Waste to energy spatial suitability analysis using hybrid multi-criteria machine learning approach

Environ Sci Pollut Res Int. 2021 Aug 10. doi: 10.1007/s11356-021-15289-0. Online ahead of print.

Abstract

Municipal solid waste is typically managed in developing countries through various disposal methods, such as sanitary landfills or dumpsites. Alternatively, waste to energy (WTE) systems have been recently adopted to provide sustainable waste management and diversify the energy mix. The abundance of remotely sensed datasets and derivatives, along with the rapid development of artificial intelligence, can offer an effective solution for WTE site selection. In this study, an analytical hierarchy process (AHP)-based framework supported by multiple machine learning algorithms (gradient boosted tree (GBT), decision tree (DT), and support vector machines (SVMs)) was established to explore the optimum location for WTE facilities. Various social, legal, environmental, economic, morphological, and land cover parameters were considered under 11 thematic geospatial raster layers. The proposed framework was applied to the 1.5-million-capita city of Sharjah, United Arab Emirates. A novel approach was developed to incorporate Gaussian dispersion modeling for the expected air pollution emissions from a WTE facility. The results showed that the accuracy performance sequence of the algorithms was 94.6, 93.9, and 91.8% for GBT, DT, and SVM, respectively. It was found that the distance from existing landfills had the most critical impact on the optimum location of the WTE facility, followed by the distance from coastline and elevation. The AHP consistency check revealed an acceptable overall criteria consistency index and the ratio of 0.0344 and 0.019, respectively. The results showed that 16.6% of Sharjah was considered extremely highly suitable areas. This research supports decision-makers in developing local guidelines for siting WTE facilities and determining the most suitable locations for such projects.

Keywords: Air pollution emissions; Analytical hierarchy process; Machine learning; Spatial suitability analysis; Waste to energy.