【摘要】 大气CO₂浓度的增加已成为无可争议的事实，在时空维度上获得足量的CO₂浓度监测数据，对准确把握城市碳排放特征并制定相应的减排措施具有至关重要的作用。在广州大学城内选择林荫大道、社区广场、茂密森林和繁忙路口四个典型景观类型，尝试结合卡尔曼滤波和物联网技术以提升低成本传感器监测CO₂浓度的精度和效率，开展城市近地面CO₂浓度时空变化的精细化分析。结果显示：（1）低成本传感器在相同环境下获取的原始CO₂浓度值一致性较低，F检验和t检验显示差异性明显，但经卡尔曼滤波处理后数据两种检验方法显示数据无差异，表明滤波后监测结果具有高度一致性；（2）林荫大道与社区广场、茂密森林与繁忙路口、林荫大道与繁忙路口之间的CO₂浓度差异非常显著，林荫大道CO₂浓度最低，平均值仅为456.74μmol/mol （±11.83μmol/mol）,繁忙路口的CO₂浓度最高，平均浓度为467.32μmol/mol（±16.04μmol/mol）;（3）不同景观之间的CO₂浓度变化规律昼夜差别明显，林荫大道与社区广场CO₂浓度差异主要发生在晚上（13.30μmol/mol）,而白天的差异很小（3.53μmol/mol）,此外，交通流量较大的繁忙路口在上午7至11时CO₂浓度最高，而下午时段与其它三种景观差别最大。研究成果可为有效解决CO₂浓度监测成本与精度之间的平衡问题提供了新的思路，并为快速高效提升碳排放监测能力提供了技术支撑。更多还原

【Abstract】 Carbon dioxide（CO₂） constitutes over 50% of the total climatic forcing attributed to long-lived greenhouse gases, serving as a principal driver of climate change characterized by shifts in surface temperature, alterations in the hydrological cycle, rising sea levels, and an increase in the frequency of extreme weather events. The increase in atmospheric CO₂ concentration has become an indisputable fact. Obtaining sufficient monitoring data of CO₂ concentration in both spatial and temporal dimensions is crucial for accurately understanding urban carbon emission characteristics and formulating corresponding reduction measures. This study selected four typical landscape types—tree-lined avenues, community squares, dense forests, and busy intersections—within the Guangzhou Higher Education Mega Center. We aimed to combine Kalman filtering and Internet of Things（IoT） technology to enhance the accuracy and efficiency of low-cost sensors in monitoring CO₂ concentration, and to conduct a detailed analysis of the spatial and temporal variations in near-surface CO₂ concentration in urban areas. The results indicated that: 1) Low-cost sensors exhibited low consistency in the raw CO₂ concentration values obtained under identical conditions, with significant differences revealed by F-tests and t-tests. However, after Kalman filtering, both statistical methods indicated no differences in the data, suggesting a high level of consistency in the filtered monitoring results; 2) Significant differences in CO₂ concentration were observed between tree-lined avenues and community squares, dense forests and busy intersections, as well as between tree-lined avenues and busy intersections. The CO₂ concentration in the tree-lined avenue was the lowest, at only 456.74 μmol/mol（±11.83 μmol/mol）, while the concentration at the busy intersection was the highest, with an average of 467.32 μmol/mol（±16.04 μmol/mol）; 3) The variation patterns of CO₂ concentration among different landscapes showed pronounced diurnal differences, with the main difference between the tree-lined avenue and community square occurring at night（13.30 μmol/mol）, while the daytime difference was minimal（3.53 μmol/mol）. Additionally, the busy intersection, which had higher traffic flow, recorded the highest CO₂ concentration between 7 and 11 AM, and the greatest discrepancy with the other three landscapes occurred in the afternoon. The findings of this study provide new insights into addressing the balance between cost and accuracy in CO₂ concentration monitoring, and offer technical support for rapidly and efficiently enhancing carbon emission monitoring capabilities.更多还原