Astro-COLIBRI: Real-Time Multi-Messenger Platform
- Astro-COLIBRI is a real-time multi-messenger platform that centralizes and harmonizes heterogeneous alerts for rapid transient follow-up.
- The system employs a modular, cloud-based architecture with a public RESTful API, real-time databases, and cross-platform clients to ensure timely data delivery.
- It integrates advanced scheduling tools like tilepy and supports citizen-science workflows to optimize observational strategies for gravitational waves, GRBs, and other phenomena.
Astro-COLIBRI—“The COincidence LIBrary for Real-time Inquiry”—is a real-time platform for multi-wavelength and multi-messenger transient astronomy that ingests heterogeneous alerts, filters them by user-specified criteria, situates each event in scientific context, and supports follow-up through web, smartphone, and API interfaces (Reichherzer et al., 2021, Reichherzer et al., 2021). Across successive descriptions by Reichherzer, Schüssler, collaborators, and related proceedings, it is presented as a top-level or downstream coordination layer within the broader alert ecosystem, covering phenomena such as gamma-ray bursts, fast radio bursts, gravitational waves, high-energy neutrinos, active galactic nucleus flares, optical transients, novae, and supernovae (Schüssler et al., 2023, Avila et al., 28 Apr 2026).
1. Scientific role and position in the transient-alert ecosystem
Astro-COLIBRI was developed to address a central operational problem of time-domain astrophysics: critical information about a transient event’s type, significance, localization, likely counterparts, and visibility is distributed across multiple alert streams, brokers, catalogs, and human-written reports. The platform centralizes and harmonizes that information in real time so that observers can move rapidly from alert receipt to an actionable follow-up decision (Schüssler et al., 2024, Schüssler et al., 2021).
The platform is repeatedly described not as a replacement for brokers such as GCN, TNS, Fink, AMON, or Four Pi Sky, but as a coordination and decision-support layer that sits downstream of them. In this formulation, upstream systems distribute discovery notices, whereas Astro-COLIBRI fuses alerts, adds contextual information, evaluates observing feasibility, and presents results through a single graphical and programmatic interface (Reichherzer et al., 2022, Avila et al., 28 Apr 2026).
Its scientific scope expanded across the 2021–2026 literature while preserving the same core purpose. Early descriptions emphasized GRBs, neutrinos, GWs, AGN flares, FRBs, and TDEs; later papers added stronger support for optical-transient photometry, discussion and coordination tools, and explicitly structured citizen-science workflows (Reichherzer et al., 2021, Schüssler et al., 9 Jul 2025, Schüssler et al., 2024). The public release is described in the 2022 “Astro-COLIBRI 2” paper as having occurred in August 2021 (Reichherzer et al., 2022).
2. Architecture, data model, and real-time dissemination
Published descriptions consistently present Astro-COLIBRI as a modular, cloud-based system built around a public RESTful API, real-time databases, a cloud-based alert system, and web plus iOS/Android clients (Reichherzer et al., 2021, Avila et al., 2 Feb 2026). The architecture evolved in deployment details—early descriptions place the Python/Flask API on Heroku, later ones on AWS or a cloud-based cluster—but the architectural pattern remains stable: continuous listening to alert streams, normalization into an internal schema, database update, contextual enrichment, and targeted user notification (Reichherzer et al., 2021, Lavergne et al., 2023, Schüssler et al., 9 Jul 2025).
| Layer | Role | Implementations described in the literature |
|---|---|---|
| Ingestion | Receive and normalize alerts | VOEvent listeners, Comet, Kafka, custom Python parsers, email parsing |
| Core services | Process, enrich, store, and serve events | Python/Flask REST API; MongoDB plus Firebase-backed real-time services |
| Clients and notification | Visualization and alert delivery | Flutter web/iOS/Android clients; Firebase Cloud Messaging |
The ingestion side handles both machine- and human-generated content. The platform is described as listening to VOEvent streams, GCN Notices, GCN Circulars, TNS notifications, email-based alerts, and brokered streams; later technical descriptions also name GCN JSON notices over Kafka, Fink, and CBAT (Reichherzer et al., 2021, Avila et al., 2 Feb 2026, Schüssler et al., 9 Jul 2025). A 2026 technical overview specifies a dedicated AWS listener that translates incoming alerts into a unified JSON-based Astro-COLIBRI format and forwards them to the API via POST requests, with updates recognized by trigger-ID de-duplication (Avila et al., 2 Feb 2026).
