Satellite & Geospatial

From AltData.wiki, The Alternative Data Encyclopedia

Satellite & Geospatial is one of the categories of alternative data covered by The Alternative Data Encyclopedia: Processed satellite and aerial imagery converted into counts and measurements of physical economic objects: vehicles in parking lots, ships at ports, volumes in storage tanks, crop condition and construction activity. The signal turns remote sensing into site-level time series that estimate stocks and activity before official statistics appear.

The Walmart parking-lot analyses circulating around 2013-15 made vehicle-counting famous on trading desks, building on earlier hedge-fund work with aerial photography. The free Sentinel-2 optical stream from 2015 and cheap cloud compute turned a bespoke trick into a product category, while Planet Labs pioneered commercially available near-daily global imaging. ICEYE demonstrated that smallsat SAR could be commercial at scale, and today constellations serve finance, insurance, government and agriculture simultaneously.

The signal

Processed satellite and aerial imagery converted into counts and measurements of physical economic objects: vehicles in parking lots, ships at ports, volumes in storage tanks, crop condition and construction activity. The signal turns remote sensing into site-level time series that estimate stocks and activity before official statistics appear.

Output is georeferenced time series of counted or measured quantities per site, with confidence scores, imagery provenance and observation dates attached to each observation. Product families include analytic feeds that detect objects automatically and derived variables such as soil moisture or vegetation indices. Radar (SAR) products add all-weather, day-night measurement of flooding, ground deformation and vessel presence.

Why investors pay for it

Physical observation leads official data: counting cars at retailers approximates same-store sales weeks before earnings, tank measurements inform crude inventory trades ahead of government releases, and crop indices drive yield estimates during growing seasons. SAR extends the edge to night, cloud and denied areas, which is why insurers buy observed flood extents rather than modeled probabilities and commodity desks track dark-fleet activity. Funds pay because the same imagery revisits any asset class location on Earth on a known schedule.

Vendors either task constellations directly or license capacity from operators, then run computer-vision pipelines that segment scenes and count or measure target objects per site. Optical providers operate near-daily global scanning constellations plus sub-daily, high-resolution tasking satellites; SAR operators advertise the world's largest commercial radar fleets with resolutions down to roughly twenty-five centimeters. Counts are calibrated against ground truth where accessible, and delivery layers ship both imagery archives and analysis-ready derived series through APIs.

Who uses it

Commodities, macro and consumer funds convert imagery into inventory, activity and yield estimates ahead of official numbers. Insurers and governments buy the same observations for catastrophe response, and industrials monitor assets and supply chains.

Questions to ask vendors in this category

What revisit cadence and resolution do you actually deliver per site of interest? How do you validate counts against ground truth, and what error rates do you publish? How do you handle cloud-cover and scheduling gaps in the series? Are derived counts licensed for redistribution, and does history ship point-in-time?

Complementary signals

This signal pairs naturally with adjacent categories of the encyclopedia:

Weather & Agriculture Supply Chain & Shipping ESG & Climate

Caveats and limitations

Cloud cover and revisit cadence create irregular sampling that complicates time-series comparison. Small samples of monitored sites rarely generalize to corporate revenue without weighting assumptions. Algorithm upgrades change object definitions over time, so backtests need frozen model versions, and calibration drift shows up as level shifts rather than noise.

Compliance and legal considerations

Imagery licenses typically restrict redistribution and resale of native-resolution pixels, while derived counts inherit lighter but real restrictions. Some jurisdictions regulate collection near critical infrastructure, and export-control rules apply to high-capability sensing systems.