VLKB archive¶

The VLKB tab of the main window — beside Data Hub and SKAVA, and reachable from the Command Palette (⌘K / Ctrl+K) as VLKB Inventory — queries the ViaLactea Knowledge Base: you give it a region of the Galactic plane, it lists every image and spectral cube that covers it, and it cuts out the piece you asked for and opens it.

Three steps, and each one tells you what it is doing.

1 · Choosing the region¶

Fill in the galactic coordinates of the centre, then give the region a size — and the size decides the shape:

Filled in	Region sent
radius	A cone: `CIRCLE l b r`.
dl and db	A box: `RANGE l₁ l₂ b₁ b₂`, with dl and db as the full widths.
neither	Nothing. Query stays disabled and says what is missing.

Under the fields, a line shows the region that will actually be sent, updated as you type:

Will query a cone
CIRCLE 40.000000 0.000000 0.300000

This is the same string the archive receives and the same one the cutout is taken with — not a separate rendering of it.

Note

The Selection radios above (None / Point / Rectangular) are not part of the query. They choose what you are allowed to draw on the all-sky view; the drawing then fills in the coordinate fields. What is queried is always what the fields say.

Longitude near the Galactic centre¶

Galactic longitude wraps at 360°, and a box around l = 0 legitimately runs from, say, 359° to 3°. VisIVO sends that as it stands — RANGE 359 3 … — which is how the archive expresses a wrapped interval, and says so under the fields. A box 360° wide is sent as RANGE 0 360, meaning every longitude.

2 · The inventory¶

The result is a window listing everything that covers the region, grouped by wavelength regime and product type, with a count in the header.

Column	What it is for
Dataset	Survey and band or line — “THOR 1.44 GHz”, “GLIMPSE I 3.6 um”.
Coverage	`full` when the dataset covers the whole region, `partial` when it only overlaps it. Full-coverage rows are also highlighted green.
Identifier	The tail of the archive identifier. An inventory routinely lists six rows that read THOR 1.44 GHz; this is what tells them apart.

Two ways to narrow it down: the filter box matches survey, band and line names, and Only datasets that fully cover the region hides everything that merely clips a corner.

3 · Downloading¶

Double-click a dataset, or select several — ⌘/Ctrl-click, or shift-click for a run — and press Download & Open, which counts what it will fetch. For a long list, consider Request as batch below instead. Each dataset is cut to your region by the archive, downloaded to the backend, and opened:

images are added as layers to the viewer this inventory session already opened, so a set of bands stacks in one window rather than scattering across six;
cubes open in the spectral cube viewer.

The viewer loads one layer at a time, so a batch queues and goes in one after another — the status line says how many are waiting. A download that fails does not cancel the rest.

Many datasets at once¶

Selecting a dataset and pressing Download & Open asks the archive for it directly, one request per dataset. That is the right thing for a handful. For more, Request as batch (N)… — enabled from two datasets up — uses the archive’s own batch API instead: the whole selection goes as one asynchronous job.

The difference is worth knowing:

	Download & Open	Request as batch
Requests	One per dataset	One job for all of them
Starts	Immediately	After the archive queues the job
If you close the window	The remaining downloads stop	The job keeps running on the archive
Results arrive	One cutout at a time	All together, in one archive

The batch window lists what was asked for and follows the job: it shows the phase (queued, executing, completed), and when the job finishes it marks each dataset ready or failed — hover a failed row for the archive’s own reason, which is usually the interesting part when nineteen of twenty worked.

Press Download & Open there to collect the results. They are downloaded to the backend and unpacked for you, then opened like any other cutout. (The legacy tool stopped at handing you a .tar.gz.)

Tip

A job the archive accepted keeps going whether or not you are watching it. If the report cannot be read — a moment of network trouble — the window offers Retry report rather than losing the job: the results are still there.

When the archive says no¶

A cutout request can fail for ordinary scientific reasons — the region falls outside the survey, the identifier needs a different endpoint — and the archive explains itself in the body of its reply, not in the HTTP status.

VisIVO checks that what came back is a FITS file, and when it is not it shows you the archive’s own words:

The service returned application/fits instead of FITS: UsageError : overlap computation could not be completed with the given arguments.

Tip

A message of that shape is about the server, not your query. It is worth sending to whoever runs the archive: it names the failure exactly.

The check is on the file’s contents, not on the type the service declares — services mislabel FITS as application/octet-stream routinely, and mislabel their error pages as FITS just as readily. A gzip or a tar archive is recognised and named as such rather than handed to the FITS reader, which could only ever report the same “No SIMPLE card found” whatever had actually gone wrong.