The storage model is likewise dual. The 2021 core description uses MongoDB as the authoritative static database and Firebase Firestore as the real-time database; later papers describe MongoDB paired with Firebase Firestore or Firebase Realtime Database for low-latency client streaming (Reichherzer et al., 2021, Schüssler et al., 14 Jul 2025, Avila et al., 2 Feb 2026). A 2026 overview states that the archive exceeds 100,000 events dating back to 2016 (Avila et al., 28 Apr 2026).
On the client side, Astro-COLIBRI uses Flutter to provide a unified codebase for iOS, Android, and web, while Firebase Cloud Messaging distributes push notifications according to user-selected topics or predefined notification streams (Reichherzer et al., 2021, Reichherzer et al., 2021). The smartphone app is described as functionally equivalent to the web interface for core operations such as notification handling, event browsing, and GW follow-up visualization (Lavergne et al., 2023).
3. Alert fusion, contextualization, filtering, and visualization
A defining characteristic of Astro-COLIBRI is that it does not merely relay alerts. It merges multiple notices referring to the same astrophysical event, attaches related notices and circulars, and enriches the event with contextual links, archival information, source catalogs, photometry, and observability summaries (Reichherzer et al., 2021, Schüssler et al., 2024). The platform’s “cone search” is a central contextualization tool: in the 2021 description its default search radius is 10 degrees, results are ordered by angular separation, and the sky map displays the alert with its uncertainty region while listing nearby sources and transients (Reichherzer et al., 2021).
The mathematical primitives used for these spatial and localization-aware operations are standard spherical-astronomy and probability-coverage relations. The 2021 and 2025 technical descriptions explicitly give the angular separation as or, equivalently, by the spherical law of cosines, and describe field-of-view probability coverage as either or its discrete HEALPix form (Schüssler et al., 2021, Schüssler et al., 9 Jul 2025).
Filtering is hierarchical and increasingly granular in later versions. Early papers describe filtering by wavelength or messenger, observatory, and event type; later interface descriptions add top-level buttons for observatories and event classes and long-press access to detailed filters such as optical magnitude at detection, neutrino “signalness,” and gravitational-wave significance, localization uncertainty, and classification probabilities including BBH, BNS, NS-BH, and MassGap (Reichherzer et al., 2021, Schüssler et al., 2024). The platform also includes a “science mode” that exposes planning-critical detail such as visibility plots, historical comparisons, and instrument links (Reichherzer et al., 2022).
Visualization combines timelines, event lists, skymaps, information panels, and external links. The web interface is described in multiple papers as including a timeline of recent transients, an event-tile or event-list panel, a central sky map with selectable coordinate systems and projections, a cone-search panel, an information panel with plots and photometry, and panels linking to official reports and external tools (Schüssler et al., 2023, Avila et al., 28 Apr 2026). The platform can display Fermi-GBM light curves for GRBs, IceCube public reconstructions for neutrino events, and photometric summaries for optical transients (Avila et al., 2 Feb 2026).
A major later addition is automatic photometric aggregation from ATLAS, ASAS-SN, and ZTF via the Fink broker. The 2025 technical overview describes publication-quality light-curve generation, CSV export, twice-daily updates for classified optical transients detected in the last 30 days, and computed “first,” “peak,” and “last” observations per telescope and filter, which then feed downstream filtering (Schüssler et al., 9 Jul 2025). That same overview also adds cross-links to SIMBAD, NED, VizieR, Gaia, and WISE color–color diagnostics for host characterization (Schüssler et al., 9 Jul 2025).
