Assessing Machine Learning Based Supervised Classifiers For Built-Up Impervious Surface Area Extraction From Sentinel-2 Images

Urban development concept requires a way forward towards smarter and easier urban development. But due to unintended urbanization, which contributes to increased impervious surface areas with several environmental issues. There was a requirement to assess and map the urban built-up impervious areas for solving various socio-environmental issues. This work attempts to evaluate and implement machine learning (ML) based supervised classifiers to semi-automate the process for extraction, quantification and mapping of urban built-up impervious areas. The work emphasised on the usage of freely available open-source high-resolution satellite datasets acquired from Sentinel-2 mission. Three different machine learning-based supervised classification techniques were evaluated for a better understanding of feature extraction methods along with suitable classifier for classification of urban impervious areas. It is a well-known phenomenon that an increase in the impervious surface contributes to declining of green cover. Also, a zonal analysis of extracted built-up impervious surfaces was conducted to understand the spatial configuration of the pilot study area. This zonal assessment of urban built-up impervious surfaces can be used as a worthy tool for better sustainable smart cities development. These can serve as a valuable resource for restoring the required urban green cover for better sustainable urban development.