Project Overview
QuARGO is a conversational interface for exploring ARGO float data. It lets users pose questions in natural language and instantly view results as maps, profile plots, and tables. Our motivation is to make high–value ocean observations more accessible for scientists, educators, and decision makers without requiring specialized coding skills.
The platform translates user intent into structured queries, retrieves the appropriate observations, and presents them in clear, interactive visualizations. Beyond discovery, users can export curated subsets of data for downstream analysis.
Core goals
- Lower the barrier to work with ARGO observations
- Provide trustworthy responses with clear provenance
- Offer fast visual analysis and export options
Tech stack
- LLM + Retrieval Augmented Generation (RAG) for query translation
- Embeddings over schema/docs for semantic grounding
- Datastores: PostgreSQL/Parquet-backed catalogs
- API & Orchestration: Next.js server routes
- Frontend: Next.js App Router + Tailwind CSS
Data Sources
QuARGO utilizes observational datasets produced by the international ARGO program and partner archives. Depending on the query, we may fetch core T/S profiles or biogeochemical extensions.
Primary portals
- Official ARGO Program — argo.ucsd.edu
- Global Data Assembly Centers (GDAC): IFREMER & FNMOC — USGODAE
- Argo Data Selection Tools — National Argo (examples)
Coverage & filters
- Date range: ~2000–present (updated daily)
- Regions: global oceans; filter by basin/lat-lon/polygon
- Parameters: temperature, salinity, pressure; BGC where available
Notes: Availability and cadence vary by float and deployment. Always consult official documentation for the most current guidance.
User Guide
Formulating queries
- 1Start with a goal: e.g., “Show temperature profiles near 30°N, 140°W in 2022.”
- 2Specify region (lat/lon or area), time period, and variables.
- 3Use comparison language for multi-set analysis: “Compare 2010 vs 2020.”
- 4Ask follow-ups to refine: “Filter for depths above 1000 m.”
Using the dashboard
- 1Map: pan/zoom, draw selection boxes, and inspect individual floats.
- 2Profiles: view temperature/salinity/BGC profiles; toggle overlays and units.
- 3Compare: select multiple results to generate side-by-side visualizations.
- 4Export: download subsets as CSV/ASCII/NetCDF for offline analysis.
Interpreting results
- 1Profiles are vertically referenced by pressure; check calibration flags.
- 2Geospatial visualizations display trajectories and profile locations.
- 3Quality flags indicate data status; apply filters for research-grade analyses.
Acronyms & Glossary
- ARGO
- A global array of autonomous floats measuring the upper ocean.
- BGC
- Biogeochemical parameters measured by some floats (e.g., oxygen, nitrate, pH).
- CTD
- Conductivity–Temperature–Depth instrumentation for ocean profiling.
- NetCDF
- Network Common Data Form; a self-describing, array-oriented data format.
- Salinity
- A measure of dissolved salts in seawater, commonly PSU or PSS-78.
- RAG
- Retrieval Augmented Generation; combines LLMs with relevant external data.
- GDAC
- Global Data Assembly Center; official ARGO data distribution hubs.
- Isopycnal
- A surface of constant density in the ocean; used to compare water masses.
For official definitions and standards, refer to the ARGO user manuals and the NetCDF conventions.