Observed data quality concerns involving low-cost air sensors.
Observed data quality concerns involving low-cost air sensors.
复制标题
DOI：
10.1016/j.aeaoa.2019.100034
复制
发表时间：
2019-07-01
影响因子：
--
通讯作者：
Williams R
中科院分区：
文献类型：
--
作者：
Clements AL;Reece S;Conner T;Williams R
美国环保署的新兴技术研究计划一直在积极评估低成本（< 2500美元）空气质量传感器的性能（新兴人工智能评估）。这项工作仔细记录了该技术类别与监管级仪器的性能，并通过可公开访问的门户网站将研究结果传达给各种利益相关者。虽然没有定义任何给定商用传感器在任何给定应用中的潜在价值，但报告了关键参数，如准确性、精度和其他响应特性。目前，这些产品的性能指标没有监管或制造商的要求（Woodall等人，2017），与更昂贵的监管仪器不同，低成本传感器通常不提供有助于保持性能的校准、流量检查或质量控制程序的手段。商业上可获得的低成本空气质量传感器的增加，加上公民科学家、社区团体和专业人员对其使用的兴趣增加，导致了大量描述环境发现的期刊文章。这些新技术提供了前所未有的能力，可以在更密集的空间尺度、更短的时间尺度和各种条件下（如移动）测量空气质量。即便如此，对最近同行评审期刊文章的广泛审查表明，数据质量问题（例如，一批传感器内的可变性能、异常值的处理、测量伪像）很少被调查、定义或报告为已发表研究结果的一部分（Williams et al., 2018）。考虑到空气质量在历史上有很好的特征，专业人员使用众所周知的最佳实践来执行质量控制检查，有时在适当的时候仔细地从分析中排除数据，这是相当令人震惊的。
The US EPA's emerging technologies research program has been actively evaluating the performance of low-cost (< $2500 USD) air quality sensors (Evaluation of Emerging Ai). This work carefully documents the performance of this technology class versus regulatory-grade instrumentation and communicates the findings to a wide variety of stakeholders via a publicly-accessible web portal. While not defining the potential value of any given commercially-available sensor for any given application, key parameters such as accuracy, precision, and other response characteristics are reported. There are currently no regulatory or manufacturer's requirements regarding performance metrics of these products (Woodall et al., 2017), and unlike more costly regulatory instruments, low-cost sensors often do not provide a means for calibration, flow check, or quality control procedures that can help maintain performance.Increased availability of commercially-available low-cost air quality sensors combined with increased interest in their use by citizen scientists, community groups, and professionals has resulted in a flood of journal articles describing environmental findings. These new technologies offer an unprecedented ability to measure air quality at denser spatial scales, shorter temporal scales, and under a variety of conditions (eg, mobile). Even so, an extensive review of recent peer reviewed journal articles indicated data quality issues (eg, variable performance within a batch of sensors, handling of outliers, measurement artifacts) were rarely being investigated, defined, or reported on as part of published findings (Williams et al., 2018). This is quite alarming considering air quality has historically been well characterized and professionals have used well-known best practices to perform quality control checks and sometimes carefully exclude data from analyses when appropriate.
影响因子：
3.8
作者：
Feinberg, Stephen;Williams, Ron;Garvey, Sam
通讯作者：
Garvey, Sam
影响因子：
2.9
作者：
Woodall GM;Hoover MD;Williams R;Benedict K;Harper M;Soo JC;Jarabek AM;Stewart MJ;Brown JS;Hulla JE;Caudill M;Clements AL;Kaufman A;Parker AJ;Keating M;Balshaw D;Garrahan K;Burton L;Batka S;Limaye VS;Hakkinen PJ;Thompson B
通讯作者：
Thompson B