Geospatial Intelligence (GEOINT)
BLUF
Geospatial Intelligence (GEOINT) is the intelligence discipline comprising the exploitation and analysis of imagery and geospatial data to describe, assess, and visually depict physical features and geographically referenced activities on Earth. It is formally defined in 10 U.S.C. § 467 as “imagery, imagery intelligence, and geospatial information.” GEOINT’s strategic value is its empirical ground truth: unlike assessments derived from intercepts or human sources, imagery-derived intelligence provides visually verifiable, location-anchored evidence that is difficult to dispute and legally defensible. Since 2015, the commercial satellite revolution — Planet Labs, Maxar, Capella Space, and open Sentinel imagery from ESA — has democratized GEOINT from a classified national-technical-means monopoly to a discipline accessible to independent analysts, civil-society organizations, and investigative journalists. This democratization is the most consequential structural shift in the intelligence landscape of the 2010s–2020s.
Historical Origins and Institutional Development
From Observation Balloons to Space Reconnaissance
Systematic aerial observation emerged during the American Civil War (Union Army balloon corps, 1861–1863) and was operationalized at scale during World War I, when the RFC and French Aéronautique Militaire flew dedicated reconnaissance sorties over the Western Front, producing imagery that underpinned artillery targeting and trench mapping.
World War II transformed aerial reconnaissance into a strategic intelligence function. The RAF’s Photographic Reconnaissance Units and the USAAF’s reconnaissance squadrons flew systematic coverage of Axis industrial targets, military installations, and port facilities. The Medmenham Air Intelligence Unit developed the systematic photo interpretation tradecraft — covering, cross-section analysis, shadow mensuration — that remains foundational to modern imagery exploitation.
Cold War: The Space Reconnaissance Revolution
The Cold War elevated GEOINT from a tactical enabler to a strategic intelligence instrument:
- CORONA Program (1960–1972): The first US reconnaissance satellite program, operated by the CIA and later DIA under the NRO. CORONA’s final missions achieved ground resolution of 1.8 meters. The program produced more imagery of the Soviet Union in its first year than all U-2 overflights combined. Declassified CORONA imagery has since become a primary source for archaeological and environmental research.
- U-2 Program (1955–present): The Lockheed U-2 high-altitude reconnaissance aircraft operated by the CIA and later USAF at altitudes above Soviet air defenses. The loss of Gary Powers’ U-2 over the Soviet Union in May 1960 triggered a major diplomatic crisis, demonstrating both the strategic value and political risk of overhead reconnaissance.
- Soviet Zenit satellites (1961–1994): The Soviet equivalent to CORONA, providing strategic imagery of NATO dispositions. The Soviet GEOINT infrastructure was institutionalized in GRU’s Military Topographic Directorate.
The NGA and Modern Doctrine
The National Geospatial-Intelligence Agency (NGA) was established in 2003, consolidating the Defense Mapping Agency, the National Imagery and Mapping Agency, and related functions. NGA’s mission: “to provide timely, relevant, and accurate geospatial intelligence in support of national security.” NGA is a combat support agency of the Department of Defense and a member of the Intelligence Community, reporting to both the SecDef and the DNI.
NGA’s GEOINT Basic Doctrine (2006, updated 2016) formally defined GEOINT as encompassing three elements: imagery, imagery intelligence, and geospatial information — establishing GEOINT as broader than pure satellite imagery and encompassing all forms of geospatially referenced data.
Collection Platform Taxonomy
Classified National Technical Means (NTM)
NTM assets operate at resolutions well below 0.3 meters and on classified collection schedules. Their existence is acknowledged; their capabilities and operational parameters are classified. For OSINT practitioners, NTM is not an accessible source — it is the context that explains why declassified GEOINT products occasionally appear with implausible specificity, and why commercial imagery of certain sensitive sites may be periodically suppressed (via Kyl-Bingaman Amendment, see below).
