Atmospheric Dispersion Research Articles

Radioxenon Releases from A Nuclear Power Plant: Stack Data and Atmospheric Measurements

An analysis of a data set consisting of 3 years of high time resolution radioxenon stack measurements from the three nuclear reactors at the Forsmark nuclear power plant in Sweden, as well as measurements of atmospheric radioxenon in Stockholm air, 110 km away, is presented. The main causes for the stack releases, such as the function of the xenon mitigation systems, presence of leaking fuel elements, and reactor operations such as shutdown and startup, are discussed in relation to the stack data. The relation between radioxenon releases and reactor operation is clearly illustrated by the correlation between the stack measurements and thermal reactor power. In general, the isotopic ratios of the Stockholm measurements, which are shown to mainly originate from Forsmark releases, agree well with stack measurements, and with a modeled reactor operational sequence. Results from a forward atmospheric dispersion calculation agree very well with observed plume arrival times and widths, and with some exceptions, also with absolute activity concentrations. The results illustrates the importance of detailed knowledge of radioxenon emissions from nuclear power plants when interpreting radioxenon measurements for nuclear test ban verification, and provide new input to this kind of analysis. Furthermore, it demonstrates the possibility to use sensitive radioxenon detection systems to remotely detect and verify reactor operation.

Modeling of the Concentrations of Ultrafine Particles in the Plumes of Ships in the Vicinity of Major Harbors.

Marine traffic in harbors can be responsible for significant atmospheric concentrations of ultrafine particles (UFPs), which have widely recognized negative effects on human health. It is therefore essential to model and measure the time evolution of the number size distributions and chemical composition of UFPs in ship exhaust to assess the resulting exposure in the vicinity of shipping routes. In this study, a sequential modelling chain was developed and applied, in combination with the data measured and collected in major harbor areas in the cities of Helsinki and Turku in Finland, during winter and summer in 2010–2011. The models described ship emissions, atmospheric dispersion, and aerosol dynamics, complemented with a time–microenvironment–activity model to estimate the short-term UFP exposure. We estimated the dilution ratio during the initial fast expansion of the exhaust plume to be approximately equal to eight. This dispersion regime resulted in a fully formed nucleation mode (denoted as Nuc2). Different selected modelling assumptions about the chemical composition of Nuc2 did not have an effect on the formation of nucleation mode particles. Aerosol model simulations of the dispersing ship plume also revealed a partially formed nucleation mode (Nuc1; peaking at 1.5 nm), consisting of freshly nucleated sulfate particles and condensed organics that were produced within the first few seconds. However, subsequent growth of the new particles was limited, due to efficient scavenging by the larger particles originating from the ship exhaust. The transport of UFPs downwind of the ship track increased the hourly mean UFP concentrations in the neighboring residential areas by a factor of two or more up to a distance of 3600 m, compared with the corresponding UFP concentrations in the urban background. The substantially increased UFP concentrations due to ship traffic significantly affected the daily mean exposures in residential areas located in the vicinity of the harbors.

Air pollution from industries and asthma onset in childhood: A population-based birth cohort study using dispersion modeling

BackgroundDespite evidence that ambient air pollution may play a role in the development of asthma, little is known about the potential contribution of industrial emissions. ObjectiveWe used a population-based birth cohort to investigate the association between asthma onset in childhood and residential exposure to industrial emissions, estimated from atmospheric dispersion modeling. MethodsThe study population comprised all children born in the province of Quebec, Canada, 2002-2011. Asthma onset were ascertained from health administrative databases with validated algorithms. We used atmospheric dispersion modeling to develop time-varying annual mean concentration of ambient PM2.5, NO2 and SO2 at participants' residence from industries. For each pollutant, we assessed the association between industrial emissions exposure and childhood asthma onset using Cox proportional hazard model, adjusted for sex, material and social deprivation and calendar year. Sensitivity analysis included adjusting for long-term regional and traffic-related ambient PM2.5 and NO2, and assessing potential confounding by unmeasured secondhand smoke. ResultsThe cohort included 722,667 children and 66,559 incident cases of asthma. For all pollutants, we found a non-linear association between childhood asthma onset and residential ambient air pollutant concentration from industries, with stronger effects at lower concentrations. A change from 25th to the 75th percentile in the mean annual ambient concentration of PM2.5 (0.13 μg/m3), NO2 (1.0 μg/m3) and SO2 (1.6 μg/m3) from industrial emissions was associated with a 19% (95% CI: 17-20%), 21% (95% CI: 19-23%) and 23% (95% CI: 21-24%) increase in the risk of asthma onset in children, respectively. For PM2.5 and NO2, associations were persisting after adjustments for long-term regional PM2.5 and traffic-related NO2 ambient concentration.Conclusion: Residential exposure to industrial emissions estimated from dispersion modeling was associated with asthma onset in childhood. Importantly, associations were stronger at lower concentrations and independent from those of other sources, thus adding up to the burden of regional and traffic-related air pollution.

