Assessing the Compatibility of Machine Learning Tools for Mapping Impervious Surfaces of Urban Areas: A Geospatial Analysis of the Dhaka Metropolitan Area (DMPA)

Rapid and unplanned urbanization in the Dhaka city area has significantly accelerated the expansion of impervious surfaces, altering the hydrological balance, intensifying surface runoff, and increasing environmental vulnerability. Therefore, the accurate mapping of impervious surfaces is crucial for evaluating urban growth patterns and designing sustainable, climate-resilient infrastructure. Hence, this study assesses the current impervious-surface landscape and evaluates the performance compatibility of manual digitization and machine-learning-based supervised classification techniques for mapping impervious surfaces across the Dhaka Metropolitan Area (DMPA) using geospatial techniques. Here, the manual digitization estimated the total impervious coverage at 138.08 km² (46.23%), while the machine learning (ML)-based supervised classification produced 155.68 km² (52.12%), revealing a 5.89% overestimation, primarily due to mixed-pixel effects at a 30 m spatial resolution. The classification achieved an overall accuracy of 88.4% and a Kappa coefficient (κ) of 0.84, indicating strong agreement with reference data. The results depict that the most urbanized areas, such as Lalbagh, Sutrapur, Kotwali, and Dhanmondi, exhibited the highest imperviousness. In contrast, planned or peripheral areas, such as Gulshan, Uttara, and Pallabi, retained larger pervious zones due to structured development and better land-use planning. The impervious coverage of Dhaka city far exceeds the ecological sustainability threshold (10%), signifying a critical hydrological imbalance and increased heat risks. The findings affirm that the MLC method is a reliable, scalable tool for urban surface analysis when complemented with high-resolution validation data. This study highlights the urgent need for prioritizing the restoration of green–blue infrastructure, the expansion of pervious surfaces, and the integration of a remote-sensing-based monitoring system to mitigate the escalating environmental pressures of rapid urban growth.