JustAir

JustAir — Cross-instrument, QA/QC-corrected air-quality time-series plus the correction metadata most networks never keep: because the platform is monitor-agnostic and ingests both research-grade regulatory instruments and low-cost sensors at the same sites, it holds paired raw-versus-corrected readings with automated error-detection flags, anomaly labels, backfill records, calibration events and timestamped field-maintenance logs. Layered on top: AI-driven pollution-source attribution labelling each exceedance as mobile, stationary or regional; community-selected monitor siting metadata (res…

JustAir offers (Alternative, Geospatial, Esg, Sentiment) — ~1-4e8 readings to date (~200-500 instruments across six states x sub-hourly reporting x ~6 measured channels x ~5 years of accumulation); plus the QA/QC flag, correction and maintenance-log tables, which are the higher-value, far-smaller component.

Low-cost-sensor calibration and correction model training — the raw-versus-corrected pairs spanning research-grade and low-cost co-located instruments are direct supervised training data, and are the scarcest artefact in this category; sensor-fault and anomaly detection model training from labelled drift/fouling/dropout events; pollution-source attribution and apportionment models trained on the mobile/stationary/regional labels; hyperlocal exposure modeling for environmental-justice and epidemiology studies in US urban industrial airsheds where regulatory networks are sparse; public-health alerting and health-effect research using alert-engagement plus exposure history; compliance and enforcement evidence datasets for state and local agencies; refinery, steel and automotive facility near-source monitoring and community-risk scoring; model validation ground truth for satellite-derived urban PM products.

The data is with 5 years of history.

Coverage spans US, Europe, Other; Environmental & Facilities Services, Application Software; alternative, geospatial, esg, sentiment; equities.