The radiological assessment, hazard evaluation, and spatial distribution for a hypothetical nuclear power plant accident at Baiji potential site

BackgroundThe safety assessment in site selection for a new nuclear power plant is an essential issue for human health. It could be improved by predicting the consequences for a hypothetical accident. This paper is contextual with the nuclear safety regarding the risk upon human health from the hazard constituted by the emission of radioactive material due to a hypothetical nuclear power plant accident. HYSPLIT model used and configured based on terrestrial and meteorological conditions for this purpose.ResultsIt presents an analysis of the time-series and spatial distribution for dispersed radioactive contaminants from a hypothetical accident at Baiji potential site. This is based on the geological and meteorological specifications of the potential site. Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model configured to simulate the atmospheric dispersion of fission emissions, then assesses the public health consequences of a hypothetical nuclear accident. Results indicate that the total individual dose intake by the population living around the potential site from the hypothetical accident exceeded 1 Sv, greater than the allowed dose limits by International Commission on Radiological Protection (ICRP). The probability of cancer incidence at regions on directions of southeast to northeast along 30 km from the potential site was high (0.0378 to 0.00131) risk/person. However, the areas that laid on the west and south-west of the site recorded higher levels of influence compared to regions on other directions. Also, the reduction of exposure dose with the distance from the site presented. The north-direction regions from the Baiji site recorded a rapid reduction of exposure dose to become zero at 40 km distance.ConclusionsFinally, to minimize the radiological impacts on population, emergency procedures are required at the regions that laid on the west and south-west of the site. These countermeasure remedy actions should include evacuation, sheltering, ban the sale of local agriculture productions, and long-range resettlement of the population. Since the plume of radioactive contaminants did not arrive at the regions on east and northeast, 40 km away from Baiji site, these areas could be categorized as a support zone.

Parameterization of the building downwash and sidewash effect using a mixture model

Widely used for permitting new emission sources (e.g., power plants) and assessing human exposure, atmospheric dispersion models do not perform well when the facility stack is relatively low and the pollution plume would be affected by the building wake, such as biomass boilers, diesel backup generators and residential woodstoves, which was also known as building downwash. The current building downwash preprocessor, Plume Rise Model Enhancements (PRIME), in a regulatory model, AMS/EPA Regulatory Model (AERMOD), tends to underestimate the ground-level concentrations (GLCs). Our flow visualization indicated that the flow pattern under the oblique wind was featured by a sidewash-downwash (S-D) vortex, which determines the near-wake concentration field. Thus, we introduced the concept of “sidewash” to highlight the effect of oblique wind conditions, which can cause higher GLCs than those under the perpendicular wind. We presented a new parameterization method, named the mixture model, to represent the concentration field in the downstream of a box-shape building with a short stack under different wind directions. This model was developed on the flow and concentration fields analysis using a computational fluid dynamics (CFD) technique, Embedded Large Eddy Simulation (ELES), and evaluated against experimental data. With both downwash and sidewash, the new parameterization showed promising performance for the low stack height under both perpendicular and oblique wind directions and successfully captured the lateral shifts of the plume under oblique wind directions and continuous concentration distribution along the longitudinal ground-level receptors, which could not be reflected by using the current PRIME. We showed that CFD-generated training data do not need to perfect to provide valuable insight into the parameterization.

