Monitoring of Vacant lands in Metropolitan Cities using VHRS data and DL ModelsA Case Study for Delhi

Vacant land - a key asset for implementing several economic development strategies in ever-growing metropolitan cities, which requires frequent monitoring as it faces land use changes due to anthropogenic activities. The spatial information about vacant land is essential for urban planners and decision-makers in order to optimally utilize the land resources. This study focuses mainly on monitoring vacant lands in metropolitan cities like Delhi using very highresolution satellite data and Deep Learning technique. Four sets of temporal satellite images were utilized in the study and about 4000 labeled training samples were generated representing five distinct LU/LC categories for developing a U-net-based DL Model. The DL model has been used for inferencing to generate the classified maps. The study indicated the use of satellite images in identifying the land use/land cover changes during the period 2006 $\left(\mathrm{T}_{0}\right)$ to $2022\left(\mathrm{~T}_{3}\right)$ for the vacant sites under DDA jurisdiction. The LU/LC map of vacant lands were customized into an open-source GIS platform and a decision support system has been developed which would help the user in the easy identification of alarm sites using a query mechanism. The DL model developed can be used to infer the new set of images procured in the future and any area within the jurisdiction of Delhi Development Authority can be analyzed/monitored.