strait — satellite vessel detection & port activity monitoring

A Python package for detecting vessels from Sentinel-1 SAR imagery and measuring port activity from satellite data — any port, any time.

The one thing

import strait

cutout = strait.Cutout(
    module="sentinel1",
    x=slice(103.4, 104.6),
    y=slice(1.0, 1.6),
    time=slice("2021-01", "2026-09"),
)
cutout.prepare()  # download + process scenes
detections = cutout.detect()  # CFAR vessel detection
monthly = cutout.aggregate(detections, zones={"eastern": (104.0, 1.24, 104.35, 1.40)})

That gives you monthly vessel counts per zone from free satellite radar.

Why this exists

Ports publish trade statistics with a 2-4 week lag. Satellite radar sees ships at anchor immediately, day or night, cloud or clear. This package turns that satellite data into economic indicators.

It was built for the Singapore Strait Observatory project, where radar-derived anchorage presence explains 48% of bunker sales variance (R²=0.478, detrended, weather-robust, validated against AIS).

Install

pip install strait-observatory

What it does

Data sources

Quick start

pip install strait-observatory
export CDSE_USER=your@email
export CDSE_PASSWORD=your_password

import strait

# 1. Define your area and time
cutout = strait.Cutout(
    module="sentinel1",
    x=slice(103.4, 104.6),  # longitude
    y=slice(1.0, 1.6),      # latitude
    time=slice("2021-01", "2026-09"),
)

# 2. Download and process (first time takes ~1h for 5 years)
cutout.prepare()

# 3. Detect vessels
detections = cutout.detect(method="trimmed_cfar")

# 4. Define anchorage zones (or use built-in Singapore zones)
zones = strait.Zones.singapore_strait()
monthly = cutout.aggregate(detections, zones, freq="MS")

# 5. Validate against AIS (optional)
ais = strait.AIS(source="aisstream", api_key="...")
match = ais.match(detections, threshold_m=500)

# 6. Correlate with official statistics (optional)
stats = strait.Stats.from_datagov_sg()
results = stats.correlate(monthly, target="bunker_sales")

Architecture (inspired by atlite)

strait/
├── __init__.py          # exports Cutout, detect, aggregate, AIS, Stats
├── cutout.py            # Cutout class (spatial/temporal abstraction)
├── detect/
│   ├── __init__.py      # detect() dispatcher
│   ├── cfar.py          # classic CFAR (v3.1)
│   ├── trimmed_cfar.py  # trimmed CFAR (v4, from SAR literature)
│   └── land_mask.py     # coastline-based land mask
├── data/
│   ├── __init__.py      # data source registry
│   ├── sentinel1.py     # Sentinel-1 via CDSE (Sentinel Hub + OData)
│   ├── ais.py           # AIS from multiple sources
│   └── official.py      # Official statistics (data.gov.sg, etc.)
├── aggregate.py         # zone × time aggregation
├── validate.py          # SAR-AIS matching, precision/recall
├── stats.py             # econometric correlation
└── zones.py             # built-in zone definitions

Built-in zones

# Singapore Strait (from the observatory project)
zones = strait.Zones.singapore_strait()

# Define your own
zones = strait.Zones({
    "my_anchorage": (104.0, 1.24, 104.35, 1.40),  # lon_min, lat_min, lon_max, lat_max
    "port_area": (103.68, 1.20, 104.02, 1.34),
})

License

MIT

Citation

If you use this in research, cite the Singapore Strait Observatory:

@software{strait_observatory_2026,
  title = {strait: satellite vessel detection and port activity monitoring},
  author = {Sivasubramanian, S.},
  year = {2026},
  url = {https://github.com/siva-sub/strait}
}