Radiological Safety Analysis for a Hypothetical Accident of a Generic VVER-1000 Nuclear Power Plant

Atmospheric dispersion modelling and radiological safety analysis have been performed for a postulated accident scenario of a generic VVER-1000 nuclear power plant using the HotSpot Health Physics code. The total effective dose equivalent (TEDE), the respiratory time-integrated air concentration, and the ground deposition concentration are calculated considering site-specific meteorological conditions. The results show that the maximum TEDE and ground deposition concentration values of 3.69E – 01 Sv and 3.80E + 06 kBq/m2 occurred at downwind distance of 0.18 km from the release point. This maximum TEDE value is recorded within a distance where public occupation is restricted. The TEDE values at distances of 5.0 km and beyond where public occupation is likely to be found are far below the annual regulatory limits of 1 mSv from public exposure in a year even in the event of worse accident scenario as set in IAEA Safety Standard No. GSR Part 3; no action related specifically to the public exposure is required. The released radionuclides might be transported to long distances but will not have any harmful effect on the public. The direction of the radionuclide emission from the release point is towards the north east. It is observed that the organ with the highest value of committed effective dose equivalent (CEDE) appears to be the thyroid. It was followed by the bone surface, lung, red marrow, and lower large intestine wall in order of decreasing CEDE value. Radionuclides including I-131, I-133, Sr-89, Cs-134, Ba-140, Xe-133, and Xe-135 were found to be the main contributors to the CEDE.

Estimating the 3D shape of volcanic ash to better understand sedimentation processes and improve atmospheric dispersion modelling

The sedimentation rate of volcanic ash through the atmosphere influences its travel distance, with important implications for aviation and health. The fall velocity of a particle depends on its size and density, but also shape, and volcanic ash is not spherical. To capture the sedimentation of ash, atmospheric dispersion models use empirical drag equations calibrated using geometric shape descriptors. However, particle shape data are scarce and there is no standard method of shape measurement. In addition, shape measurements are not always available during an eruption, when dispersion models are used operationally to forecast ash hazard. We assess the variability in the shape of volcanic ash from Icelandic eruptions using X-ray computed tomography. To consider how good different drag equations and shape descriptors are at representing the sedimentation of volcanic ash we compare calculated fall velocities to measured fall velocities of volcanic ash in air in a settling column. We then suggest the best drag equations and shape descriptors for use in atmospheric dispersion models. We find that shape-dependent drag equations produce more accurate results than a spherical approximation. However, accurate drag calculations based on the shape descriptor sphericity, which is a function of surface area, require the imaging resolution to be within the range of 102 - 105 voxels per particle (where a voxel is a volumetric pixel) as surface area is sensitive to imaging resolution. We suggest that the large-scale form of the particle impacts sedimentation more than small-scale surface roughness. Shape descriptors based on ratios between principal axis lengths are more practical as they are less variable among particle size classes and much less sensitive to imaging resolution. Finally, we use particle shape data from this study and literature sources to make recommendations on default values for use with atmospheric dispersion models where no shape data are available.

TRANSBOUNDARY RADIOLOGICAL EFFECTS OF A HYPOTHETICAL NUCLEAR ACCIDENT IN NORTHEAST ASIA.

Offsite radiological consequence investigation using computerized software has been considered as an important quantitative risk communication in order to recognize and discuss public concerns about nuclear safety and health risk in case of hypothetical nuclear accidents around specific nuclear power plants (NPPs), with guideline of lessons learned from previous nuclear disasters. In this study, Northeast Asia nuclear accident simulator (NANAS) developed by Nuclear Safety and Security Commission (NSSC) in Korea was used to quantify the offsite radiological consequences from Haiyang unit 1 NPP in China and to examine the emergency protective measures for the public around regions of Korea as NPPs operating in Northeast Asia countries contributed to about 25% of the industry. Broad simulations of radiological source term estimation, atmospheric dispersion analysis and radiation dose assessment to the public have been performed in case of hypothetical nuclear accident involving source term of radionuclides release taken from Fukushima accident.

Origin of Bluetongue Virus Serotype 8 Outbreak in Cyprus, September 2016.

In September 2016, clinical signs, indicative of bluetongue, were observed in sheep in Cyprus. Bluetongue virus serotype 8 (BTV-8) was detected in sheep, indicating the first incursion of this serotype into Cyprus. Following virus propagation, Nextera XT DNA libraries were sequenced on the MiSeq instrument. Full-genome sequences were obtained for five isolates CYP2016/01-05 and the percent of nucleotide sequence (% nt) identity between them ranged from 99.92% to 99.95%, which corresponded to a few (2–5) amino acid changes. Based on the complete coding sequence, the Israeli ISR2008/13 (98.42–98.45%) was recognised as the closest relative to CYP2016/01-05. However, the phylogenetic reconstruction of CYP2016/01-05 revealed that the possibility of reassortment in several segments: 4, 7, 9 and 10. Based on the available sequencing data, the incursion BTV-8 into Cyprus most likely occurred from the neighbouring countries (e.g., Israel, Lebanon, Syria, or Jordan), where multiple BTV serotypes were co-circulating rather than from Europe (e.g., France) where a single BTV-8 serotype was dominant. Supporting this hypothesis, atmospheric dispersion modelling identified wind-transport events during July–September that could have allowed the introduction of BTV-8 infected midges from Lebanon, Syria or Israel coastlines into the Larnaca region of Cyprus.

