Speaker
Description
Ground-based networks and space-based archives are served by largely disjoint client stacks: sunpy, radiospectra and Fido with per-station file conventions on one side; CDAWeb, AMDA and HAPI clients with ISTP metadata on the other. Comparing a type-III burst observed simultaneously by a ground spectrograph and a spacecraft receiver therefore requires manual work for each event. We report on serving both from a single interactive analysis environment, and on the interoperability obstacles this exposed. The work follows from the Lorentz Center workshop Bridging Gaps in Heliospheric Radio Data Analyses (Leiden, May 2026).
SciQLop is an open-source, cross-platform environment for multi-mission in situ data analysis, with archive access through the Speasy library and an extensible plugin system. Its radio plugin serves 38 curated dynamic-spectrum products from space-based receivers spanning five decades: Wind/WAVES, STEREO-A and -B SWAVES, Solar Orbiter/RPW, PSP/FIELDS RFS, and the planetary receivers of Juno, Cassini, Galileo and Voyager 1 and 2. It serves ground-based e-CALLISTO, I-LOFAR (mode-357 beam-formed) and RSTN through radiospectra's Fido clients. All are exposed as time-windowed products that fetch and cache on demand as the user pans and zooms across a shared time axis. The approach is not specific to radio: an equivalent plugin serves FDSN seismic networks through ObsPy, and a SuperMAG provider being added to Speasy itself exposes that network's roughly 600 ground magnetometer stations as individual products, discoverable by IAGA code and returned in the local geomagnetic or the geographic frame.
The integration exposed several obstacles of general interest. Ground-based products carry no ISTP-equivalent presentation metadata, requiring a separate plot-hints layer for them to render consistently with archive-served data. EOVSA spectrograms moved behind a registration wall during development, motivating an explicit "listed but not retrievable" product state rather than a silent absence. SuperMAG publishes no per-station temporal coverage, reporting instead which stations were operating over a requested interval, so a client cannot advertise a station's real extent without an extra query. Search attributes never fully identify a channel. A station attribute exists only in the affiliated package, sunpy's generic one having been removed, and the token separating concurrent signal chains at one site, such as an e-CALLISTO focus code, has no attribute at all, so channel identity is reconstructed for every instrument by re-filtering the returned rows on their columns. Getting this wrong is silent: at I-LOFAR, where every timestamp ships one file per linear polarisation, the two channels were merged into a single spectrogram, and no structural check could object because they share an identical frequency grid. I-LOFAR mode-357 files decode into three co-temporal frequency bands requiring reassembly, and carry their start time only in the filename, so a cache that renames files loses it.
We close by asking what minimal agreed metadata ground-based networks would need to publish for a generic client to consume them as it already consumes CDAWeb and AMDA.