One Year Evaluation of Three Low-Cost PM2.5 Monitors.
One Year Evaluation of Three Low-Cost PM2.5 Monitors.
复制标题
三款低成本 PM2.5 监测仪的一年评估。
DOI：
10.1016/j.atmosenv.2020.117615
复制
发表时间：
2020
期刊：
影响因子：
--
通讯作者：
Koehler,Kirsten
中科院分区：
文献类型：
--
作者：
Zamora,MistiLevy;Rice,Jessica;Koehler,Kirsten
由于传感技术的进步、城市污染意识的增强以及公民科学的兴起，过去几年中，用于家庭使用的低成本监视器的可用性迅速增加。用户友好的软件包使其在研究级室内暴露评估中具有吸引力，但公开市场上的许多监测仪尚未进行严格的科学评估，这导致数据有效性的不确定性。此外，之前的许多传感器研究都是在相对较短的时间内进行的，这可能无法捕捉到此类仪器随着时间的推移可能表现出的变化（称为传感器老化）。我们在 12 个月的时间里对有人居住的无烟住宅中的三台显示器（AirVisual Pro、Speck 和 AirThinx）进行了评估，以评估传感器、内置校准以及为实现长期部署的高精度而对额外数据的需求。对每种类型的监测器的两个单元进行评估，以评估单元之间的精度，并将带有过滤器的个人 DataRAM (pDR-1200) 在家中放置约 20% 的采样周期（例如，每月约一周），以评估随时间变化的准确性。与 pDR 并置期间的平均 PM2.5 质量浓度为：重量校正 pDR（以下简称 pDR 校正）为 5.31 μg/m3，AirVisual Pro 装置为 5.11 和 5.03 μg/m3，Speck 装置为 13.58 和 22.68 μg/m3，以及 7.56 和 7.56 μg/m3。 AirThinx 装置为 7.57 µg/m3。与过滤器相比，AirVisual Pros 表现出最佳准确度，约为 86%，与过滤器重量 (84%) 相比，略好于 pDR 的浊度组件。 Speck（-174 和 -405%）和 AirThinx（42 和 40%）监测器的精度要低得多。当按空气质量指数 (AQI) 对 1 分钟平均 PM2.5 质量浓度进行分类时，pDR 校正与 AirVisual Pro、Speck 和 AirThinx 垃圾箱的匹配率分别约为 97%、40% 和 87%。对于 AirVisual Pro、Speck 和 AirThinx 装置，装置对与 pDR 之间的 Pearson 相关系数 (R2) 分别为 0.90/0.90、0.50/0.27 和 0.92/0.93。 AirVisual Pro、Speck 和 AirThinx 的同一类型设备之间的 R2 分别为 0.99、0.17 和 1.00。除了制造商的校正例程之外，通过添加滤波器校正和后处理以校正已知偏差，所有监视器都可以实现更好的精度。每月校准可产生最高的精度，但对于许多应用，只需对 Air Visual Pro 和 AirThinx 进行一两次校准即可实现几乎同样高的精度。总的来说，这种新型低成本监视器在科学研究中显示出令人兴奋的潜力。然而，三台监测仪中只有一台在没有任何后处理或额外测量的情况下表现出高精度（真实质量浓度的 20% 以内），因此在使用数据自信地评估住宅暴露之前，必须对每台监测仪进行评估。
The availability of low-cost monitors marketed for use in homes has increased rapidly over the past few years due to the advancement of sensing technologies, increased awareness of urban pollution, and the rise of citizen science. The user-friendly packages can make them appealing for use in research grade indoor exposure assessments, but a rigorous scientific evaluation has not been conducted for many monitors on the open market, which leads to uncertainty about the validity of the data. Furthermore, many previous sensor studies were conducted for a relatively short period of time, which may not capture the changes this type of instrument may exhibit over time (known as sensor aging). We evaluated three monitors (AirVisual Pro, Speck, and AirThinx) in an occupied, non-smoking residence over a 12-month period in order to assess the sensors, the built-in calibrations, and the need for additional data to achieve high accuracy for long deployments. Two units of each type of monitor were evaluated in order to assess the precision between units, and a personal DataRAM (pDR-1200) with a filter was placed in the home for about 20% of the sampling period (e.g., about a week each month) to evaluate the accuracy over time. The average PM2.5mass concentration from the periods of colocation with the pDR were 5.31 μg/m3for the gravimetric-corrected pDR (hereafter pDR-corrected), 5.11 and 5.03 μg/m3for the AirVisual Pro units, 13.58 and 22.68 μg/m3for the Speck units, and 7.56 and 7.57 μg/m3for the AirThinx units. The AirVisual Pros exhibited the best accuracy compared to the filter at about 86%, which was slightly better than the nephelometric component of the pDR compared to the filter weight (84%). The accuracies of the Speck (−174 and −405%) and AirThinx (42 and 40%) monitors were much lower. When the 1-min averaged PM2.5mass concentrations were categorized by air quality index (AQI), the pDR-corrected matched the AirVisual Pro, Speck, and AirThinx bins about 97, 40, and 87% of the time, respectively. The Pearson correlation coefficients (R2) between the unit pairs and the pDR were 0.90/0.90, 0.50/0.27, and 0.92/0.93 for the AirVisual Pro, Speck, and AirThinx units, respectively. The R2between units of the same type were 0.99, 0.17, and 1.00 for the AirVisual Pro, Speck, and AirThinx, respectively. All of the monitors could achieve better accuracy by adding filter corrections and post-processing to correct for known biases in addition to the manufacturer's correction routine. Monthly calibrations yielded the highest accuracies, but nearly as high of accuracies could be achieved with only one or two calibrations for the Air Visual Pro and the AirThinx for many applications. In general, this type of new low-cost monitor shows exciting potential for use in scientific research. However, only one of the three monitors exhibited high accuracy (within 20% of the true mass concentration) without any post processing or additional measurements, so an evaluation of each monitor is essential before the data can be used to confidently evaluate residential exposures.