4. Observability, scheduling, and gravitational-wave follow-up
From its earliest papers, Astro-COLIBRI is described as computing observing conditions for a large selection of observatories around the world in real time (Reichherzer et al., 2021, Reichherzer et al., 2021). The 2021 core description states that detailed 24-hour visibility plots include target altitude as a function of time, Sun altitude, Moon altitude, Moon phase, and Moon–source separation, alongside a monthly overview for planning (Reichherzer et al., 2021). Later work broadens this to a curated set of relevant facilities, the IAU database of more than 2600 professional and amateur observatories, and arbitrary user-defined sites, with observability windows also available in JSON (Schüssler et al., 9 Jul 2025).
Custom observatory handling is central to both professional and amateur use. The interface exposes pre-defined observatories grouped by wavelength range, an IAU-database search with autocompletion, and custom observatories based on device GPS or manually entered coordinates and parameters; saving custom entries requires a free account associated with a valid email address (Schüssler et al., 2023, Schüssler et al., 2024). A 2026 interface description adds multi-observatory, multi-night, and multi-source visibility curves, as well as watchlists and manually created sky-position entries (Avila et al., 28 Apr 2026).
The most developed scheduling capability concerns poorly localized events, especially gravitational waves. Astro-COLIBRI integrates the tilepy API as a cloud-based service and exposes it directly in the web and mobile interfaces, returning optimized schedules in JSON and displaying pointings or tiles in a table and as overlays on the localization map (Schüssler et al., 2023, Lavergne et al., 2023). Different interface descriptions mention localization contours of 50% and 90% containments for GW events and, in later web layouts, 68% and 95% contours for broadly localized events (Schüssler et al., 2023, Avila et al., 28 Apr 2026).
The tiling logic is described qualitatively in the 2023 GW-follow-up paper. In the 2D approach, tilepy uses the GW sky-localization map directly, iteratively selecting the pointing that maximizes covered localization probability and “masking” the selected probability before choosing the next tile. In the 3D approach, it combines the GW distance posterior with the GLADE+ galaxy catalog, assigning probabilities to galaxies and selecting tiles with the highest cumulative galaxy probability. The paper states that the 3D approach is effective up to a few hundred Mpc, beyond which catalog completeness degrades (Lavergne et al., 2023).
This integration inherits a substantial observational pedigree. Tilepy was developed for the H.E.S.S. gravitational-wave rapid follow-up program, was used operationally during the LIGO/Virgo O2 and O3 runs, and is described as currently employed by CTA’s LST-1 in O4 (Lavergne et al., 2023). The same paper states that tilepy’s optimized strategies enabled H.E.S.S. to be the first ground-based facility to point to the true counterpart location of GW170817, with observations starting 5.3 hours after the merger; the subsequent campaign covered 0.22 to 5.2 days post-merger and detected no significant very-high-energy gamma-ray emission (Lavergne et al., 2023).
5. Citizen science, pro-am workflows, and community infrastructure
A persistent theme in the Astro-COLIBRI literature is that some transient classes are accessible to small-aperture telescopes and geographically distributed observers, making the platform suitable for structured citizen-science participation (Schüssler et al., 2023, Schüssler et al., 2024). The dedicated citizen-science papers emphasize two alert streams in particular: “Bright optical transients (mag<18)” and the Unistellar stream for “bright and early optical transients,” whose magnitudes at detection are usually required to be below 16.3 (Schüssler et al., 2023). The first stream is explicitly defined by the magnitude at detection and does not directly account for subsequent source evolution (Schüssler et al., 2023).
These amateur-facing workflows do not suppress caveats. The Unistellar early-bright candidate stream is selected by the ALeRCE broker, and the paper explicitly warns that this procedure has “large uncertainties” because it often relies on limited information, typically only a single flux measurement exceeding archival observations; as a result, the fraction of events due to mis-classifications can be “sizable” (Schüssler et al., 2023). That limitation is one of the clearest places where the literature distinguishes accessibility from reliability.
Citizen-science coordination broadened markedly after the initial public release. Later papers describe a discussion forum, collaboration with RAPAS and BHTOM, a weekly Top-10 transient-events list published on the forum in collaboration with RAPAS, and integration with networks such as ProAm Gemini, the IAU Citizen Science Program, and the United Nations Office for Outer Space Affairs’ Open Universe initiative (Schüssler et al., 2024, Schüssler et al., 14 Jul 2025). Astro-COLIBRI is also described as active in the IAU Executive Committee Working Group on Professional–Amateur Relations in Astronomy and as the alert-distribution engine for the Unistellar citizen-science program supported by the SETI Institute (Schüssler et al., 14 Jul 2025).
