OTTER: Open Multiwavelength Transient Repository

• OTTER is a multiwavelength transient event repository that unifies published metadata and photometry, focusing initially on tidal disruption events (TDEs).
• It uses a JSON-based, document-centric data model optimized for storing heterogeneous photometric data across the full electromagnetic spectrum.
• OTTER’s open-source API and web interface enable streamlined curation, retrieval, and integration into established astrophysical transient analysis workflows.

The Open mulTiwavelength Transient Event Repository (OTTER) is a publicly available catalog of published transient event metadata and photometry designed to make multiwavelength analyses of astrophysical transients more efficient and effective. Its data schema is optimized for the storage of multiwavelength photometric datasets spanning the entire electromagnetic spectrum, and the infrastructure includes open source software, an application programming interface (API), and a web application for viewing, accessing, and analyzing the dataset. In its initial release, OTTER presents the largest ever photometric archive of tidal disruption events (TDEs), including $\gtrsim 80{,}000$ observations of 232 TDEs spanning from radio to X-ray wavelengths (Franz et al., 5 Sep 2025).

1. Scope and repository concept

OTTER is defined as an infrastructure for archiving published transient event metadata and photometry in a form tailored to multiwavelength work. The repository is explicitly motivated by the centrality of multiwavelength analyses to the physics of astrophysical transients and by the need to assemble heterogeneous measurements into a unified framework (Franz et al., 5 Sep 2025).

A distinguishing property of the repository is that its schema is optimized for photometric datasets across the full electromagnetic spectrum rather than for a single bandpass regime or a narrow class of events. The initial scientific focus is the TDE population, but the infrastructure was built with the goal of archiving additional transient data, including supernovae, gamma-ray bursts, fast blue optical transients, and fast radio bursts, in the future (Franz et al., 5 Sep 2025).

A common misconception would be to treat OTTER as only a static TDE list. The released system is broader than that: the TDE catalog is the initial release and the first large-scale scientific demonstration, whereas the underlying repository, software, and upload workflow are intended to support continuing maintenance and future ingestion of newly published data (Franz et al., 5 Sep 2025).

2. Data model and database architecture

OTTER stores each transient in a single JSON document. The top-level keys are name, coordinate, distance, date_reference, classification, host, and photometry. This document-centric model places object metadata and measurement arrays in the same record, allowing a transient to be represented as a single structured entity rather than as a set of disconnected tables (Franz et al., 5 Sep 2025).

Within the photometry array, entries can encode optical and UV magnitudes, radio measurements, and other wavelength-domain observations through fields such as mjd, uaname, wavelength, freq, value, valueerr, and flux, together with associated units and citation keys. The schema also stores correction states in boolean flags: corr_host for host-subtraction, corr_k for k-correction, corr_ext for Galactic extinction, and corr_mb for Milky Way broadband applying (Franz et al., 5 Sep 2025).

The representation explicitly links filter identity to effective wavelength or frequency and includes the AB-system conversion

$$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$

with $F_\nu$ defined as the flux density per unit frequency. This makes the schema suitable for analyses that require conversion between magnitude and flux representations or aggregation of observations from different facilities into a common spectral-energy-distribution view (Franz et al., 5 Sep 2025).

The backend is an ArangoDB document store. Each JSON document is directly ingested as one ArangoDB document, and documents are automatically sharded across the cluster. The photometry array is stored as an embedded sub-document, enabling AQL queries that iterate over a transient’s measurements while filtering by units or wavelength-domain attributes (Franz et al., 5 Sep 2025). This suggests that OTTER prioritizes co-location of event metadata and time-domain measurements over a fully normalized relational design.

OTTER also defines classification semantics at two levels. Each classification entry carries an integer confidence flag $C \in \{0,\dots,4\}$, where $C=0$ denotes an unverified or intermediate candidate, $C=1$ a photometric-only classification including ML algorithms, $C=2$ a spectroscopic classification in a non-refereed venue, $C=3$ a bronze-quality spectrum in a refereed publication, and $C=4$ a silver/gold spectroscopic quality in a refereed publication. In addition, automatically derived meta-flags per transient include discovery_method in {"optical","xray","radio"}, spec_classed in $\{0,\dots,3\}$, and unambiguous in {true,false}. These flags allow refined sample cuts, such as selecting only unambiguous, gold-spectro TDEs (Franz et al., 5 Sep 2025).

