关键词: 化工园区, 无组织排放, 傅里叶红外光谱, 扩散模型, 排放通量

Abstract: Objective　Industrial parks discharge complex mixtures of air pollutants, including sulfides, nitrogen oxides, particulate matters, volatile organic compounds (VOCs) and other hazardous gaseous contaminants. The diversified and complicated characteristics of pollution factors, coupled with extensive unorganized emissions and frequent leakage points throughout the production processes of chemical industrial zones, have led to recurrent air pollution risks and safety accidents. Therefore, timely pollution monitoring and accurate source tracing are urgently required to identify high-concentration emission points, control pollutant discharge in a targeted manner, and reduce environmental hazards. However, conventional monitoring methods primarily rely on manual sampling with steel cylinders or air bags, combined with offline laboratory gas chromatography-mass spectrometry (GC-MS) or online gas chromatography analysis. Although these point-monitoring devices guarantee high detection accuracy, they are incapable of regional flux monitoring and early warning of directional pollution. Meanwhile, the widely used atmospheric dispersion models such as WRF-CHEM and CALPUFF are designed for medium and large-scale simulation covering hundreds of kilometers, and cannot meet the small-scale tracing demands of chemical industrial parks with limited land areas. To address the aforementioned technical limitations, this study aims to construct an integrated monitoring and source apportionment system by combining advanced optical monitoring equipment with a small-scale diffusion model. It intends to realize real-time online detection, precise pollutant source localization and emission flux quantification within a 1-kilometer range, so as to improve the technical system for air pollution control in chemical industrial parks and provide scientific basis for refined environmental management and green industrial transformation. Methods　In this work, an open-path Fourier transform infrared spectroscopy (OP-FTIR) system was combined with the backward Lagrangian stochastic (BLS) diffusion model to carry out field monitoring and source analysis. The experimental site was located in a typical chemical industrial park in northern China, with a total area of 42.38 km² and multiple highpollution industries including coal power, metallurgy and chemical manufacturing. The OP-FTIR device was installed at the center of the industrial zone, with an optical path of 164 m and an effective monitoring radius of 1 km. A high-precision meteorological sensor was independently installed in proximity to the detector of the OP-FTIR device, which synchronously collected real-time data of wind speed, wind direction, temperature, humidity and air pressure at a one-minute interval, providing basic meteorological parameters for subsequent model simulation. For spectral analysis, the nonlinear least squares (NLLS) fitting algorithm was adopted to break through the limitation of Beer's law nonlinear deviation in actual measurement. A composite baseline correction method integrating empirical mode decomposition (EMD) and median filtering was proposed to eliminate spectral baseline drift caused by environmental interference and instrument fluctuation. The characteristic absorption bands of eight typical pollutants, namely hydrogen chloride (HCl), sulfur dioxide (SO2), ammonia (NH3), acetone, formic acid, ethylene, acetylene and methane, were selected based on the HITRAN and QA soft spectral databases to avoid mutual gas interference and water vapor absorption disturbance. In terms of source tracing, the BLS model embedded in the WindTrax software was applied to simulate short-range turbulent diffusion and particle backward trajectory. By filtering valid meteorological data and eliminating invalid low-wind-speed samples, the downwind concentration data monitored by OP-FTIR were imported into the model to calculate the emission flux of key factories and achieve quantitative source apportionment. Results and Discussion　A continuous 14-day field monitoring experiment was successfully conducted in the chemical industrial park with stable operation of the OP-FTIR system, and the eight typical gaseous pollutants were effectively detected. The spectral inversion results showed that the root mean square errors of spectral fitting for all pollutants were maintained at a low level, verifying the high accuracy and stability of the NLLS fitting and baseline correction algorithm. Obvious diurnal variation characteristics were observed for the eight pollutant concentrations. HCl, SO2, acetone and formic acid presented higher concentrations in the afternoon and lower values at dawn, while ammonia and several hydrocarbon gases showed opposite variation trends, which reflected the periodicity of industrial production activities. According to the concentration variation patterns, pollutants were divided into two homologous emission groups. Based on wind field analysis, it showed that high pollutant concentration values mainly occurred in the southwest and northwest regions, which were identified as core potential emission areas. Further quantitative calculation using the BLS model revealed distinct emission contributions of different production zones. Specifically, Factory 1 was the dominant emission source for HCl, formic acid and acetone, Factory 3 contributed the most to SO2 emission, and Factory 2 showed relatively low comprehensive emission level. Conclusions　The combination of OP-FTIR line monitoring technology and small-scale BLS model perfectly overcomes the defects of traditional point-source monitoring and large-scale dispersion models. This combination method requires no complex layout of monitoring points, and is particularly suitable for the complex building distribution and unorganized leakage environment of chemical industrial parks. The research findings confirm that the collaborative application of optical spectral monitoring and backward diffusion simulation can achieve high-precision regional monitoring, accurate source location and quantitative flux evaluation. The proposed combination method fills the technical gap in small-scale air pollution source tracing of chemical industrial parks, and offers reliable technical support for targeted pollution reduction, unorganized emission supervision and sustainable environmental development of industrial parks.

Key words: industrial parks, unorganized emissions, Fourier transform infrared spectroscopy, diffusion model, emission fluxes