Mareas - Tide & Weather Dashboard

A Django app that displays live tide height and weather data for Argentine river stations, sourced from two official APIs (INA and SMN), with a BigQuery + dbt + Airflow layer behind it for historical analysis.

No database. No configuration. Clone and run.

---

What it does

• Current tide height, trend (rising / falling / stable), and next high/low tide
• Weather: temperature, wind speed & direction, rainfall
• SVG tide chart, hand-built in vanilla JS
• 3 stations: San Fernando, Rosario, Zárate (up to 4 days of forecast)
• Visual theme shifts with time of day (dawn / day / sunset / night)

---

Data pipeline

INA API + SMN feed → local JSON cache → Django app (live)
                    ↘ BigQuery (raw) → dbt → mart → Airflow (daily)

• The live dashboard reads a local per-station JSON cache: zero added latency, no dependency on the warehouse
• A separate loader does an incremental MERGE into BigQuery, building a real time series instead of a rolling snapshot
• dbt builds two staging models (INA, SMN) and one mart, mareas_por_estacion, joining tide and weather by station/date/hour
• Tests cover freshness, duplicates, and forecast staleness
• Airflow runs the pipeline daily: refresh → load → dbt run → dbt test

See dbt_mareas/README.md for the full dbt setup.

---

Stack

Python 3.11 · Django 4.2 · Vanilla JS · CSS custom properties

Data layer: BigQuery · dbt · Airflow · Docker

---

Architecture

mareas/
├── services/
│   ├── stations.py         # station catalog
│   ├── refresh.py          # fetches INA + SMN
│   ├── transform.py        # interpolation, extremes, merge
│   ├── landing.py          # view context
│   └── bigquery_export.py
└── management/commands/     # mareas_actualizar_datos, mareas_cargar_bigquery

dbt_mareas/         # staging + mart models on mareas_raw
airflow_dags/        # mareas_dag.py

---

Run locally

git clone https://github.com/chipap-dev/mareas.git
cd mareas
python -m venv venv
source venv/bin/activate      # venv\Scripts\activate on Windows
pip install -r requirements_mareas.txt
python manage.py runserver

Open http://localhost:8000: loads instantly from the cached data already in the repo.

python manage.py mareas_actualizar_datos   # refresh live data

To run the BigQuery + dbt layer, see dbt_mareas/README.md.

---

Built by Claudia Cáceres · LinkedIn · Buenos Aires, Argentina