3. Programmatic and web access

The API is exposed at https://otter.idies.jhu.edu/api/v1. Major endpoints include GET /transients, which supports parameters such as name, ra, dec, zmin, zmax, classification, and has_photometry; GET /transients/{name} for the full JSON document of a transient; GET /transients/{name}/photometry with query parameters including filter, wave_min, wave_max, tmin, tmax, and units=("mag","flux"); POST /contributions/upload; and GET /contributions/{upload_id} for vetting status and diffs. GET /transients/{name}/spectrum is listed as coming soon (Franz et al., 5 Sep 2025).

Authentication uses an API key supplied via the HTTP header X-OTTER-APIKEY after free registration. The default rate limit is 500 requests/day, 5 req/min per key. Pagination is handled through page and per_page in the query string, with total_pages and total_results returned in response metadata (Franz et al., 5 Sep 2025). These details indicate that OTTER is designed for direct integration into analysis workflows rather than only manual browsing.

The web application provides a complementary interface. Its home page includes an interactive skymap of all OTTER transients using Leaflet and Aladin Lite. The search page offers a GUI form, a table of results, and export as zip(JSONs). Transient detail pages expose a metadata panel with default values and citation counts, interactive photometry plots, and a citations-and-contribution view. The light-curve display uses log–log axes and band-based color coding, while spectral-energy-distribution snapshots allow selection by $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$0, $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$1, $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$2, and either $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$3 or $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$4 (Franz et al., 5 Sep 2025).

The contribution interface supports both single-object and bulk JSON upload and performs live validation with Pydantic models. A private vetting dashboard includes a diff viewer together with approve/reject controls (Franz et al., 5 Sep 2025). A plausible implication is that OTTER attempts to couple public accessibility with controlled curation rather than permitting unreviewed direct insertion into the production archive.

4. Tidal disruption event catalog

The initial OTTER release is centered on TDEs and is described as the largest ever photometric archive of that class, with $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$5 observations of 232 TDEs extending from radio to X-ray wavelengths (Franz et al., 5 Sep 2025). This scale is central to the repository’s scientific role because TDE studies often require synchronized treatment of optical, UV, X-ray, infrared, and radio measurements, yet those measurements are typically distributed across separate publications and archives.

The catalog’s internal quality-control structure is encoded through its confidence and meta-flag system. Because classifications can be filtered by verification level, spectroscopy quality, discovery method, and ambiguity status, the same archive can support both permissive population studies and more conservative subsamples (Franz et al., 5 Sep 2025). This directly addresses a recurrent issue in transient-population work: sample definition can dominate downstream inference.

OTTER’s photometric representation is also tuned for heterogeneous TDE phenomenology. The example JSON structure includes both optical AB magnitudes and radio flux-density measurements, each with explicit units, uncertainties, correction flags, and literature provenance. The repository therefore stores the observational context needed for multi-regime modeling rather than only an abstracted light-curve summary (Franz et al., 5 Sep 2025).

Another potential misconception is that a large TDE archive automatically implies uniform selection or uniform reduction. OTTER is instead described as a catalog of published transient event metadata and photometry, with citation-linked entries and correction flags indicating whether host subtraction, k-correction, or extinction corrections have been applied (Franz et al., 5 Sep 2025). This suggests that the repository standardizes representation and access while preserving provenance and some aspects of upstream heterogeneity.

