MooringMind

Ocean mooring insights, instantly.

An interactive Streamlit dashboard for exploring and forecasting conditions in the California Current Ecosystem (CCE) mooring array (Scripps / SIO). It turns multi-sensor mooring telemetry into:

• Mission Control health checks and headline metrics
• Analytics (coverage, climatology, spectral / FFT, anomalies)
• Data Quality views for gaps, sensor issues, and sanity checks
• AI Predictions: simple forecasting and “soft sensor” modelling
• Optional narrative insights (works offline; keys only enable the LLM buttons)

Primary reference: CCE Mooring Array.

Run locally

Requirements: Python 3.10+ (3.11+ recommended).

git clone https://github.com/aanyalari/datahacks.git
cd datahacks

Installation

This project is packaged as cce-hack under src/. The recommended install is editable so changes to the package are picked up instantly while you develop the app.

• Recommended (editable install):

pip install -U pip
pip install -e .

• From requirements (alternative):

pip install -U pip
pip install -r requirements.txt

• Optional extras:
  • Docs: pip install -r requirements-docs.txt
  • NetCDF/OceanSITES: pip install -e ".[netcdf]"

Windows (PowerShell)

python -m venv .venv
.\.venv\Scripts\Activate.ps1
pip install -U pip
pip install -e .
streamlit run Home.py

macOS / Linux

python3 -m venv .venv
source .venv/bin/activate
pip install -U pip
pip install -e .
streamlit run Home.py

---

Data inputs

You can use the repo’s included CSVs, or bring your own:

• Upload a CSV in the app sidebar, or
• Drop a merged *.csv under data/raw/ (the loader picks a default)

If charts look empty, set the sidebar time window to All data (mooring sensors often do not overlap perfectly in time).

Expected CSV shape (quick guide)

The app works best with:

• a timestamp column named time (UTC recommended), and
• sensor columns as numeric fields (temperature, salinity, pH, oxygen, nitrate, chlorophyll, wind, etc.)

If your raw files are separate per sensor, use the fetch/merge script below (or merge them yourself into one “wide” table).

Optional: generate a tiny sample file

python scripts/generate_sample_data.py

Optional: fetch and merge a deployment directly from the mooring site

python scripts/fetch_cce_deployment_csv.py ^
  --csv-base https://mooring.ucsd.edu/cce2/cce2_19/csv/ ^
  --mooring-id CCE2 ^
  --out data/raw/cce2_19_merged.csv

On PowerShell, use a single line or backtick continuation instead of ^ (cmd.exe).

---

Features (what to try in the app)

What you can do with MooringMind

• Mission Control (Home):
  • quick “health” view of the selected time window
  • headline sensor tiles (latest vs a trailing baseline)
  • map view of reference sites
• Analytics:
  • coverage and gaps across variables
  • climatology / seasonality views
  • spectral structure (FFT / periodicity)
  • multi-sensor comparisons in real units
• Data Quality:
  • missingness summaries and “is this usable?” checks
  • quick outlier/sanity scans that help spot sensor problems early
• AI Predictions:
  • forecasting experiments using lagged sensor history + calendar features
  • anomaly detection with plain-English explanations
  • “soft sensor” models to estimate harder-to-measure variables from easier ones
• Ecosystem context (if you use the included pages/data):
  • CalCOFI comparisons (ship surveys vs high-frequency mooring telemetry)
  • species observation tie-ins (optional pages)

Pages overview

• Home: Mission Control entry point (Home.py)
• Analytics: core plots + story-style tabs (pages/1_📊_Analytics.py)
• AI Predictions: forecast/anomaly/soft sensor tabs (pages/2_🤖_AI_Predictions.py)
• Analysis Lab: one-at-a-time advanced modules (pages/3_🧪_Analysis_Lab.py)
• Data Quality: completeness, gaps, and QA checks (pages/4_📋_Data_Quality.py)

---

Project layout (for contributors)

• Home.py: Streamlit entry point
• pages/: multipage dashboard screens
• src/cce_hack/: reusable package (loading, ingest, features, pipeline helpers)
• scripts/: data fetch / processing utilities
• data/raw/, data/processed/: example inputs and derived panels (large local drops stay gitignored)
• docs/: Sphinx docs (optional)
• notebooks/: Marimo EDA (optional)

---

Dev extras

Common scripts

• scripts/fetch_cce_deployment_csv.py: download a deployment’s csv/ directory and merge channels into a single wide table
• scripts/build_processed_panel.py: build processed panels (when raw folders are present)
• scripts/process_mooring_daily.py: daily summary pipeline utilities
• scripts/download_oceansites_by_variable.py: OceanSITES helper (optional)