GWOSC Event Portal

• GWOSC Event Portal is a public web application that facilitates galaxy-targeted electromagnetic follow-up of gravitational wave events.
• It integrates probabilistic galaxy ranking, observability filtering, and interactive mapping to optimize detection of decaying transients in vast localization regions.
• The system employs advanced algorithms for crossmatching galaxies and computing composite probability scores to deliver near real-time prioritized target lists.

The GWOSC Event Portal is a publicly accessible web application designed to facilitate the galaxy-targeted electromagnetic follow-up of gravitational wave (GW) events detected by detectors such as LIGO and Virgo. The portal addresses the central challenge of multi-messenger astronomy—efficient detection and characterization of decaying transients within the large localization regions typical for GW events—by integrating rapid probabilistic ranking of galaxy hosts, observability constraints, and a user-friendly interface for target selection and data access.

1. Functional Overview and Web Application Features

The GWOSC Event Portal provides an interactive interface for astronomers to retrieve, filter, and download ranked lists of galaxy targets most likely to host a detected GW event. Users can select the gravitational wave event and specify the desired containment probability for the localization region (e.g., 99%, 90%, or 50%). The interface visualizes the selected galaxies and GW localization contours on an interactive map.

Further filtering is available based on observability at user-specified coordinates and times, calculated using the astroplan package. An additional detectability option allows users to determine, for each candidate, whether a kilonova of assumed minimum brightness ($M_{\text{kn,min}} = -17$) would be observable at that distance, indicated by a dedicated table column. The system is designed to deliver these features rapidly to any modern HTML5-compatible web browser, requiring no installation and supporting public, immediate access for community follow-up campaigns (Salmon et al., 2019).

2. Algorithmic Framework for Galaxy Ranking

The core backend algorithm executes galaxy crossmatching and prioritization in response to GW triggers:

• Galaxy Sample Selection: The process begins by filtering the GLADE V2 galaxy catalogue to include only galaxies within both the 99% sky localization region (as defined by the LIGO/Virgo probability sky map) and a distance window spanning $\text{DISTMEAN} \pm 5\cdot \text{DISTSTD}$, where $\text{DISTMEAN}$ and $\text{DISTSTD}$ are the mean and standard deviation of the GW source distance estimates.
• Probability Scoring: Every candidate galaxy is evaluated using a composite probability score $S$:
  • Localization Probability ($S_{\text{loc}}$): The product of the GW sky map pixel probability $p_{\text{loc}}$ and a Gaussian distance probability $p_{\text{dist}}$ centered at $\mu_{\text{dist}}$ with spread $\sigma_{\text{dist}}$,

  $$p_{\text{dist}} = N_{\text{dist}} \cdot \exp\left( -\frac{(D - \mu_{\text{dist}})^2}{2\sigma_{\text{dist}}^2} \right).$$ - Luminosity Weight ($S_{\text{lum}}$): The B-band luminosity of the galaxy, derived from apparent magnitude and distance, normalized by the sum over the candidate sample,

  $S_{\text{lum}} = \frac{L_B}{\Sigma L_B}.$ - Overall Ranking: The final score is the product of these components, normalized so that $\sum S = 1$ across all selected galaxies:

  $$S = S_{\text{loc}} \cdot S_{\text{lum}} = (p_{\text{loc}} \cdot p_{\text{dist}}) \cdot S_{\text{lum}}.$$

Probabilities are summarily normalized after computation, and only the top 100 entries by rank are retained for display and export (Salmon et al., 2019).

3. Execution Performance and Automated Operations

The algorithm has been validated on events from LIGO/Virgo O3 and earlier observing runs. Typical execution times for producing a ranked catalog are 20–30 seconds, influenced primarily by FITS file size and network conditions. Subsequent static database upload to Amazon S3 and update of the Heroku-hosted website require an additional $\sim$360 seconds.

Profiling via Python’s cProfile module has demonstrated significant performance improvements with version V1 of the algorithm, notably by confining processing to the highest-probability sky regions and adjusting distance cuts. These optimizations minimize runtime and risk of missing high-priority targets.

The backend is triggered automatically in response to GW triggers during observing runs, enabling near real-time updating of target lists for community observers (Salmon et al., 2019).

4. Technical Architecture and Platform Integration

The system is implemented entirely in Python using the Flask web framework. Major architectural components include:

• Back-End Services: Galaxy retrieval and scoring logic using ligo.skymap for sky map manipulation and crossmatching, pandas for data operations, and astroplan for observing window calculations.
• Data Storage: Ranked galaxy lists are stored in a PostgreSQL database hosted on Amazon S3.
• Front-End Delivery: Interactive tables and maps are rendered via the Flask web server hosted on Heroku. Integration with Aladin DSS enables side-by-side astronomical image visualization.
• Alert Handling: The pyGCN package allows for real-time ingestion of GW triggers via NASA’s Gamma-ray Coordinates Network (GCN).

All functionalities are accessible via the public URL gwtool.watchertelescope.ie (Salmon et al., 2019).

5. User Interaction and Data Utilization

Users interact with the GWOSC Event Portal through a browser-based interface. The workflow includes:

1. Selection of GW event and localization probability threshold.
2. Optional filtering by observatory location, time, and limiting magnitude/observability.
3. Extraction and visualization of the ranked galaxy sample, including names, coordinates, distances, B-band magnitudes, ranking scores, and images.
4. Download of result tables in JSON or ASCII formats for use in follow-up scheduling and observation planning.

The user interface is optimized for rapid public access, intentionally limiting dynamic tables to 100 entries for responsiveness but allowing full catalogue exports when required (Salmon et al., 2019).

6. Planned Enhancements and Future Prospects

Planned enhancements for the GWOSC Event Portal include integration of advanced scheduling and tiling algorithms to optimize follow-up coverage, migration to private server infrastructure to reduce database update latency, and refinement of the spatial selection methodology to employ the full three-dimensional localization credible volume rather than two-dimensional projections with distance limits.

Such improvements are expected to increase the speed, fidelity, and scientific return of coordinated multi-messenger observations in future GW observing runs, particularly by strengthening the coupling between real-time GW alerts and electromagnetic follow-up infrastructure (Salmon et al., 2019).