5. Demonstrated scientific analyses

The repository paper demonstrates four example analyses of the TDE population. The first measures the delay between optical/UV and X-ray peaks in optically discovered TDEs with at least two detections in both bands. The analysis defines $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$6, extracts peak times and uncertainties from adjacent bins, and uses a Kaplan–Meier estimator for the censored $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$7 distribution. The reported finding is a median $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$8 days, implying constraints on circularization time of approximately $$F_\nu = 10^{-0.4\,(m_{\rm AB} + 48.6)}\ \mathrm{erg\,s^{-1}\,cm^{-2}\,Hz^{-1}},$$9–$F_\nu$0 days, with censored-data treatment described as essential. The survival function is written as

$F_\nu$1

where $F_\nu$2 is the number of events at $F_\nu$3 and $F_\nu$4 the number at risk (Franz et al., 5 Sep 2025).

The second analysis compares host infrared colors and radio environments for extreme coronal line emitters (ECLEs) and non-ECLE TDEs. Using WISE queries via the OTTER API and astroquery, it computes peak $F_\nu$5 and peak radio luminosity and applies a two-sample KS test. The reported result is that ECLEs are significantly redder, with $F_\nu$6, interpreted as evidence for a dust-rich circumnuclear medium; the paper further states that this suggests dense gas fosters coronal lines and reprocessing (Franz et al., 5 Sep 2025).

The third analysis uses MOSFiT to infer SMBH mass $F_\nu$7 from TDE light curves. OTTER photometry is retrieved in ABmag units, converted into the form expected by the tde MOSFiT model, and fitted with parallel nested sampling. Two example results are provided: AT 2020zso with $F_\nu$8 and AT 2021sdu with $F_\nu$9 (Franz et al., 5 Sep 2025). These examples show that OTTER is designed not merely for retrieval but for direct ingestion into established transient-modeling pipelines.

The fourth analysis validates the FLEET photometric classifier on 83 spectroscopically confirmed TDEs with at least four ZTF $C \in \{0,\dots,4\}$0 points. OTTER is used to extract ZTF light curves, which are then scored by FLEET for probabilities of SN Ia, SLSN, and TDE classes. The findings reported are that the most probable class is TDE for 69% of the spectroscopic TDEs, 89% have TDE in the top-3 probabilities, and jetted TDEs fail as expected (Franz et al., 5 Sep 2025). This case study illustrates OTTER’s utility as a benchmark corpus for classifier testing under explicitly defined sample cuts.

6. Ingestion, curation, and extension pathway

OTTER includes a structured contribution workflow. In the web interface, single-object upload requires your_name, your_email, transient_name, koordinate (ra,dec), and bibcode. Optional metadata include redshift, distance, classification, and host information. Photometry fields include time, filter, magnitude or flux with uncertainty, wavelength or frequency, units, corrections, and bibcode. Bulk upload uses two files, one for metadata and one for photometry (Franz et al., 5 Sep 2025).

Automated checks include JSON schema validation through Pydantic, coordinate-format and unit-consistency checks, and existing name or coordinate deduplication through cone search. After submission, an upload goes to a Pending table visible to trusted vetters. Vetters compare the upload against existing data and can approve or reject it. On approval, the data are merged into the backend, identical entries are deduplicated with new citations appended, and rejected contributions leave an audit trail (Franz et al., 5 Sep 2025). This workflow indicates that OTTER treats literature aggregation as a curation problem with explicit provenance and change tracking.

Planned extensions include additional transient classes, with the schema described as agnostic to event type; adding supernovae, gamma-ray bursts, or fast radio bursts would involve defining new classification flags and metadata sub-keys such as t90 for GRBs. The API and web forms are said to be easily extended via modular Pydantic models. Further plans include storage of spectroscopic metadata such as resolution, wavelength range, and line measurements, linking to WISeREP, and integration of host photometry through Astro Datalab queries and VizieR cross-matches (Franz et al., 5 Sep 2025).

The future roadmap also includes on-the-fly MOSFiT fitting for new arrivals, web-hooks for real-time community alerts such as via the Transient Name Server, and a scalability model based on Kubernetes and Docker, with the claim that the setup can scale to hundreds of nodes and that ArangoDB auto-sharding will handle millions of transients (Franz et al., 5 Sep 2025). This suggests that OTTER is intended to evolve from a curated archival resource into a broader platform for analysis and community-facing transient-data services.