

This page is an overview of that model: how Skynet represents targets, observation configurations, data requests, and results. By describing observations in terms of what data is needed rather than how a particular instrument obtains it, Skynet can schedule one request across instruments that vary widely in capability and control system — and store the results in a consistent, comparable form.


Target Definitions: Specifying Astronomical Objects

Section titled “Target Definitions: Specifying Astronomical Objects”

Observations require precise definitions of targets, which are represented in a unified model applicable to all observation types.

Each target includes a position that can be one of several forms:

  • Fixed Position – Defined by equatorial (RA/Dec), ecliptic, galactic, horizontal, or topocentric coordinates.
  • Catalog Position – A reference to an object in Skynet’s moving-object catalog (major solar system bodies, planetary satellites, comets and asteroids from the Minor Planet Center, NEOCP candidates, and NORAD satellites), whose position is computed when needed.
  • Ephemeral Position – Time-dependent coordinates for moving objects (e.g., asteroids, comets).
  • Orbital Position – Derived from Keplerian elements for solar system objects.

Optical imaging observations also carry models of the target’s brightness. These live on the optical imaging configuration (below), not on the target itself, because they are exposure references rather than properties of the object:

  • Brightness Model – Defines the target’s brightness and is referenced when calculating exposure times from an optical imaging data request’s target signal-to-noise ratio (SNR) or target full-well fraction. Brightness models can optionally include temporal and spectral components that describe how the target’s brightness evolves as a function of time and frequency.
  • Saturation Model – Sets upper limits on exposure times to help prevent overexposure and saturation of sources during optical imaging.

Observation configurations define environmental and operational constraints that apply to specific types of data requests.

The optical imaging configuration specifies environmental and field constraints used for all optical imaging data requests. It includes:

  • Field of View (FOV) constraints
  • Moon phase and moon separation requirements
  • Sun elevation and target elevation limits
  • Tiling and coverage strategy
  • The brightness and saturation models described above

These parameters ensure that optical imaging observations are scheduled under suitable conditions and maintain consistent image quality across the network.

The radio tracking configuration defines parameters for tracking observations, including:

  • Whether data should be collected on-source only, or on-source and off-source (on then off, or off then on)
  • The offset of the off-source position from the target, in beamwidths by default

Radio tracking data requests consist of the following properties:

  • Spectroscopy windows
  • Minimum spectral resolution
  • Polarization requirements
  • Maximum integration time
  • Observation duration
  • Continuum requirement

The radio mapping configuration defines how data are collected across multiple pointings to construct a map of a sky region. It includes properties describing the offset pattern used during observation:

  • Daisy Pattern – Optimized for compact or point-like sources.
  • Raster Map – Used for larger or extended regions of the sky.

Radio mapping data requests include:

  • Spectroscopy windows
  • Minimum spectral resolution
  • Polarization requirements

These configurations ensure that both optical and radio observations can be defined consistently while remaining flexible enough to accommodate the unique requirements of each observation type.


Observation Requests: A Unified Way to Define Data Needs

Section titled “Observation Requests: A Unified Way to Define Data Needs”

Skynet defines a unified data request format that lets users specify what kind of data they need rather than how to obtain it. This separation allows Skynet’s scheduling system to determine which instruments and configurations can fulfill the request, no matter how different those instruments may be.

  • Sized by exactly one of: a fixed exposure time, a target signal-to-noise ratio (SNR), or a target fraction of full-well depth. The observation editor offers the first two; the full-well mode is available through the API.
  • When using SNR-based or full-well exposure, a brightness model is required.
  • Can be a set of independent images, or a series: one contiguous run of images on one camera for a set duration.
  • Include spectroscopy windows (center frequency and bandwidth), a minimum spectral resolution, a maximum integration time, and the duration.
  • Choose a record mode: spectrometer output, or pulsar search or fold.
  • Can require polarization (linear or circular) and a continuum spectrometer.

Observation Results: Consistent, Comparable Data Output

Section titled “Observation Results: Consistent, Comparable Data Output”

As an observation runs, Skynet stores its results in a consistent structure, so data from different instruments and observatories can be compared and combined. Each result is one sample of one data request in one epoch: a single image, or a whole series.

Observation results may include:

  • Captured exposures – Raw and processed image files for optical observations.
  • Spectral data – Spectrograms and calibrated data for radio spectroscopy.
  • Final result files – Combined or stacked data products after post-processing.

