Towards a hybrid Earth observation and social media framework for near-operational detection of temporary urban accessibility barriers

Temporary disruptions to pedestrian infrastructure—seasonal snow and ice, short-term construction, flooding and unplanned road closures—impose recurring but poorly quantified mobility constraints on urban populations. Conventional Earth observation is insufficient for detecting the sub-weekly, localized disturbances that define transient accessibility barriers, while existing social media analytics lack the spatial grounding required for operational urban monitoring. This paper proposes the Earth Observation and Social Media Accessibility Framework (EOSMAF), a five-stage conceptual pipeline that integrates natural language processing (NLP) of geolocated social media data with multitemporal Sentinel-1 synthetic aperture radar (SAR) and Sentinel-2 multispectral imagery. The stages comprise social media corpus construction with bilingual NLP barrier-topic extraction; satellite change detection via log-ratio SAR analysis and spectral-index computation; spatial fusion through a weighted evidence-scoring model; and graph-theoretic pedestrian-network impact assessment. The framework is illustrated through a conceptual scenario representative of northern Canadian urban conditions and is positioned as a near-operational, latency-bounded system rather than a real-time one. Its modular architecture supports transfer across urban contexts, with relevance to inclusive mobility planning, smart-city monitoring and climate-adaptive infrastructure management. Validation requirements, parameter calibration and geolocation uncertainty are discussed critically.