Applying conditional analysis methods to manage fugitive air pollution from regulated industry sites.

环境保护局对英格兰和威尔士境内可能排放大量有害空气污染物的工业场所进行监管。EA意识到，测量和数据分析技术现在可以用来量化某些类型的工业过程的排放，例如从点源(烟囱、烟囱等)排放的排放。这有助于更好地监管和控制它们。然而，人们也认识到，其他类型的排放，即所谓的逃逸排放，还没有得到很好的理解。Speranza(1)指出，逃逸排放的例子是指来自建筑物、设备、尘土飞扬的道路或储存堆周围的大量非常小的、不同的位置的排放。例如，最近的DEFRA(2)报告强调了对逃逸颗粒物排放量估计的高度不确定性，理由既有测量困难，又高度依赖于地形和气象条件，而地形和气象条件本身在空间和时间上高度多变。这项工作旨在减少与工业现场逃犯释放的量化、归属和管理相关的不确定性。该项目分为三个主要阶段。第一阶段是回顾现有的关于工业排放、逃逸排放的文献，以及以前适用于逃逸排放的现有测量和评估方法。第二阶段致力于分析现有数据，并制定适当的框架，将条件统计分析方法应用于工业场地周围的空气质量数据。在应用条件分析方法时，还必须研究与每个步骤相关的假设和产出的不确定程度。因此，该项目的第三阶段包括研究如何将条件分析和不确定性分析结合起来，以减少最终量化逃逸排放源的不确定性。第二和第三阶段的结合还可以为常规优化条件分析的输出和调查更主观的模型输入参数的影响提供基础。我的硕士论文项目涉及使用分类和回归树技术分析和建模环境时间序列数据。这使我能够审查环境模拟技术，了解将污染源的排放与在接收器测量的空气质量水平联系起来的复杂的潜在过程，并了解考虑大气扩散过程中涉及的不同气象和背景变量可能具有多么大的挑战性。在过去的三年里，我发展了‘R统计和计算’软件的工作经验，并使用它来分析和模拟环境数据。我目前还在使用它开发软件，对时间序列交通和排放数据进行分析和建模。在我的整个博士学位期间，我一直感兴趣的是研究模型输入和模型参数的不确定性如何影响模型输出，例如使用随机化和敏感性分析方法。将EA/LEC研究人员在工业处理/法规和环境分析技术方面的经验与我在环境数据分析、不确定性分析和‘R’编程方面的现有知识相结合，可以帮助我在该领域发展知识，并完成该项目的主要阶段。此外，我将非常感谢有机会更充分地参与“从数据分析到决策制定”的过程，我认为这个项目可以提供这一点，因为它非常注重数据分析，以支持EA逃逸排放管理活动。(1)Speranza，P.A.(1993)美国专利号5,206,818。华盛顿特区：美国专利商标局。(2)DEFRA。(2014)。空气质量污染物清单，英格兰、苏格兰、威尔士和北爱尔兰。

EA regulates industrial sites located within England and Wales which have the potential to emit significant amounts of harmful air pollutants. EA realises that measurement and data analysis techniques are now available to quantify emissions from certain types of industrial processes such as those emitted from point sources (stacks, chimneys, etc.) which help in better regulating and controlling them. However, it is also recognized that other types of emissions, referred to as fugitive emissions, are yet to be as well understood. Speranza (1) identified example fugitive emissions as those originating from a large number of very small, diverse locations around a building, a piece of equipment, a dusty road, or a storage pile. Recent DEFRA (2) report highlights the high uncertainty in the estimates of fugitive emissions of particulate matter, for instance, citing both measurement difficulties and their high dependence on topographical and meteorological conditions which are themselves highly variable in space and time. This work is intended to contribute towards reducing the uncertainties associated with the quantification, attribution, and management of fugitive releases from industrial sites.This project is divided into three main stages. The first stage is to review existing literature on industrial emissions, fugitive emissions, and existing measurement and evaluation methods previously applied to fugitive emissions. The second stage is dedicated to analysing existing data and developing an appropriate framework to applying the conditional statistical analysis method to air quality data surrounding industrial sites. In applying the conditional analysis method, it is also essential to study the level of uncertainty of the assumptions and outputs associated with each step. Therefore, the third stage of this project involves investigating how conditional analysis and uncertainty analysis can be integrated to reduce the uncertainty in the final quantification of fugitive emission source. The combination of the second and third stages could also provide a basis for routinely optimising the outputs of conditional analysis and investigating the impact of more subjective model input parameters.My MSc thesis project involved analysing and modelling environmental time series data using a classification and regression trees technique. This has allowed me to review environmental modelling techniques and understand the complex underlying processes linking emissions at the source to air quality level measured at the receptor and to understand how challenging it can be to account for the different meteorological and background variables involved in the atmospheric dispersion process. In the last 3 years, I have developed working experience of 'R statistical and computing' software and have used it to analyse and model environmental data. I am also currently using it to develop software to analyse and model time-series traffic and emissions data. Throughout my PhD as well, I have been interested in studying how uncertainty in the model inputs and model parameters has an influence on models outputs using for instance randomisation and sensitivity analysis methods. Combining EA/LEC researchers' experience of industrial processing/regulations and environmental analysis techniques with my existing knowledge in environmental data analysis, uncertainty analysis, and 'R' programming can help develop my knowledge in the field and accomplish the main stages of this project. In addition, I would greatly appreciate the opportunity to engage more fully in the 'data analysis to decision making' process, something I think this project, with its strong focus on data analysis to support EA fugitive emissions management activities, could provide.(1) Speranza, P.A. (1993). US Patent No. 5,206,818. Washington, DC: US Patent &Trademark Office.(2) DEFRA. (2014). Air Quality Pollutant Inventories, for England, Scotland, Wales & Northern Ireland.