Assessment of the potential Total Effective Dose (TED) and Ground Deposition (GD) following a hypothetical accident at the proposed Rooppur Nuclear Power Plant

The radiological consequences following a severe hypothetical accident at the proposed Rooppur Nuclear Power Plant in Bangladesh, have been assessed using the health physics code HotSpot 3.1, developed by Lawrence Livermore National Laboratory, USA in this study. The source term for this approach has been calculated analytically. Atmospheric dispersion characteristics have been branded using the site-specific meteorological data of 30 years, obtained from Bangladesh Meteorological Department. The TED and GD values have been calculated in the downwind distance for six individual cases along the plume centerline from the release point up to 200 km. The cases include three seasons namely Summer, Rainy and Winter with three stability classes A, B and F prevailing around the region. The highest value of TED and GD observed is for Rainy season (Stability Class F). The three TED contours have been drawn for 1.00, 0.50 and 0.05 Sv respectively which reflect dose limits for permanent relocation, emergency worker and evacuation, as established by IAEA-TECDOC-953. Demographic analysis reveals that the highest number of affected population corresponds to Rainy season, as it is more spread.

Development and uncertainty analysis of radionuclide atmospheric dispersion modeling codes based on Gaussian plume model

It is necessary to assess the radiological consequences of radioactive leakage accident in the planning and operation of a nuclear power plant, especially an atmospheric radioactive material spill that has a rapid and broad impact on public health. Uncertainty analysis of the assessment results will help to reduce the probability of making mistakes in the emergency response after accident. The Gaussian plume model is the most widely used computational model for atmospheric diffusion assessment. Based on this model, the FORTRAN computer language is used to compile Radionuclides Atmosphere Dispersion Codes (RADC). Calculation results based on RADC are compared with HotSpot Health Physics Codes to verify its calculation accuracy. Based on the Bayesian Markov Chain Monte Carlo method, uncertainty of the Gaussian plume model is analysed, and the influence of observation error on the confidence interval is calculated. The results show that the greater the air concentration of radioactivity, the wider the confidence interval; the observation error has a great impact on the confidence interval. Meanwhile, the small observation error will cause a large change in the width of the confidence interval.

Vehicle atmospheric pollution evaluation using AERMOD model at avenue in a Brazilian capital city

One of the most complex environmental problems is the air pollution, and automotive vehicles are one of the main sources of urban air pollution. Aracaju-SE, Sergipe’s capital in the northeast of Brazil, faces frequent congestion in traffic and does not have a monitoring network of air quality, so mathematical models are useful for impact assessment. This work consisted of an unusual application of AERMOD View software for vehicular pollution evaluation at Tancredo Neves Avenue in Aracaju’s city. The modeling was performed for the two avenue tracks, considered as urban linear sources. The rate of emission source was calculated from emission factors, average speed, and the number of vehicles accounted for footage circulating on the promenade at peak times. Concentrations distributions of total suspended particles (TSP), carbon monoxide (CO), and nitrogen oxide (NOx) on the mesh receptors were determinated from weather, topographic, and sources of emission data. The dispersion maps showed that the pollutants were concentrated around the sources; the estimated TSP concentrations were within the standards of CONAMA 491/2018 law. The CO concentration values exceeded the standard due to the high rate of emission sources. NO2 concentrations also exceeded the standard for hourly average, attributed to the contribution of heavy vehicles and the emission rates of light vehicles and motorcycles. The simulations showed that the meteorological and topographical conditions of Aracaju favor the atmospheric pollutants dispersion, that vehicles significantly affect air quality in the region and that the mathematical modeling is a useful tool for the study of atmospheric dispersion.

