Dipartimento di Scienze Statistiche Carlo Cecchi, Università degli Studi di Bari, Bari, Italy.
Environ Monit Assess. 2010 Mar;162(1-4):177-90. doi: 10.1007/s10661-009-0786-z. Epub 2009 Mar 11.
In this paper, an analysis of air quality data is provided for the municipal area of Taranto (southern Italy) characterized by high environmental risks as formally decreed by the Italian government in the 1990s with two administrative measures. This is due to the massive presence of industrial sites with elevated environmental impact activities along the NW boundary of the city conurbation. The aforementioned activities have effects on the environment and on public health, as a number of epidemiological researches concerning this area reconfirm. The present study is focused on particulate matter as measured by PM10 concentrations at 13 monitoring stations, equipped with analogous instruments based on the Beta absorption technology, either reporting hourly, two-hourly, or daily measurements. Daily estimates of the PM10 concentration surfaces are obtained in order to identify areas of higher concentration (hot spots), possibly related to specific anthropic activities. Preliminary analysis involved addressing several data problems: (1) due to the use of two different validation techniques, a calibration procedure was devised to allow for data comparability; (2) imputation techniques were considered to cope with the large number of missing data, due to both different working periods and occasional malfunctions of PM10 sensors; and (3) reliable weather covariates (wind speed and direction, pressure, temperature, etc.) were obtained and considered within the analysis. Spatiotemporal modelling was addressed by a Bayesian kriging-based model proposed by Le and Zidek (2006) characterized by the use of time varying covariates and a semiparametric covariance structure. Advantages and disadvantages of the model are highlighted and assessed in terms of fit and performance. Estimated daily PM10 concentration surfaces are suitable for the interpretation of time trends and for identifying concentration peaks within the urban area.
本文对意大利南部塔兰托市（Taranto）的空气质量数据进行了分析。该城市由于 20 世纪 90 年代意大利政府的两项行政措施，被正式宣布为具有高环境风险的地区。这是由于城市城市群西北边界沿线存在大量具有高环境影响活动的工业场所。上述活动对环境和公众健康都有影响，这一点在该地区的许多流行病学研究中得到了再次证实。本研究主要关注颗粒物，使用 13 个监测站的 PM10 浓度进行测量，这些监测站配备了基于 Beta 吸收技术的类似仪器，可报告每小时、两小时或每日的测量值。为了识别可能与特定人为活动有关的高浓度区域（热点），获得了 PM10 浓度表面的每日估计值。初步分析涉及解决几个数据问题：（1）由于使用了两种不同的验证技术，因此设计了校准程序以实现数据可比性；（2）考虑了插补技术，以应对由于不同工作周期和 PM10 传感器偶尔出现故障而导致的大量缺失数据；（3）获得了可靠的天气协变量（风速和风向、压力、温度等），并在分析中进行了考虑。时空建模采用了 Le 和 Zidek（2006 年）提出的基于贝叶斯克里金的模型，该模型的特点是使用时变协变量和半参数协方差结构。突出并评估了模型的优缺点，从拟合度和性能方面进行了评估。估计的每日 PM10 浓度表面适合解释时间趋势，并识别城市区域内的浓度峰值。
