Artificial Intelligence, Machine Learning and Geographic Information Systems for Environmental Monitoring and Climate-Resilient Urban Planning
Folorunso Segun Stephen
Geological Resources and Geological Engineering, College of Geography and Planning, Chengdu University of Technology, Chengdu, 610059, China.
Ebuka Victor Ezelobe
Department of Civil and Environmental Engineering, University of Toledo, Toledo, Ohio, United States.
Chijioke George Edeh
Department of Civil Engineering, Purdue University, West Lafayette, United State of America.
Elorm Adjoa Gomli
Department of Geographic Information Science, Michigan Technological University, Houghton, Michigan, United States of America.
Julius Odemi Brown
Department of Economics, University of New Mexico, Albuquerque, United States of America.
Ifeoluwa Odunayo Olofinsao
Department of Marine Biology, University of Lagos, Lagos, Nigeria.
Confidence Adimchi Chinonyerem *
Abia State Polytechnic, Abia State, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
Urban climate risks are produced through interactions between changing hazards, built form, ecological conditions, infrastructure and unequal social vulnerability. Artificial intelligence (AI), machine learning (ML), geographic information systems (GIS) and Earth observation now allow these interactions to be represented at increasingly fine spatial and temporal scales. Yet improved environmental maps do not necessarily constitute reliable predictions of human exposure, defensible estimates of intervention benefits or evidence of greater urban resilience. This critical narrative review examines the transition from geospatial sensing to planning decisions, focusing on urban heat, flooding, air quality and land-cover change, together with cross-cutting questions of transferability, uncertainty, equity and institutional use. Peer-reviewed studies and authoritative assessments published principally from 2010 to 23 August 2026 were selected through topic-focused scholarly searches and reference checking, with foundational earlier works retained where necessary. Evidence is strongest for extracting environmental patterns from satellite imagery, integrating heterogeneous measurements and identifying spatial inequalities. It is less consistent for out-of-city prediction, causal inference about adaptation measures, the treatment of compound hazards and proof that automated information improves planning outcomes. Surface temperature, near-surface air temperature and heat stress are frequently conflated; flood susceptibility and event-specific inundation answer different questions; and sensor-calibrated pollutant estimates remain vulnerable to network imbalance. Random cross-validation can exaggerate model portability when neighbouring observations share spatial structure. An integrated, decision-oriented appraisal is developed that links the target phenomenon and measurement process to external validation, distributional consequences, feasible interventions and post-implementation evaluation. The evidence supports a qualified role for AI-enabled GIS as an aid to diagnosis, scenario comparison and monitoring, rather than as a substitute for physically informed models, local knowledge or accountable planning. Research priorities include prospective inter-city tests, uncertainty-aware compound-hazard models, equity-sensitive ground validation and evaluations of actual adaptation outcomes.
Keywords: GeoAI, spatial cross-validation, urban heat exposure, pluvial flooding, environmental sensing, spatial justice, decision support, adaptation evaluation