Commercial Satellite Platforms
| Provider | System | Resolution | Revisit | Spectral | Access |
|---|---|---|---|---|---|
| Planet Labs | Dove (optical) | 3–5m | Daily global | RGB + NIR | Education free; commercial subscription |
| Maxar Technologies | WorldView-2/3/4 | 0.3–0.5m | 1–4.5 days (variable) | Multispectral + SWIR | Commercial; government prime |
| BlackSky | Optical constellation | 1m | Multiple per day (key sites) | RGB | Commercial |
| Satellogic | Optical + hyperspectral | 0.7–1m | Configurable | Multispectral + hyperspectral | Commercial |
| Capella Space | SAR (X-band) | 0.5m | On-demand | SAR | Commercial API |
| Umbra | SAR (X-band) | 0.25m | On-demand | SAR | Commercial API |
| ICEYE | SAR (X-band) | 1m | Persistent dwell capability | SAR | Commercial |
| Synspective | SAR (X-band) | 1–3m | Regional | SAR | Commercial |
Assessment: The combination of Planet Labs (daily global optical, free education tier) + Sentinel-1 SAR (free, cloud/night-independent) + Sentinel-2 optical (free, 10m, 5-day revisit) provides approximately 85–90% of OSINT-grade GEOINT capability for conflict monitoring at zero cost. Commercial platforms add value for sub-5m resolution requirements and on-demand tasking of high-priority AOIs.
Open-Access Satellite Sources
| Source | Resolution | Revisit | Primary use | Access |
|---|---|---|---|---|
| Sentinel-1 (SAR, ESA) | 10–20m | 6–12 days | Cloud/night-independent monitoring; ship detection; construction; ground change | Copernicus Dataspace (free) |
| Sentinel-2 (optical, ESA) | 10m (visible/NIR), 20m (SWIR) | 5 days | Visual confirmation; vegetation; urban damage; burn scar | Copernicus Dataspace (free) |
| Landsat 8/9 (USGS/NASA) | 15–30m | 16 days | Historical archive from 1972; long-baseline change detection | USGS EarthExplorer (free) |
| MODIS (Terra/Aqua) | 250m–1km | Daily | Fire/smoke monitoring; large-scale atmospheric events; rapid SA | NASA Worldview (free) |
| GOES (NOAA) | 0.5–2km | 5–15 min | Near-real-time weather; fire; maritime weather context | NOAA (free) |
Key access portals:
- Sentinel Hub EO Browser (
apps.sentinel-hub.com/eo-browser): Sentinel-1, Sentinel-2, Landsat, MODIS in one interface; free 30,000 processing units/month; registration via Copernicus Dataspace - NASA Worldview (
worldview.earthdata.nasa.gov): MODIS/VIIRS near-real-time; no registration; fires, aerosols, storms - USGS Earth Explorer (
earthexplorer.usgs.gov): Full Landsat archive from 1972; free registration - Copernicus Dataspace (
dataspace.copernicus.eu): Full Sentinel catalog with STAC API for programmatic bulk download
Airborne and UAV Platforms
- U-2S (USAF): Current-generation high-altitude reconnaissance; operational at >70,000 ft; cameras, SIGINT, and SAR payloads; not accessible to OSINT
- RQ-4 Global Hawk: Long-endurance UAV; imagery and SIGINT; NATO Alliance Ground Surveillance uses RQ-4B; not OSINT-accessible
- Bayraktar TB2 (commercial export): Armed UAV with Wescam MX-15D EO/IR camera widely deployed in Ukraine, Libya, Azerbaijan, and Sahel conflicts; footage disseminated via open Telegram channels and social media has become a significant OSINT-accessible GEOINT source in conflict documentation
- Commercial drones: DJI and equivalent platforms under 400m AGL produce high-resolution ground-truth imagery widely shared by conflict participants — a major source of real-time, geolocatable conflict imagery accessible via SOCMINT collection
Legal Constraint — Kyl-Bingaman Amendment
The Kyl-Bingaman Amendment (Section 1064 of the FY2009 NDAA, as amended) restricts the commercial sale of satellite imagery of Israel and the Occupied Palestinian Territories at resolutions more precise than those available from non-US commercial sources. Historically interpreted as restricting US commercial satellite imagery of Israel/Palestine to 2m resolution. Since 2020, the restriction has been eased (no better than “best available” from foreign commercial sources) but continues to affect what is commercially available for OSINT work on this region. Non-US commercial providers (Airbus Defence, Kompsat) and ESA Sentinel imagery are not subject to the amendment.
GEOINT Tradecraft — Open-Source Workflow
Area of Interest (AOI) Definition
GEOINT collection begins with a precisely defined AOI: a geographic bounding box or polygon defining the area to be monitored. AOI definition requires:
- Threat/analytical focus: what military activity, infrastructure development, or event is being monitored?