The platform supports direct operational pathways from alert to telescope. One 2025 paper describes a community-built N.I.N.A. plugin that connects to the Astro-COLIBRI API and auto-fills N.I.N.A.’s framing assistant and sequencer. The same paper states that a single tap in the Astro-COLIBRI app can command a Unistellar telescope to slew and start collecting data, with observations uploaded to a SETI Institute science server for professional analysis (Schüssler et al., 14 Jul 2025).
Case studies illustrate the scientific value of this pro-am integration. For SN 2023ixf in M101, the 2023 advanced-platform paper reports that Unistellar telescopes of 11.4 cm aperture produced a dense light curve with 3.3-hour average sampling over 35 days, while early amateur detection constrained the explosion time window to about one hour (Schüssler et al., 2023). For SN 2025coe, a 2025 citizen-science paper describes amateur discovery by Koichi Itagaki, immediate dissemination by Astro-COLIBRI, RAPAS follow-up, and a densely sampled multiband light curve produced in combination with ZTF and ATLAS photometry (Schüssler et al., 14 Jul 2025).
6. Performance, limitations, and development trajectory
The most quantitative performance characterization appears in the 2021 core technical paper. There, push-notification delivery through Firebase Cloud Messaging is given as approximately 335 ms, a transient cone search through Firestore as ms, a manual cone search via the API as ms, monthly visibility-plot generation as ms, and detailed 24-hour visibility plotting as ms (Reichherzer et al., 2021). The contemporaneous conference paper also states that the platform achieved almost 100% uptime during an extensive beta-testing phase and that push notifications arrived well under one second from alert ingestion (Schüssler et al., 2021). Later papers usually replace benchmarks with qualitative phrases such as “few hundred milliseconds,” “negligible latency,” “minimal cadence,” or “extremely high availability” (Reichherzer et al., 2021, Avila et al., 2 Feb 2026, Schüssler et al., 9 Jul 2025).
Several recurring limitations are explicit in the literature. Human-generated reports, particularly GCN Circulars and ATels, introduce unavoidable upstream latency and inconsistency relative to machine-generated notices (Reichherzer et al., 2021, Reichherzer et al., 2022). The 2021 conference paper states that weather is not part of the current public platform, so visibility emphasizes geometric and celestial constraints rather than meteorological conditions (Schüssler et al., 2021). On the GW side, the tilepy 3D strategy depends on GLADE+ completeness and degrades beyond a few hundred Mpc (Lavergne et al., 2023). On the citizen-science side, the ALeRCE-selected Unistellar stream carries a sizable mis-classification risk (Schüssler et al., 2023).
The platform’s source-code status is also unevenly documented. Astro-COLIBRI consistently exposes a public RESTful API, but several papers state that the platform’s code availability is not specified, while the 2022 “Astro-COLIBRI 2” paper explicitly says that the source code is currently private, with public releases planned in the future (Reichherzer et al., 2022, Schüssler et al., 2024). By contrast, tilepy is openly available and documented as public code with a hosted API (Schüssler et al., 2023, Lavergne et al., 2023).
The development trajectory is oriented toward deeper automation and broader interoperability. Planned or ongoing items across the literature include direct parsing of emailed ATels, additional broker integrations, more customizable push-notification criteria, new scheduling strategies in tilepy, enhanced filtering, collaboration-specific ToO submission, richer photometric tools, and deep learning-based natural language processing for human-written alerts such as GCN Circulars, ATels, and TNS AstroNotes (Reichherzer et al., 2022, Lavergne et al., 2023, Reichherzer et al., 2021). Taken together, these papers portray Astro-COLIBRI not as a static broker interface but as an evolving coordination layer whose defining characteristics are real-time fusion, observer-centric contextualization, and operational follow-up support across the full multi-messenger transient landscape.