The Dataset

An open-source structured dataset documenting every documented salvo of the largest missile defence engagement in military history. Covering all four rounds (TP1–TP4), the dataset provides machine-readable records of attack salvos, munitions, targets, interception outcomes, and international reactions. Scraped with AI and OSINT from open sources.

Rounds Tracked

4

TP1 (2024) through TP4 (2026)

Salvos Documented

40+

Individual attack salvos across all rounds

Fields per Salvo

76+

Timing, weapons, targets, interception & more

Reactions Tracked

210

States and multilateral organisations

Access the Dataset

Canonical Source

Full relational database with commit history, documentation, and structured JSON exports. This is the primary source that feeds this site.

danielrosehill/Iran-Israel-War-2026-Data

Data Science Platform

Download flat CSV/JSON exports. Run notebooks and explore in Kaggle's data science environment.

kaggle.com/danielrosehill/iran-israel-war-2026

May lag behind GitHub

Flat Export / ML

Download flat exports or load directly with the datasets library for ML pipelines and LLM fine-tuning.

huggingface.co/datasets/danielrosehill/Iran-Israel-War-2026

May lag behind GitHub

Schema & Documentation

Data Dictionary

Field-by-field documentation of every property in the salvo data JSON schema, with types, descriptions, and examples.

Entity-Relationship Diagram

Interactive schema ERD. Click entities to inspect fields and relationships.

Explore the Data

All four rounds of Iranian strikes (TP1–TP4).

Browse salvos by round — weapons, timing, targets, and interception data.

All salvos in reverse chronological order, filterable by operation.

Interactive Leaflet map with launch sites, targets, and military bases.

World Reactions

210 international reactions with interactive choropleth map and statements.

OSINT charts: timing, weapons mix, targets, interception, escalation.

AI-generated content for informational purposes only. Data scraped with AI and OSINT from open sources and should be independently verified.