Note: Jobs are managed by the Watchdog service.
Enable/disable jobs here; the Watchdog executes them.
Container Status
Loading containers...
Stats refresh automatically every 10 seconds.
Restart buttons are disabled for the API container.
User Management
Loading users...
AI Data Scopes
Each scope is a named set of data tables that
non-admin AI queries may read — enforced by
Postgres itself. Assign a scope per user
above. Admins are always unrestricted, and
system tables (settings, users, logs) are
never grantable.
Loading scopes...
New Scope
New User
Reset Password
Confirm
Global Settings
Access
When on, anyone can view the read-only map
(tracks, layers, modals) without an
account. AI chat, admin, and all editing
stay locked to logged-in users. The Mapbox
token becomes publicly readable — restrict
it by URL in your Mapbox account.
Default Map View
Touch devices only. Leave blank to use the
default zoom. Lower = more area visible on phones.
Zoom level when a search result is selected.
Lower = wider view. 6 ≈ 100 km scale.
Save then hard-refresh the map to preview.
Satellite styles show bathymetry; others may
not.
Tip: Use the main map to find coordinates.
Contact Panel
Welcome Panel
Learn More Panel
Built-in Layers
System data sources. Toggle on/off and set
coverage here — no code change needed.
Detailed harvest tuning (refresh, markers,
workers) still lives in each source's own tab.
User-Added Layers
Pick a Mode, tune that mode's box; the Shared
Renderer box styles whichever layer runs.
After saving, reload the map to apply renderer
changes. Re-run the Streamlines harvest job to
apply fetch (stride/bbox/mode) changes.
ARGO float harvest settings and map marker styling.
Re-run the ARGO harvest job to apply bbox changes.
Marker changes apply on next map reload.
Bounding Box
Days of profile history to store
Markers
Layer Control
NDBC buoy harvest settings and map marker styling.
Re-run the NDBC harvest job to apply bbox changes.
Marker changes apply on next map reload.
Bounding Box
Days of obs history to store
Points per station after thinning
Markers
Layer Control
Copernicus Marine in-situ moored buoys and drifting
buoys (federated) plus Taiwan CWA data buoys
(direct): one harvest, one layer, one map toggle
per sub-layer. Region, source and harvest changes
apply on the next Copernicus Buoys / Taiwan CWA job
run; marker changes on the next map reload.
Sources
Copernicus coverage regions
A platform lands in the first region, top to
bottom, whose box holds it, so keep a global
catch-all last. Each region is harvested and
swapped on its own and its rows carry the region
name. Every region on = global coverage: about
2,500 platforms and a ~5 h run, so schedule the
Copernicus Buoys job daily in background mode.
Longitudes are signed degrees (west negative).
On
Name
Label
Lat min
Lat max
Lon min
Lon max
Held
Harvest
Days of history kept per platform
(one file per platform per day)
Points per platform after thinning
Markers
Layer Control
Copernicus Marine (CMEMS) tile harvest settings.
Re-run the Copernicus harvest job to apply changes.
Fleet positions, tracks and port calls come
from Global Fishing Watch (free,
non-commercial): hourly positions about three
days behind real time, history back up to
three years. The token is server-side
(GFW_API_TOKEN) and never entered here.
Fleet roster
Named fleets, one MMSI per line. Each fleet
is its own map toggle; every source reads
this list.
Ship markers
Defaults for any fleet that sets no
colours of its own above.
Port calls & events layer
Port visits, loitering, encounters and
AIS gaps, harvested nightly; shown per
vessel from its popup ("Show port
calls"). CC BY-NC; the map credits GFW.
A live commercial AIS source can be added
later; while it runs its fixes are newer
and own the markers.
GOBLIN glider deployments: the glider's GPS
track and its Rice's whale detections, one
bundle per deployment from
goblin-map
(goblin_olorin_<tag>.json).
Uploading a deployment again replaces it.
Detection positions are interpolated between
surfacings; the map says so.
Upload a deployment bundle
Stored deployments
Deployment
Glider
Species
First fix
Last fix
Fixes
Detections
AIS traffic
Map layer
One toggle per deployment in the layer
menu; the track in one colour, detections
coloured from the low to the high score
colour.
Live gliders (and later Saildrones) from any
ERDDAP server: every active dataset, every
sensor except the excluded prefixes and names,
harvested by the "ERDDAP vehicles" job (its
interval on the Jobs page is the cadence).
Add a server as a row; nothing else to
configure.
ERDDAP sources
One layer toggle per source. The dataset
filter is a regex on dataset IDs (needed
for servers whose catalogue is mostly not
vehicles, e.g. PMEL); leave blank for the
Glider DAC. The look here (track colour,
width, opacity, icon size) is the base for
every vehicle of the source; a type panel or
a vehicle's own Edit overrides it.
Name
ERDDAP URL
Dataset filter
On
Track color
Width
Opacity
Icon px
Map layer
The look (track colour, width, opacity, icon
size) is set per source above; a vehicle
type panel below overrides its source, a
vehicle's own Edit overrides its type.
Vehicle types
One panel per platform type, named exactly
as the ERDDAP server names it (the harvested
table shows each vehicle's type): its icon,
icon size, track colour, width and opacity,
and the default gauge / plot sensors for a
vehicle of that type that has no Sensor
config of its own (comma-separated; blank =
the glider defaults). A type without a
panel uses the defaults above.
Type
Icon
Icon px
Track color
Width
Opacity
Flight sensors
Science sensors
Vehicle window
Click a vehicle on the map for its window:
flight gauges and science plots. Which of
its sensors play which role is set per
vehicle: Edit a row below, "Sensor config".
A vehicle without its own config uses its
type's default lists (Vehicle types above),
else the seed glider set.
Harvested deployments
Edit a row to set its look or hide it
from the map; the data always stays.
Dataset
Type
Institution
First
Last
Fixes
Obs
Sensors
Active
Shown
External data-source connectors run by
the watchdog. Credentials are configured
server-side via environment variables
and never entered here.
MAN CEON Engine Telemetry
REST / NDJSON
Checking credential…
Vessel Registry
A new hull starts harvesting from its
Start date. Hulls with harvested data
can be disabled but not removed.
Configure the AI assistant for natural
language queries about vehicles, data,
and operations.
Managed via ANTHROPIC_API_KEY
in the server environment
(never stored in the database).
Click Refresh Models after
entering a base URL.
Tool calling (run_sql,
get_logs) needs a
tool-capable model
(e.g. qwen2.5, llama3.2,
mistral). Web search is
disabled for local
models.
Hard ceiling on total tokens
consumed. AI is disabled when
the limit is reached.
Unsaved changes will be lost!
Vehicle
Sensor config
Every sensor this vehicle carries. Flight = a gauge
in its window, Science = a plot tab (tabs follow
this order); neither = not shown. One role each.