Smartphone-based Urban Heat Mapping for Real-time Heat Island Mitigation

Thesis Abstract

Urban heat islands (UHIs) exacerbate heat stress, energy demand, and air pollution in rapidly urbanizing environments, particularly where field-based meteorological networks are sparse and costly. This study develops and evaluates a smartphone-based framework for real-time urban heat mapping and mitigation guidance, addressing gaps in spatial resolution, temporal immediacy, and community-driven data collection. The aim is to (i) create a citizen-science platform that collects high-resolution surface temperature proxies using built-in smartphone sensors and calibrated photogrammetric methods, (ii) integrate crowd-sourced data with municipal land-surface models to produce actionable heat maps, and (iii) inform targeted mitigation strategies through real-time feedback to residents and planners. Specific objectives include (1) designing an open-source mobile app that records ambient temperature proxies, GNSS-derived location, time, and contextual imagery; (2) validating smartphone-derived heat indicators against reference infrared thermography and meteorological stations across 20 distinct urban microclimates; (3) developing a calibration and data fusion workflow using Bayesian hierarchical modeling to reconcile heterogeneous data streams; (4) applying machine learning to interpolate spatiotemporal heat patterns at 10?meter resolution and generate near-real-time heat island indices; (5) assessing the potential of the platform to influence behavior and policy through pilot feedback sessions with local residents and city planners. The study adopts a mixed-methods approach within a pragmatic research design, drawing on the theoretical lens of Information for Action and Technological Empowerment to link data quality, user engagement, and decision-support effectiveness. The population comprises urban residents and municipal planners in a mid-sized metropolitan area with diverse land cover and microclimates. A stratified purposive sample of 300 residents participates in app-based data collection over six months, complemented by technical validation at 40 controlled roadside and park-site locations. Data collection instruments include a mobile sensing app, portable reference-grade infrared thermometers during validation campaigns, a structured survey instrument to capture contextual factors, and semi-structured interviews with planners. Validity and reliability are enhanced through cross-validation with satellite-derived land surface temperature products (MODIS/VIIRS) and a calibration protocol that accounts for device heterogeneity, ambient moisture, shading effects, and time-of-day variability. Data analysis employs (i) descriptive statistics and quality-control metrics to assess data reliability; (ii) regression analysis and Bland-Altman checks to compare smartphone proxies with reference measurements; (iii) Bayesian data fusion to integrate smartphone, ground-truth, and satellite data; (iv) random forest and gradient boosting machines for high-resolution spatial interpolation and heat index computation; (v) time-series analysis to identify diurnal and weekly heat dynamics; and (vi) thematic analysis of interview transcripts to gauge stakeholder perceptions and policy implications. The anticipated findings include (a) high concordance between calibrated smartphone proxies and reference measures (mean absolute error under 0.8°C in controlled tests; under 1.2°C in field conditions); (b) robust 10-meter resolution heat maps capturing fine-scale UHIs linked to land cover, solar exposure, and building morphology; (c) evidence that near-real-time heat maps enhance situational awareness among residents and provide tangible inputs for micro-cooling or shading interventions; and (d) a validated workflow for scalable deployment in other urban contexts. The study contributes to knowledge by (i) advancing a cost-effective, participatory method for obtaining high-resolution urban temperature fields; (ii) integrating citizen-generated data with formal urban climate models through a transparent calibration and fusion framework; (iii) demonstrating the practical utility of near-real-time heat maps for targeted mitigation planning and behavioral change. The main conclusion posits that smartphone-based urban heat mapping, when coupled with rigorous data fusion and user-centered design, can deliver timely, actionable heat information at fine spatial scales to support both community resilience and evidence-based urban design. Recommendations include expanding sensor integration (e.g., lightweight thermal cameras), extending the approach to multi-city comparisons to test transferability, and embedding the platform within municipal climate adaptation strategies to institutionalize citizen-science contributions and curb heat-related risks.