Introducing Delta Signals: Geospatial Event Intelligence for Satellite Monitoring
Quick Answer: Delta Signals is Off-Nadir Delta's geospatial event-intelligence layer. It folds news reports of the same real-world event into one event and adds events built from satellite and official measurements. Each event is mapped next to satellite imagery, with an AI-written summary, a rule-based Delta GEOINT score level and named actors, and the record is updated continuously through the day. It is evidence-based: each event keeps the reports it was built from, and its location is placed only as precisely as the reporting allows — country, town, or block and facility — and says which. It helps analysts connect satellite-observed changes to the real-world events that caused them.
Satellite imagery tells you what changed. Delta Signals tells you why.
Delta Signals is the event-intelligence layer inside Off-Nadir Delta. It places geolocated geopolitical and security event signals directly on the map, alongside your satellite data, so the report and the imagery sit in one view. It is also the event layer behind the situational-awareness dashboard — the broader picture this feeds into. This post explains what it does and how analysts use it.
The Gap Between Pixels and Ground Truth
When you monitor an area with satellite imagery, you can detect changes with remarkable precision: burned vegetation, flooded roads, shifting coastlines, new construction. But the imagery alone rarely tells you the cause.
- Did that vegetation loss happen because of drought — or because of conflict-related displacement?
- Is that sudden drop in nighttime lights an economic disruption, or something else entirely?
- What triggered the movement you detected near that port?
Intelligence analysts have always bridged this gap manually, correlating satellite observations with open-source reporting. Delta Signals automates that correlation, bringing it directly into the map interface.
What Delta Signals Provides
Delta Signals reads publicly available reporting — armed conflict, strikes, military activity, maritime incidents, natural disasters, infrastructure failures, protests, humanitarian crises, and more — and adds events made from measurements rather than reports (for example, thermal anomalies at facilities and unusual drops in strait or canal transits). The AI analysis produces:
For every event:
- A summary written from the reports that make up the event
- The actors named, the event category, and the place, resolved only as finely as the reporting supports
- A Delta GEOINT score from 1 to 5, looked up by fixed rules from how much physical harm the reports establish (the Delta Impact Scale) and whether satellite imagery can check the event (Delta GEOINT Readiness); how it is worked out
- Threat vector (from the air, on the ground, at sea, cyber, economic, political)
- Escalation trend (escalating / de-escalating / stable / new), judged from what the reporting itself says about the change
- Impact scope (local / regional / international)
At the map level:
- Event points colored by category and sized by Delta GEOINT score level
- A stated location level for every event, so a country-level report is not drawn as a pinpoint
- Continuous updates through the day, with consolidation: the same event reported by many outlets appears once, not hundreds of times
Designed for Analysts, Not Data Engineers
Delta Signals is built to require zero configuration. On the map, open the Signals panel from the sidebar, choose a time window, and the events appear as points on the map. Click one to open the event and assess its relevance to what the satellite is showing.
The categories correspond directly to satellite monitoring use cases:
| Signal Category | Satellite Evidence to Correlate |
|---|---|
| Kinetic / Armed Conflict | SAR backscatter changes, building damage, displacement |
| Maritime | Ship detection, port activity, coast guard movements |
| Protest / Civil Unrest | Crowd density, road blockages, nighttime light anomalies |
| Humanitarian | Settlement expansion, IDP camp emergence, supply convoy routes |
AI Enrichment: More Than a Summary
A news report is often vague about place, and a headline alone says little about what happened. The analysis goes further:
Location refinement: Many reports name only a country or a large region. The analysis reads the most specific place the text names for where the event happened and resolves it as far as that place allows: country or province, town or city, or a block or named facility. An event that only resolves to a country is not drawn as a point.
Consolidation: A significant event generates dozens of news articles. Our deduplication logic identifies events that describe the same real-world occurrence — even when the framing, actors named, or time of reporting differ — and consolidates them into a single canonical signal. What you see on the map reflects the actual number of events, not the volume of coverage. An event appears from its first admitted report rather than waiting for many outlets to pick it up — a low headline count early on is not the same as low importance. Merges are recorded rather than overwritten, so a wrong one can be undone.
