Integrated mechanistic and data-driven modeling for risk assessment of greenhouse gas production in an urbanized river system

Surrounded by intense anthropogenic activities, urban polluted rivers have increasingly been reported as a significant source of greenhouse gases (GHGs). However, unlike pollution and climate change, no integrated urban water models have investigated the GHG production in urban rivers due to system complexity. In this study, we proposed a novel integrated framework of mechanistic and data-driven models to qualitatively assess the risks of GHG accumulation in an urban river system in different water management interventions. Particularly, the mechanistic model delivered elaborated insights into river states in four intervention scenarios in which the installation of a new wastewater treatment plant using two different technologies, together with new sewage systems and additional retention tanks, were assessed during dry and rainy seasons. From the insights, we applied fuzzy rule-based models as a decision support tool to predict the GHG accumulation risks and identify their driving factors in the scenarios. The obtained results indicated the important role of new discharge connection and additional storage capacity in decreasing pollutant concentrations, consequently, reducing the risks. Moreover, among the major variables explaining the GHG accumulation in the rivers, DO level was considerably affected by the reaeration capacity of the rivers that was strongly dependent on river slope and flow. Furthermore, river water quality emerged as the most critical variable explaining the pCO2 and N2O accumulation that implied that the more polluted and anaerobic the sites were, the higher were their GHG accumulation. Given its simplicity and transparency, the proposed modeling framework can be applied to other river basins as a decision support tool in setting up integrated urban water management plans. [Display omitted] •A novel integrated framework is proposed to assess the risks of GHG accumulation.•A mechanistic model delivered elaborated insights into river states.•Fuzzy models predicted GHG accumulation risks and identified their driving factors.•DO level was among the most important variables explaining the GHG accumulation.•River water quality affected the pCO2 and N2O accumulation rates.