A Bayesian network model for suitability evaluation of underground space development in urban areas: The case of Changsha, China

Abstract

Developing underground spaces has emerged as a crucial strategy for mitigating the pressure on urban land resources. Therefore, an appropriate evaluation of underground space is imperative for the sustainable utilization of urban underground space resources (UUSR). In this study, the Bayesian Network model was employed as a decision-support tool to evaluate the suitability of UUSR in Changsha. A spatial database was constructed to comprehensively understand the socio-economic, geological condition, and the current construction status that impacts UUSR suitability. Furthermore, the analytic hierarchy process (AHP) method was utilized to weigh the contributions of these factors, thereby constructing the conditional probability tables (CPTs). The Changsha urban area was rasterized into 116,554 evaluation units with a size of 100 m × 100 m, and the suitability of each unit was evaluated using Bayesian inference. Finally, the suitability of the UUSR in Changsha was classified into four levels: very high, high, low, and very low. They represented 16%, 30%, 37%, and 17% of the total area. It was discovered that the areas with high suitability were predominantly located in the Changshaxian district, whereas those with low suitability exhibited an inverted Y-shaped distribution in the study area. Based on Bayesian inference, it was revealed that an increase in socio-economic value is associated with a maximum 20% increase in the suitability of underground space. This study has enhanced the available assessment tools for decision-making support when evaluating UUSR and is a valuable reference for urban planners and other professionals working in related fields.