Carbon emissions, wastewater treatment and aquatic ecosystems

As a nexus of environmental pollution, fossil fuel consumption and the global warming, carbon emissions are critical in China's long-term environmental strategies. In the water cycle, carbon is released during wastewater discharge, wastewater treatment, and subsequent changes in aquatic ecosystems. To gain a comprehensive understanding of this entire process, we investigate the intricate connections using balanced panel data from 261 prefecture-level cities in China spanning the period from 2000 to 2020. Each sample is quantified using 48 features derived from hydrosphere, biosphere, anthroposphere, atmosphere, pedosphere and lithosphere. This paper contributes to the relevant studies in the following ways: Firstly, to analyze the basic interaction within the water cycle, we utilize Structural Equation Modeling (SEM). Our results indicate a weak linear relationship between wastewater treatment and carbon emissions. We also substantiate the crucial role of the aquatic ecosystems in carbon fixation. Secondly, in order to comprehend the intricate interactions within the Earth system, we employ eight machine learning models to predict carbon emissions. We observe that extremely randomized trees (ET) exhibit the highest predictive accuracy among these models. Thirdly, in interpreting the ET model, we utilize Explainable artificial intelligence (XAI) techniques, including Shapley Additive Explanations (SHAP) and Accumulated Local Effects (ALE). Our 3D-SHAP analysis reveals heterogeneity in the emission effects of wastewater treatment across different sub-groups, indicating that emissions are especially sensitive to increased wastewater treatment in agricultural and tourism cities. Furthermore, 3D-SHAP analysis of the aquatic ecosystems exhibits a series of spikes, signifying that aquatic plants will abruptly lose their carbon storage ability once the degradation of the aquatic ecosystems exceeds a certain threshold. Finally, our ALE evaluation, depicting the dispersion tendency of feature importance, identifies the uncertainty of wastewater carbon release in agricultural and tourism cities, while also affirming the vulnerability of the aquatic ecosystems.