This study provides a comprehensive analysis of greenhouse gas (GHG) emissions in free surface constructed wetlands (FSCWs) by combining qualitative insights from literature with machine learning-based quantitative analysis. Key factors influencing CO₂, CH₄, and N₂O emissions were identified, with a focus on the effects of seasonal variation, vegetation, substrate, and influent characteristics. The qualitative review highlights vegetation and substrate as critical drivers, noting that specific plant traits, such as root oxygenation, significantly impact methane dynamics. Influents with high nutrient loads, such as agricultural runoff and municipal wastewater, were found to increase GHG emissions, underscoring the importance of influent composition in emissions management. The machine learning analysis, using tree-based models (Random Forest, Gradient Boosting, CatBoost, and XGBoost), further quantified variable importance, revealing that presence of vegetation and All Sky Surface Shortwave Downward Irradiance (SW DNI) are primary drivers for CO₂ and CH₄ emissions, while the age of the Wetland is an important determining factor for N₂O emissions due to its impact on nitrogen cycling. Earth skin temperature, the thermal metric used in this study, showed low importance. This may reflect the fact that SW DNI better captures surface energy inputs that influence microbial activity, or that different thermal variables, such as air temperature, are more relevant in other contexts. These findings emphasize the need for targeted management strategies to optimize GHG mitigation in FSCWs, supporting sustainable wetland design that balances wastewater treatment with climate mitigation goals.