Atmospheric-dispersion database system that can immediately provide calculation results for various source term and meteorological conditions

ABSTRACTWe have estimated source term and analyzed processes of atmospheric dispersion against atmospheric discharge of radioactive materials due to the Fukushima Daiichi Nuclear Power Station (FDNPS) accident by atmospheric-dispersion calculation using the Worldwide version of System for Environmental Emergency Dose Information (WSPEEDI). On the basis of this experience, we developed an atmospheric-dispersion calculation method that can respond to various needs for dispersion prediction in a nuclear emergency and provide useful information for emergency-response planning. By this method, if a release point, such as a nuclear facility, is known, it is possible to immediately obtain the prediction results by applying provided source term (released radionuclides, release rate, and release period) to the database of dispersion-calculation results prepared in advance without specifying source term. With this function, it is easy to compare results by applying many kinds of source term with monitoring data, and to find out the optimum source term. By preparing a database by this calculation with past long-term meteorological data, we can immediately get dispersion-calculation results for various source term and meteorological conditions. This database is useful for pre-accident planning, such as optimization of a monitoring plan and understanding of events to be supposed in considering emergency countermeasures.

Health effects of PM2.5 emissions from on-road vehicles during weekdays and weekends in Beijing, China

Mobile source emissions have significantly contributed to fine particulate matter (PM2.5) pollution in urban atmospheric environments. Few studies have explored on-road vehicular PM2.5 emissions and the health effects attributed to these emissions under different traffic conditions. Based on driving data obtained from 20000 taxi receipts, a motor vehicle emission simulator (MOVES) model was used to estimate the PM2.5 emission levels of motor vehicles in the urban area within the Sixth Ring Road of Beijing (SRRB) on weekdays and weekends, respectively. Two different PM2.5 exposure scenarios were further simulated using an atmospheric dispersion model. Subsequently, the health effects attributable to traffic-related PM2.5 exposure were quantified by using exposure-response function to calculate the population acute morbidity and premature mortality during different time periods. We found that PM2.5 emission levels of motor vehicles on normal weekdays were overall higher than those on weekends. The median vehicular PM2.5 dispersion concentration in the study area was 2.68 μg/m3 on weekdays and 1.82 μg/m3 on weekends. Annually, there were 4435 premature deaths attributed to vehicle emissions (95% confidence interval (CI): 3655, 4904) under weekday exposure conditions, and this number sharply decreased to 3462 (95% CI: 3052, 4011) on weekends. Considering the hourly measurements, road traffic emissions have the greatest impact on public health during morning rush hour (8:00 a.m.). Total PM2.5 emissions were closely associated with road traffic conditions and largely determined the magnitude of the health impacts caused by traffic-related PM2.5 exposure. These findings provide information to aid in formulating reasonable public health policies to address vehicular PM2.5 emission-induced health implications.

Volcanic ash forecast using ensemble-based data assimilation: an ensemble transform Kalman filter coupled with the FALL3D-7.2 model (ETKF–FALL3D version 1.0)

Abstract. Quantitative volcanic ash cloud forecasts are prone to uncertainties coming from the source term quantification (e.g., the eruption strength or vertical distribution of the emitted particles), with consequent implications for an operational ash impact assessment. We present an ensemble-based data assimilation and forecast system for volcanic ash dispersal and deposition aimed at reducing uncertainties related to eruption source parameters. The FALL3D atmospheric dispersal model is coupled with the ensemble transform Kalman filter (ETKF) data assimilation technique by combining ash mass loading observations with ash dispersal simulations in order to obtain a better joint estimation of the 3-D ash concentration and source parameters. The ETKF–FALL3D data assimilation system is evaluated by performing observing system simulation experiments (OSSEs) in which synthetic observations of fine ash mass loadings are assimilated. The evaluation of the ETKF–FALL3D system, considering reference states of steady and time-varying eruption source parameters, shows that the assimilation process gives both better estimations of ash concentration and time-dependent optimized values of eruption source parameters. The joint estimation of concentrations and source parameters leads to a better analysis and forecast of the 3-D ash concentrations. The results show the potential of the methodology to improve volcanic ash cloud forecasts in operational contexts.

Identifying patterns and sources of anthropogenic trace metals in the Argentine Central Andes by using snow samples and an atmospheric dispersion model.

