NSF Ocean Observatories Initiative · Regional Cabled Array

aRCADA A Regional Cabled Array Data Assistant

Answer questions, surface published research, and generate ready-to-run Python data pull scripts — all from plain-language requests against the OOI Regional Cabled Array's decade-spanning seafloor archive.

Request Data How It Works
10+ Years of Data
900m Avg. Depth
98+ Instruments
∞ Open Access
Scroll
The Challenge

Valuable data,
interdisciplinary impact

The RCA hosts continuous observational data spanning geophysics, geochemistry, biology, and oceanography — yet reliably accessing it remains a significant barrier to interdisciplinary research.

ooi-api — researcher-session
$ query_instrument --type seismometer --site axv01
→ Fetching data stream... OK
→ 3 streams found, metadata complete
 
$ query_instrument --type mass-spectrometer --site ce04
→ ERROR: No programmatic interface found
→ Data stored on PI HTML page (no metadata)
→ Manual extraction required
 
$ query_instrument --type scanning-sonar --site hyv01
→ ERROR: No API endpoint registered
→ 847 raw sensor files, unlabeled
 
$
01
Inconsistent API Coverage

Seismometers and pressure gauges offer seamless access; PI-operated instruments like scanning sonar and mass spectrometers do not.

02
Fragmented HTML Storage

Many instruments store unprocessed sensor readings on standalone HTML pages with no metadata or programmatic interface.

03
Siloed Disciplines

Few researchers integrate data across domains — methane seeps, hydrothermal vents, and volcanic activity remain separately studied despite deep connections.

04
Wasted NSF Investment

Scientists invest substantial time locating and contextualizing data before any analysis, reducing the scientific return on a decade of infrastructure investment.

Architecture

RAG-based agentic
query resolution

Three modes, one interface — ask questions, review literature, or generate data pull scripts from plain-language requests.

1
Query + Mode
Researcher types a plain-language request and selects a mode: Ask a Question, Literature Review, or Access Data.
User Interface
→
2
Hybrid RAG Retrieval
BM25 + semantic search over 98+ instruments, published papers, site context, and data-access patterns.
Gemini Embeddings
→
3
Intent Routing
Routes to the appropriate path: conversational Q&A, citation-formatted literature summary, or structured data plan.
Cloudflare Workers
→
4
Script Assembly
For data requests: instrument IDs, stream names, and time range are filled into a standalone Python script — no server-side pull needed.
Gemini 2.0 Flash
→
5
Local Download
Researcher downloads the .py script, runs it locally, and data is saved directly to disk — OOI M2M, EarthScope FDSN, or PI portal.
Researcher
Capabilities

What aRCADA does

// ask
Ask a Question
Get plain-language answers about instruments, sites, and science — what a BOTPT measures, what happened at Axial in 2015, how hydrothermal venting and seismicity relate.
// literature
Literature Review
Search a curated library of RCA-related publications. Ask "what research exists on methane seeps?" and get cited summaries drawn from indexed papers.
// script
Python Script Generation
For data requests, aRCADA builds a standalone .py script pre-filled with instrument IDs, stream names, and your time range — download and run locally.
// access
Unified Instrument Coverage
Bridges API-accessible instruments (seismometers, pressure gauges, CTDs) with PI-operated portal instruments (sonar, mass spectrometers, DAS) — all in one query.
// scale
Decade-Scale Archive
Access continuous observational records spanning over ten years — methane seep activity, hydrothermal vent chemistry, and Axial Seamount eruption cycles.
// open
Open & Extensible
Built on Gemini API, Cloudflare Workers, and GitHub Pages. The RAG index, instrument catalog, and data-pull scripts are all open-source and community-extensible.
Interface Preview

Three modes,
one interface

Switch between Q&A, literature review, and data access without changing tools.

aRCADA · OOI Regional Cabled Array Preview
Ask a Question Literature Review Access Data
AI
Welcome to aRCADA. I can answer questions about RCA instruments and sites, surface relevant published research, or generate a ready-to-run Python script for any data request. Select a mode above to get started.
You
Pull seismic and bottom pressure data from Axial Seamount for two weeks after the April 2015 eruption
AI
Got it — you're looking for seismic waveform and seafloor pressure data at Axial Seamount in the two weeks following the April 24, 2015 eruption onset.
AI
Data Pull Script
Seismic and bottom pressure records, Axial Seamount — Apr 24 → May 8 2015
Axial Seamount Seismometer (AXAS1)EarthScope FDSN
Axial Base Bottom Pressure (BOTPT)OOI M2M API
Axial Base Seismometer (AXAS2)EarthScope FDSN
Period 2015-04-24 → 2015-05-08  ·  Est. size ~840 MB
Download Script (.py)
Coverage

Cascadia Margin
instrument network

The RCA spans the Juan de Fuca tectonic plate, hosting instruments across methane seeps, hydrothermal vents, and an active submarine volcano — a complete biogeochemical system.

API-accessible
HTML-stored (PI)
Ocean-Bottom Seismometers (OBS-BB) EarthScope
Seafloor Pressure Gauges & Tiltmeters (BOTPT) API
CTD Profilers & Fluorometers API
Low-Frequency Hydrophones (HYDLF) API
pCO₂ Sensors (PCO2W) API
Vent Fluid Thermometers (THSPH) API
Acoustic Doppler Current Profilers (ADCP) API
Dissolved Oxygen Sensors (DOSTA) API
HD Video Cameras (CAMHD) HTML
Scanning Sonar (DIDSON / ZPLSC) PI Portal
Roadmap

Future directions

Scoped initially to the OOI platform, designed from the start to expand.

01
Expanded Literature Corpus
Index full-text PDFs from the RCA research literature, moving beyond abstracts to enable paragraph-level retrieval and broader interdisciplinary synthesis.
Near-term
02
User Persona Modes
Tailor responses for expert scientists, interdisciplinary researchers, and the general public — same data, different depth and framing.
Near-term
03
Notebook Integration
Generate Jupyter notebooks rather than bare scripts — with inline data loading, exploratory plots, and suggested analyses appropriate to the instrument type.
Phase 2
04
COSZO Integration
Adapt aRCADA as the data assistant for the forthcoming Cascadia Offshore Subduction Zone Observatory to support earthquake and tsunami early warning research.
Long-term
Researchers

The team

As users of seafloor geophysical instrument data, we know studying ocean systems is deeply interdisciplinary. We want to decrease the barrier to entry and advance researchers' ability to work across datastreams.

Michael Hemmett
UW ESS Graduate Student
Uses ocean bottom seismometer data from the Regional Cabled Array to study Axial Seamount volcano. Works with Marine Denolle in UW Earth and Space Sciences and William Wilcock in the UW School of Oceanography.
Anjani Mirchandani
UW AMATH Undergraduate Student
Uses seafloor distributed acoustic sensing (DAS) in Alaska to image the subsurface. Works with Marine Denolle in UW Earth and Space Sciences. Intends to pursue graduate work in Applied Mathematics or a related field.
References

[1] NSF Ocean Observatories Initiative · Regional Cabled Array · ooinet.oceanobservatories.org
[2] Kelley, D.S. et al. (2014). The ALOHA and PAPA time-series stations and the Ocean Observatories Initiative. Oceanography, 27(1).
[3] Boore, D.M. et al. Cascadia Offshore Subduction Zone Observatory — proposed instrumentation framework.