Because results share a common definition, data collected from one observatory can be compared, combined, or analyzed alongside data from another—without translation or reformatting.


This model exists to solve one of the most fundamental challenges in modern astronomy: how to make a globally distributed, diverse set of instruments behave as a single coherent observatory.

Without a shared model, every telescope and control system would remain siloed—each speaking its own language, with its own data structures and observation formats. That would make it impossible to automatically coordinate or distribute observations across the network.

By giving observations, instruments, and data requests a shared vocabulary, Skynet’s data model is the foundation that lets a network of heterogeneous observatories act as one agile, adaptive, and globally aware instrument.


This page is a conceptual overview of the data model. For a field-level breakdown of every record the public API accepts and returns, see the API reference, which is generated from the API’s OpenAPI document at build time.

---
config:
  flowchart:
    defaultRenderer: elk
  layout: elk
---
flowchart TB
  %% End-to-end lifecycle with feedback loop
  subgraph U["Create Observation – Multiple Inputs"]
    direction TB
    WebAdv["Website · Advanced Editor (pro)"]
    WebProj["Website · Project-Based Wizard (guided)"]
    Script["Scripts & Notebooks"]
    Third["Third-Party Apps"]
    Integr["Integrations (e.g., GCN)"]
  end

  subgraph A["Access Control"]
    direction TB
    Account["Observing Accounts"]
    Grants["Grants & Quotas"]
  end

  subgraph Q["Scheduling"]
    direction TB
    Queues["Telescope Queues (TOO, General, etc.)"]
    Tasks["Scheduled Tasks"]
  end

  subgraph T["Execution"]
    direction TB
    Telescopes["Telescopes"]
    Upload["Raw Data Upload"]
  end

  subgraph P["Processing"]
    direction TB
    Pipeline["Pipelines (calibration, stacking, photometry, etc.)"]
    Products["Processed Data Products"]
  end

  subgraph D["Data Tools"]
    direction TB
    ProTools["Analysis Tools (time-series, photometry, catalogs)"]
    EduTools["Education/Projects (color composites, cluster age/metallicity)"]
    APIs["APIs & Downloads"]
  end

  subgraph F["Results"]
    direction TB
    Results["Results & Visualizations"]
  end

  U --> A --> Q --> T --> P --> D --> F
  F -->|plan the next observation| U
---
config:
  flowchart:
    defaultRenderer: elk
  layout: dagre
---
flowchart TB
  %% "For Everyone" interface map
  subgraph FE["Interfaces & Goals"]
    direction LR
    subgraph WebUI["Website Interfaces"]
      direction TB
      Adv["Advanced Editor (pros)"]
      Proj["Project-Based Wizard (guided)"]
    end
    subgraph Prog["Programmatic"]
      direction TB
      Scripts["Scripts/Notebooks"]
      Apps["Third-Party Apps"]
      Integrs["Integrations (alerts, etc.)"]
    end
    Goals["User Goals"]
  end

  Goals -->|science-grade| Adv
  Goals -->|guided learning| Proj
  Goals -->|automation| Scripts
  Goals -->|ecosystem| Apps
  Goals -->|event-driven| Integrs

  subgraph Tools["Data Tools Spectrum"]
    direction TB
    Edu["Education/Project Tools\n• Color composites\n• Globular cluster age/metallicity"]
    Pro["Professional Tools\n• Time-series photometry\n• Catalog cross-match\n• Pipeline QC"]
  end

  Proj --> Edu
  Adv --> Pro
  Scripts --> Pro
  Apps --> Pro
  Integrs --> Pro
sequenceDiagram
  autonumber
  participant HS as High-School User
  participant PA as Pro Astronomer
  participant Web as Website
  participant S as Scheduler/Queues
  participant Tel as Telescope
  participant Stor as Data Storage
  participant Pipe as Pipeline
  participant Tools as Data Tools

  HS->>Web: Choose guided project (e.g., color composite)
  PA->>Web: Configure advanced observation
  Web->>S: Submit observation via Account → Queue
  S->>Tel: Assign & dispatch tasks
  Tel->>Stor: Upload raw images/data
  Stor->>Pipe: Trigger processing
  Pipe->>Tools: Publish processed products
  Tools-->>HS: Education-oriented outputs
  Tools-->>PA: Analysis products (photometry, catalogs)