Data Sources

Methodology Notes

• Baseline period: 15 Jan – 27 Feb 2026 (pre-war)
• % change: (current − baseline) / baseline × 100
• Primary analytical lens: shock → buffer → service → welfare. Missing evidence lowers coverage and widens the reported bounds; it is not treated as zero stress.
• Counterfactual order: same-season prior years, then a date-matched control, then a robust pre-event median. The selected method and sample count travel with every estimate.
• Legacy compound index: retained for continuity while the layered model is validated; correlated inputs should be interpreted through the new evidence-family grouping.
• DNB L1B caveat: No lunar illumination or straylight correction — absolute values less reliable than relative day-to-day trends
• Nightlight quality: Black Marble Mandatory Quality Flag 0 pixels only; cloud-affected or thin observations are marked partial.
• Ramadan caveat: The 15 Jan–27 Feb baseline includes Ramadan from mid-February. Future baselines should stratify pre-Ramadan and Ramadan behavior.
• Infrastructure evidence: FIRMS thermal anomalies near a facility are proximity evidence only. A detection does not establish cause or damage; an absence cannot distinguish normal operations from source-coverage and detection limits.

Download Data

All data is available as JSON: meta · nightlights · fires · fires (GeoJSON) · compound risk · internet · flights · thermal · SAR vessels · vegetation / NDVI · fuel security · validated fuel ensemble · analytical lenses

Citation

Gulf Humanitarian Systems Monitor. Crisis Evidence Pipeline. [Date accessed]. Available at: https://danwaterfield.github.io/satint-public/