- Spatial extent: large enough to capture relevant activity, small enough to be manageable with free-tier processing quotas
- Coordinate format: WGS84 decimal degrees for interoperability with GIS tools and satellite access APIs
Source Selection Logic
| Condition | Primary source | Rationale |
|---|---|---|
| Cloud cover > 50% | Sentinel-1 SAR | Cloud-penetrating; night-capable |
| Ground resolution requirement > 10m | Sentinel-2 | Free, 5-day revisit |
| Ground resolution requirement < 5m | Planet Labs (education/commercial) | Daily coverage |
| SAR ship/vehicle detection | Sentinel-1 GRD | Bright radar returns from metallic objects |
| Historical baseline (>5 years) | Landsat 8/9 or CORONA (declassified) | Long archive |
| Near-real-time fire/weather | MODIS/GOES via NASA Worldview | Sub-daily refresh |
| Very high resolution (< 1m) | Maxar / Capella / Umbra | Commercial; AOI-specific tasking |
Change Detection Methodology
The primary analytical technique for GEOINT-based conflict monitoring:
- Acquire pre-event baseline: Pull the most recent cloud-free imagery of the AOI before the event of interest.
- Acquire post-event imagery: Pull imagery from after the event or as close to as achievable.
- Pixel-level comparison: In Sentinel Hub EO Browser, use the Change Detection layer or export scenes for comparison in QGIS (free, open-source GIS).
- Feature identification: Changed pixels likely represent: new construction (bright returns in SAR, changed spectral signature in optical), demolition (rubble spectral signature), vehicle concentration (multiple bright SAR returns), vegetation damage (NDVI decrease), fire scars (SWIR anomaly in Sentinel-2).
- Confidence assignment: Changes visible in both SAR and optical, corroborated by independent sources (social media imagery, news reporting), receive High confidence. SAR-only or optical-only with no corroboration receives Medium. Single-source, single-scene receives Low.
Mensuration
Shadow-based mensuration allows analysts to estimate structure heights and object dimensions from satellite imagery:
- Shadow length method: Shadow length (pixels × GSD) / tan(solar elevation angle) = object height. Solar elevation angle at time of imagery acquisition is available from metadata.
- Building footprint: Roof area (pixels × GSD²) provides structure size estimate. Useful for estimating missile silo dimensions, hangar capacity, and bunker footprint.
- SunCalc.net provides solar elevation and azimuth for any location/date/time combination — critical for shadow-based mensuration.
Integration with ADS-B and AIS
GEOINT achieves its highest analytical value when corroborated with flight tracking (ADS-B) and maritime vessel tracking (AIS). The methodology:
- Define AOI bounding box over crisis theater
- Query ADS-B Exchange (no filtering, includes military callsigns) for flights over AOI in target time window
- Query Global Fishing Watch for vessel tracks and AIS gap events in AOI
- Compare with latest SAR imagery for vessel wake confirmation (SAR detects wake without AIS signal — “dark vessel” detection)
- Corroborate military flight callsigns with Sentinel-1 imagery showing ground activity at airfields
See OSINT Toolkit Essentials for the full ADS-B and AIS tool stack.
AI and Computer Vision in GEOINT
Object Detection and Classification
Computer vision models applied to satellite imagery automate what previously required human analyst review of every image frame:
- Building footprint detection: Automatically segments building footprints from high-resolution imagery — used for population estimation, damage assessment
- Military equipment classification: YOLO and ResNet variants trained on labeled satellite imagery can detect and classify tanks, artillery, trucks, and aircraft at medium resolution
- Change detection automation: Sentinel Hub’s Statistical API enables scripted comparison of time-series imagery — anomalous pixels above a threshold generate automated alerts
Commercial platforms: Picterra (object detection + change detection, SaaS), SpaceKnow (industrial facility monitoring), Orbital Insight (population/economic activity from imagery).
SAR-Optical Fusion
Combining Sentinel-1 SAR with Sentinel-2 optical provides cross-modal confirmation:
- SAR detects metallic objects (vehicles, shipping containers, aircraft) even under cloud or at night — optical provides visual identification and feature classification
- SAR ship wakes persist after vessel transit — when corroborated with optical imagery showing the vessel type, combined confidence for vessel activity assessment rises to High
Adversarial ML Countermeasures
State adversaries have deployed countermeasures against CV-based GEOINT:
- Adversarial patches: Markings applied to vehicles that cause YOLO-class detectors to misclassify them — demonstrated in academic research, deployment in operational environments inferred
- Radar-absorbent materials (RAM): Reduce SAR bright return from military vehicles and installations
- Camouflage netting: Defeats optical classification; partially defeats SAR depending on netting construction and frequency
Camouflage, Concealment, and Deception (CC&D) Against GEOINT
State actors with mature GEOINT programs have developed systematic CC&D doctrine:
- Russian maskirovka: Inflatable decoy tanks and artillery, retroreflective road markings to mimic armored vehicles in SAR imagery, false radio traffic to support deceptive GEOINT picture. Deployed extensively in 2022 Ukraine invasion preparation — with limited success against the volume and cadence of commercial satellite coverage.