Threat and impact assessment: Each event is evaluated on dimensions that matter to monitoring analysts: Is this escalating? Does it cross a border? Is there infrastructure at risk? These assessments are produced the same way for every event, so they can be compared across the feed — but they are model readings, and the reports behind them are what you check.
Coverage and Scope
Delta Signals focuses on events with a physical footprint — a report is admitted when its subject is a recent event that has left, or is likely to leave, a trace on the ground that a satellite could in principle look at. Reports that are only about statements, negotiations, courts, elections, markets or opinion are left out, as are anniversary pieces and retrospectives.
Coverage is global and updates through the day, covering:
- Kinetic incidents and armed conflict (strikes, attacks, front lines)
- Military activity and weapons tests
- Maritime events (incidents at sea, seizures, groundings, spills)
- Natural disasters and infrastructure failures
- Protest and civil unrest with geographic specificity
- Humanitarian emergencies with displacement or supply implications
Alongside news reporting, some events come from official feeds and measurements — for example volcanic ash advisories and significant earthquakes — which news reporting can miss entirely.
The record begins on 23 September 2026 and is kept, so you can correlate events with your monitoring time series; how far back you can browse depends on your plan.
Evidence and Auditability
Delta Signals is not a black box. Every signal is built to be checked, not just read — the same principle that separates rigorous open-source intelligence from rumor. When an open-source finding can be shown, shared, and checked by others, it holds up; a claim you cannot trace does not. Each Delta Signal carries the evidence to do exactly that:
- The reports it was built from. Every event built from reporting keeps the articles it was made from. The event's Sources section lists them with links to the original articles (which plans open that section is shown on the pricing page) — the summary never stands in for the source.
- Reports, consolidated. When many outlets report the same event, they are merged into a single event. What you see reflects real-world events, not the raw volume of coverage. Each event counts the distinct outlets that carried it — breadth of reporting, not independent confirmation, because syndicated reprints of one source are not collapsed (see the methodology).
- A resolved location, with its precision stated. The location is read from the article text — the place the event actually occurred, not where it was reported from — and resolved as far as the text supports. An event whose place resolves only to a country, falls outside the country the reporting names, or cannot be placed at all stays at the
reportedstage and is not drawn as a point. Above it,localizedmeans a town or city andpinpointedmeans a block or a named facility. The API carries the resolved level (location_level) and how the coordinate was obtained (geo_status), so a broad answer is never read as a precise one. - An admission test. Reports whose subject is not a physical event — statements, courts, markets, opinion, retrospectives — are left out before anything is built from them, so the feed stays events, not chatter.
The result is an event layer you can defend. When an event informs a decision, you can open its reports, see how precisely its location was resolved, and hand the same evidence to a colleague — the audit trail travels with it.
Using Delta Signals with Monitoring
The most powerful workflow combines Delta Signals with area monitoring:
- Draw a monitoring area around any region of interest
- Enable a satellite index (SAR backscatter, NDVI, nighttime lights, etc.) to track changes over time
- Open Delta Signals and navigate to the same date range
- Look for signals that coincide — spatially and temporally — with anomalies in your satellite time series
When the satellite shows an anomaly and Delta Signals shows a kinetic event in the same area on the same day, you have a correlated observation that is far more actionable than either source alone.
Where to See It
The fastest way to see Delta Signals is the Delta Watchfloor — the situational-awareness dashboard, which lists events with how many outlets carry each one, how precisely it is located, and whether a satellite could see it. Browsing the Watchfloor is free; how much of each event's detail opens, how far back you can look, and how much you can do with the Delta Agent depend on your plan (see pricing).
You can also work signals directly on the map: open one, pull Sentinel-1 SAR or Sentinel-2 optical over its location, and correlate it with a monitored area's time series. For the wider discipline this sits inside — geolocating, corroborating, and confirming open-source reports with imagery — see what is geospatial OSINT.
Delta Signals uses publicly available reporting and open data, analysed by Off-Nadir Delta's own AI processing. The platform is an independent project and is not affiliated with any organization or institution.

Remote sensing specialist with 10+ years in satellite data processing and AI. Founder of Off-Nadir Lab. Ph.D. in Informatics (Yokohama National University, 2026). Master's in Earth System Science and Technology (Kyushu University). Co-author, Remote Sensing Encyclopedia. More about the author →