Change Detection for Sustainable Redevelopment of AI Identified Brownfields from Satellite Imagery using Statistical and Clustering Techniques - A Case Study on Supply Chain Location Analysis

In the area of Supply Chain Network Design and Location Analysis, it is critical to find an optimal geographic location for production and logistics with a desired area size, usually a large piece of land near the industrial estates or urban settlements, to develop it into a distribution center or a warehouse etc. The reasons may include ease of access to the potential end customers or other business partners for vertical integration of their supply chains. Considering United Nations Sustainable Development Goal 11 and other existing urban planning regulations, it may not always be feasible to locate a suitable greenfield site. One approach to address this problem is by identifying the existing brownfields using high resolution satellite and aerial images. A Machine Learning(ML)-based image classification algorithm is being developed which can make a classification on all available land parcels into brownfields or other active sites with the help of these high resolution images. However, these high resolution images are expensive and difficult to be frequently collected. An economically affordable alternative are more frequently available low resolution images which can be used to validate the classification with an appropriate change detection method. This paper introduces a detailed method for detecting changes in brownfield sites. The process includes initial classification with DOP21 (high resolution digital orthophoto captured around 2021) imagery, enhancement using SPOT21 and SPOT23 (Satellite pour l'Observation de la Terre) images for low resolution analysis and a series of steps. The experimental outcome shows the effectiveness of the proposed method to distinguish physical changes within the areas of interest, demonstrating substantive applicability in large-scale analysis.