- PLAN carrier operations: Chinese aircraft carriers are periodically camouflaged with netting during shipyard maintenance periods to limit imagery analysis of modification work.
- Iranian Fordow nuclear facility: Built underground at 80m+ depth, designed to survive all but the most penetrating bunker-busting munitions. Satellite imagery confirms the surface presence but cannot image the underground infrastructure — creating a persistent intelligence gap exploited by Iran as strategic ambiguity.
- DPRK underground tunnels: 38 North (Stimson Center), using commercial satellite imagery from Planet and Maxar, has mapped North Korean underground tunnel network entrances and egress infrastructure at missile bases and nuclear facilities. The tunnel interiors are inaccessible to imagery intelligence — another deliberate CC&D design.
Assessment: Commercial GEOINT has materially reduced state actors’ ability to conceal large-scale military operations (force concentrations, construction, equipment movement). It has not resolved the intelligence gap against underground facilities, small-unit operations, or activities conducted at night under cover structures.
Case Studies
Case Study 1: Cuban Missile Crisis (1962)
On 14 October 1962, a USAF U-2 (piloted by Major Richard Heyser) photographed the San Cristóbal area of Cuba, capturing imagery that CIA photo interpreters identified as medium-range ballistic missile (MRBM) infrastructure consistent with Soviet R-12 missile sites. The imagery provided President Kennedy with the certainty required to implement a naval quarantine and present unambiguous evidence to the UN Security Council, denying Soviet denials. Soviet Ambassador Zorin’s public denial (“Am I talking to the court of the United States?”) was met by Adlai Stevenson’s display of the imagery — the first public use of classified satellite imagery for diplomatic effect. The crisis established the precedent: GEOINT, when credible, is diplomatically non-deniable.
Case Study 2: South China Sea Island Construction (2013–2016)
The CSIS Asia Maritime Transparency Initiative (AMTI), using Planet Labs and DigitalGlobe commercial imagery, documented the construction of approximately 3,200 acres of artificial island infrastructure in the Spratly Islands — creating airstrips, radar installations, and port facilities on features previously at or below sea level. The GEOINT documentation predated and was more specific than any official government disclosure, forcing PRC acknowledgment and enabling targeted diplomatic responses at ASEAN and UN forums. The case demonstrated that commercial GEOINT can impose diplomatic accountability for activities that a state intends to conduct covertly and deny publicly.
Case Study 3: Ukraine Invasion Preparation (2021–2022)
From November 2021 through February 2022, US and European governments and independent analysts (Maxar Technologies, Planet Labs, GeoConfirmed) documented Russian force buildup along three axes — northern Belarus, eastern Donbas, and southern Crimea. The satellite imagery showed armored vehicle concentrations, field hospital establishment, fuel depot pre-positioning, and railway logistics activity. This GEOINT, combined with OSINT (social media geolocation of unit movements) and SIGINT context, produced high-confidence assessments of imminent invasion that were publicly released — denying Russia the strategic surprise on which its operational plan depended. The Maxar imagery releases became major news events, demonstrating that commercial GEOINT can function as a strategic communication instrument.
Key Connections
Sub-disciplines and complements: IMINT — imagery intelligence as GEOINT’s historical precursor Signals Intelligence — SIGINT-GEOINT fusion for target location OSINT — GEOINT is a primary open-source collection vector Geolocation Methodology — the analytical methodology for image-based location verification OSINT Toolkit Essentials — ADS-B, AIS, satellite access tools
Active conflicts where GEOINT is primary: Ukraine War | US-China Strategic Competition | South China Sea
Institutional actors: CIA | National Geospatial-Intelligence Agency (NGA) | ESA Copernicus Program
Applications in current investigations: The IDF’s Kill Machine — GEOINT-supported targeting methodology IDF Kill Machine Thematic Study — AI integration in GEOINT-based targeting
Pattern-of-life complement: Pattern of Life Analysis — GEOINT provides the location-anchoring vector for POLA