Data Science Applications in Environmental Monitoring Using Remote Sensing Data, Machine Learning Algorithms, and IoT-Based Sensors

Abstract

This study explores the use of advanced information technology methods—satellite remote sensing, machine learning, and Internet of Things (IoT)-based sensors—for environmental monitoring in Lahore, Pakistan. The objectives were to collect and preprocess multi-source environmental data, apply machine learning models (Random Forest, Support Vector Machine, and Artificial Neural Networks) for air quality classification and land surface temperature (LST) prediction, and evaluate their performance using standard metrics. Data were obtained from USGS and ESA satellite platforms, IoT-based air quality sensors, and the Pakistan Meteorological Department. Preprocessing included data cleaning, normalization, and geospatial feature extraction using Python tools. Results show that Random Forest performed best in air quality classification, achieving 91.4% accuracy, while the ANN model provided the most accurate LST predictions with an RMSE of 1.84°C. Key contributing features included NDVI, PM2.5 concentration, and land use classification. The findings demonstrate that integrating remote sensing, machine learning, and IoT technologies provides a reliable and scalable framework for real-time urban environmental monitoring, with important implications for public health and environmental policy in Lahore and similar urban regions.