Integrating Remote Sensing, Geographic Information Systems, and Machine Learning for Environmental Monitoring and Sustainable Resource Management
Ebuka Victor Ezelobe
Department of Civil and Environmental Engineering, University of Toledo, Toledo, United State of America.
Chijioke George Edeh
Department of Civil Engineering, Purdue University, West Lafayette, Indiana, United State of America.
Elorm Adjoa Gomli
College of Forest Resources and Environmental Science, Michigan Technological University, Michigan, United States of America.
Julius Odemi Brown
Department of Marine Biology, University of Lagos, Lagos, Nigeria.
Kayode Elizabeth Omolara
Department of Civil engineering, University of Ilorin, Ilorin, Nigeria.
Ahmed Idowu Agbelejoye
Department of Mathematics and Statistics, Mississippi State University, Mississippi State, United States of America.
Confidence Adimchi Chinonyerem *
Abia State Polytechnic, Aba, Abia State, Nigeria.
*Author to whom correspondence should be addressed.
Abstract
Environmentally sensitive coastal areas are rapidly becoming urbanised, increasing pressure on vegetation, wetlands, surface-water resources and the natural resource base. The current study combines remote sensing, Geographic Information Systems (GIS) and machine learning to evaluate changes in Land Use/Land Cover (LULC) and environmental conditions in the Lekki–Ajah axis of Lagos State, Nigeria. Built-up areas, vegetation, wetlands, surface water and bare/exposed land were mapped using multitemporal satellite imagery and spatial datasets. Random Forest, Support Vector Machine and Extreme Gradient Boosting algorithms were used to evaluate their performance in land-cover classification, while GIS-based spatial analysis was used to map temporal changes and environmentally sensitive areas. The Random Forest model achieved an overall accuracy of 92%, indicating good performance in discriminating the major land-cover categories within the study environment. The results show substantial urban growth and continued shrinkage of natural and semi-natural vegetation. Historical evidence for Eti-Osa shows that built-up land increased from 2.94% in 1984 to 43.34% in 2002 and 68.38% in 2014, while vegetation declined from 65.53% to 36.87% and 14.86%, respectively. Mangrove cover also declined from 15.21% to 5.04% and 2.98%, respectively, over the same period. Total built-up land cover in the broader Lekki Peninsula rose from 0.44% to 17.98%, while vegetation cover fell from 68.62% to 55.07% between 1984 and 2014. Recent data also show ongoing urbanisation, with an increase in urbanised land from 5.36% in 2002 to 42.37% in 2022, and approximately 49 km of coastline reported as being affected by coastal retreat. The results demonstrate the potential of remote sensing, GIS and machine learning as effective tools for monitoring the rapidly changing Lekki–Ajah coastal area. Increasing developed surfaces and declining vegetation and coastal ecosystems emphasise the need for ongoing geospatial monitoring, protection of remaining wetlands and vegetation resources, environmentally sensitive development planning, and improved coastal-resource management. The study provides a spatial basis for supporting sustainable urban and environmental planning in the rapidly growing Lekki–Ajah corridor.
Keywords: Remote sensing, geographic information systems, machine learning, environmental monitoring, sustainable resource management