AI-Powered 3D Urban Morphology Intelligence (3D-UMI): A Scalable Framework for Monitoring Vertical and Horizontal Growth in Global South Cities (Case Study: Lucknow, India)

Conventional urban planning in the Global South relies heavily on 2D land-use maps, which fail to capture vertical densification, a critical dimension of contemporary urban growth. This study introduces the 3D Urban Morphology Intelligence (3D-UMI) framework, an integrated, scalable, and open-source pipeline that combines AI-derived building footprints, cloud-based geospatial processing, spatial statistics, and interactive 3D visualization to bridge this data gap. We applied the framework to Lucknow, India, a rapidly growing Tier-2 city, to analyze spatiotemporal urban dynamics from 2016 to 2023. Using the Google Open Buildings dataset (AI-extracted footprints with estimated heights), Sentinel-2 imagery for spectral validation (NDBI), and Geographically Weighted Regression (GWR) to model spatially varying drivers, we uncovered a "two-speed" city. Horizontally, total built-up area expanded by 36.9% (from 75.3 km² to 103.0 km²), primarily along major transport corridors. Vertically, the urban core experienced significant intensification, with average building heights rising by 40% (from 6.8 m to 9.5 m) within the central 50 km². The GWR model (R² = 0.95) revealed that proximity to roads strongly drives suburban sprawl but has negligible influence in the already dense historic center, demonstrating the necessity of localized planning policies. Accuracy validation against independent reference data (N = 150 buildings sampled across varied urban density zones via field rangefinder measurements and high-resolution DSM ground truth) yielded a height RMSE of 2.28 m, confirming the AI data is fit for city-scale trend analysis. Critically, an interactive 3D dashboard (built with Streamlit and PyDeck) overlays actual building volumes against the Lucknow Master Plan 2031 zoning, automatically flagging compliance violations (e.g., high-rises in low-density zones and encroachment on green buffers). The framework demonstrates that advanced 3D urban analytics are no longer exclusive to well-funded cities; leveraging free, open-source tools (Google Earth Engine, OpenStreetMap, and open AI datasets), it offers a transparent, repeatable, and cost-effective solution for intelligence-driven governance in data-scarce environments across the Global South.