Assessment of UAV-Derived Urban Change in Coastal Cities Using Time-Series LiDAR and Imagery

Thesis Abstract

Coastal urban areas face persistent challenges from rapid development, shoreline erosion, and sea-level rise, necessitating accurate, timely, and high-resolution assessments of urban change to inform planning and resilience strategies. This study addresses the gap in longitudinal, multi-sensor monitoring that integrates unmanned aerial vehicle (UAV) derived time-series LiDAR and optical imagery to quantify spatiotemporal urban dynamics in coastal cities, with a focus on land-use transitions, vertical urban growth, and shoreline modification. The aim is to develop and validate a robust framework for detecting and characterizing urban change over a five-year period (2019–2024) using synchronized UAV-based LiDAR and high-resolution multispectral imagery. Specific objectives are (1) to generate harmonized time-series 3D city models and spectral composites from UAV platforms; (2) to quantify vertical and horizontal urban growth indicators alongside shoreline and flood-prone zone alterations; (3) to assess the influence of urban development on coastal vulnerability indices; (4) to evaluate the performance of change detection algorithms under coastal conditions and diverse land-cover classes; and (5) to provide actionable recommendations for coastal urban planning under climate risk scenarios. A mixed-methods approach combines a quantitative, empirical design with a limited qualitative component for contextual interpretation. The population comprises the urbanized coastal corridor of Lagos Metropolitan Region, Nigeria, selected for its rapid growth and data-rich environment. A stratified random sampling approach yields a 60-kilometer transect with 150 sample plots distributed across land-use zones (residential, commercial, industrial, green/blue-blue infrastructure) and shoreline interfaces. Data collection employs (i) UAV surveys conducted biannually over a four-year window (2019–2022) and a follow-up 2024 survey, delivering LiDAR point clouds (return-encoded, with mean point density 20–25 points/m2) and calibrated 4-band multispectral imagery (R, G, B, NIR) at 5 cm spatial resolution; (ii) archival satellite imagery and existing terrestrial LiDAR where available to extend temporal coverage; and (iii) ancillary datasets including national cadastral maps, hydrological and sea-level rise projections. Instrumentation includes a RTK-enabled UAV platform, a lightweight LiDAR payload, calibrated radiometric instruments, and standardized ground control networks for centimeter-level georeferencing. Validity and reliability are ensured through cross-sensor co-registration (root-mean-square error < 5 cm for LiDAR- imagery fusion), radiometric normalization, and inter-sensor calibration procedures. Data analysis proceeds through a three-tier pipeline. First, time-series 3D urban models are produced via LiDAR-derived digital elevation models (DEMs), digital surface models (DSMs), and normalized difference height indices, integrated with spectral packages to produce thematic maps of land-use change. Second, change detection employs multi-temporal LiDAR-based height difference, canopy/ built-up height metrics, and object-based image analysis (OBIA) to classify transitions with accuracy assessments (producer and user accuracy target > 85%). Third, statistical analyses—structural equation modeling (SEM) and spatial autoregressive (SAR) models—examine relationships between urban change indicators and coastal vulnerability metrics (inundation risk, erosion rates, and flood exposure), while regression analyses identify predictors of vertical growth versus horizontal expansion. A conceptual model drawing on the land-use/land-cover change framework and the theory of urban resilience in coastal contexts underpins interpretive analyses. Expected findings indicate that coastal urban areas exhibit pronounced vertical densification in peri-urban corridors, coupled with shoreline retreat and increased exposure of informal settlements. The integration of time-series LiDAR with imagery is anticipated to yield higher detection accuracy for subtle changes in elevation and built-form compared with single-sensor approaches, with stronger explanatory power for coastal vulnerability indices when including vertical growth metrics and shoreline dynamics. The study contributes to knowledge by (1) presenting a validated, transferable methodology for UAV-based, time-series coastal urban change monitoring; (2) demonstrating the added value of LiDAR-derived height metrics in conjunction with spectral features for robust change detection; (3) linking urban morphologic evolution to coastal risk indicators within a resilience framework informed by theories of urban morphology and socio-ecological resilience. Implications for practice include improved planning for flood risk management, shoreline protection, and land-use zoning in rapidly expanding coastal cities, as well as guidance for policy-makers on integrating UAV-based monitoring into coastal hazard mitigation programs. The main conclusion posits that timely, multi-sensor UAV monitoring significantly enhances the detection of urban change and its implications for coastal vulnerability, providing a scalable toolkit for evidence-based decision-making; recommendations emphasize standardized data protocols, capacity-building for local authorities, and ongoing refinement of change-detection algorithms under diverse coastal environments.