AI in Urban Planning: Transforming City Development

Planning Has Always Been About Uncertainty Urban planners manage decisions about land use, infrastructure investment, and growth corridors with incomplete information and long time horizons. A decision made today about the alignment of a road or the zoning of a new district will shape a city for a generation. The ability to reduce uncertainty — to know with greater confidence where the ground is stable, where infrastructure is deteriorating, where population density is likely to shift — fundamentally improves the quality of planning decisions. Geospatial AI is increasingly the tool that provides that certainty. Land Subsidence: A Hidden Urban Risk One of the most consequential and underappreciated risks in coastal and delta cities is ground subsidence — the gradual sinking of the land surface due to groundwater extraction, sediment compaction, or loading from new construction. In low-lying cities, subsidence of even a few centimetres per year can dramatically increase flood exposure, damage buried utilities, and destabilise foundations. InSAR-derived deformation maps now provide urban planners with annual or sub-annual surface deformation data at spatial resolutions that allow individual city blocks to be assessed. Cities across North Africa and the Mediterranean — many built on Quaternary sediments with high subsidence susceptibility — are ideal candidates for InSAR-based urban risk mapping. From Monitoring to Prediction The next frontier is not observation but prediction. Deep learning architectures trained on multi-year InSAR time series can learn the temporal signature of accelerating subsidence and issue early warnings before deformation reaches critical thresholds. This predictive capability transforms subsidence monitoring from a reactive audit exercise into a proactive planning tool. Spatial AI Beyond Subsidence The applications of geospatial AI in urban planning extend well beyond deformation monitoring. Land-cover change detection algorithms applied to Sentinel-2 time series can track the pace and direction of urban expansion, identifying encroachment on agricultural land, wetland loss, and informal settlement growth — all critical inputs to sustainable master planning. Machine learning models trained on multi-source spatial data — satellite imagery, road networks, census data, building footprints — can produce granular spatial predictions of population growth, service demand, and infrastructure need at the neighbourhood scale. The Role of the Geospatial Consultant Translating satellite data and AI model outputs into planning-grade deliverables requires expertise that bridges geomatics, data science, and urban engineering. Raw deformation values need to be contextualised against geological data, infrastructure asset registers, and building records. Risk classifications must be defensible in a regulatory context. Visualisations must communicate effectively to non-technical decision-makers. ElGharbawi Geospatial Consulting has worked with government agencies and urban development authorities to produce InSAR-based urban risk assessments and geospatial AI analysis reports. Our work bridges the gap between satellite observation and actionable planning intelligence.