This study presents the first local measurements of metals in snow from the Argentine Central Andes. Cu, Pb and Zn were selected as specific tracers of anthropogenic local emission sources in the study area. Snow samples were collected during winter (2014, 2015 and 2016) at two sites with different characteristics: Punta de Vacas and Vallecitos. The samples were analyzed by ICP-MS, and the average concentrations found were 1.4 μg L-1 (Cu), 2.2 μg L-1 (Pb) and 14.2 μg L-1 (Zn) in the snow samples from Vallecitos, and 5.3 μg L-1 (Cu), 2.6 μg L-1 (Pb) and 24.0 μg L-1 (Zn) in the snow samples from Punta de Vacas. These data and the snowfall amount in each sampling site were statistically analyzed. Results suggested that the levels of metals in the snow are influenced not only by the atmospheric concentrations, but also by the amount of precipitated snow. Additionally, an atmospheric dispersion model, CALPUFF, was applied in the study area to determine the origin of the studied metals and to understand the spatial distribution pattern of the concentrations found. Results showed that the metal atmospheric loads in the study area originated from local sources, especially vehicular traffic and mining. This work represents an important contribution to the knowledge and preservation of the local Andean cryosphere, and it could be used as an input to develop protection policies in the area through the combination of different and complementary tools for the evaluation of air quality.

Operator splitting method for numerical solving the atmospheric pollutant dispersion problem

The paper discusses the numerical modeling of transport and diffusion of air pollutants in the atmospheric boundary layer. There was developed a mathematical model of industrial emissions spread in the atmosphere taking into account the deposition velocity of fine particles. The model is described by multidimensional partial differential equations with appropriate initial and boundary conditions. The basic laws of hydrothermodynamics were used in deriving the model. In order to obtain the numerical solution of the problem, we used one of the splitting methods according to physical processes involved (transport, diffusion and absorption), as well as a second-order implicit finite-difference scheme in time. Analysis of numerical results showed that the developed computational algorithm provides sufficient accuracy of the problem solution compared with field measurement data and it has a certain advantage over other numerical methods. In the course of computational experiments, there was determined the degree of influence of such parameters as wind speed and direction, absorption coefficient and physicomechanical properties of particles on the process of atmospheric air pollutants dispersion.

Numerical simulation on aerosol pool scrubbing

In the event of severe accident for nuclear power plant (NPP), radioactive aerosols may be released into spent fuel pool (SFP) through filtered-containment-venting-system (FCVS), which are entrained by mixing gas from containment, in order to prevent the overpressure of containment. The efficiency of aerosol pool scrubbing determines the radioactive threats of spent fuel compartment induced by containment venting, and the probable amount of radioactive substance during atmospheric dispersion later on. Therefore, it is necessary to study the typical flow phenomena during the process of aerosol pool scrubbing in the SFP of PWR nuclear power plants, and figure out the important regularities and mechanisms, which can provide reference for evaluations of radioactive threats of spent fuel compartment and provide technical supports for new type of designs for SFP and venting system. In this paper, Fluent is used to establish two kinds of numerical models of SFP, including horizontal injection and vertically downward injection configurations, according to the geometrical dimensions of Qinshan 2 NPP’s SFP. TiO2 is used as the substitute of radioactive substance and coupled numerical models of VOF and DPM are introduced for qualitative and quantitative studies on the effects of diameter of aerosol particle, injection velocity, initial water depth and injection direction. The decontamination factor (DF) is determined by quantifying the mass concentration of aerosol particles which escape and are traced from the SFP under different conditions, respectively. Based on the simulation results, it can be seen that with the increase of particle diameter, DF is going to decrease at first and then increase. There should be a negative correlation between injection velocity and DF. By contrast, there is a positive association between initial water depth and DF. Besides, DFs under vertically downward injection conditions are much lower than that under horizontal injection, due to the appearance of contra-vortex flow adjacent to the free water surface.

Enhancing the Data Learning With Physical Knowledge in Fine-Grained Air Pollution Inference

Fine-grained air pollution monitoring has attracted increasing attention worldwide. Even with an increasing amount of both static and mobile sensing systems, an inference algorithm is still essential to achieve a comprehensive understanding of the urban atmospheric environment. Conventional physical model-based methods are unable to involve all the influencing factors with limited prior knowledge, and data-driven methods lacking physical interpretation may result in bad generalization ability. This paper presents a multi-task learning scheme, which combines the physical model and the data-driven model with both merits. It enhances the data learning of a neural network with the aid of prior knowledge on atmospheric dispersion, and also controls the impact of the knowledge with a tunable weighting coefficient. Evaluations over a real-world deployment in Foshan, China show that, with the resolution of 500m $\times 500\text{m}\times 15$ min, the proposed method outperforms the state-of-the-art ones with 7.9% error reduction and 6.2% correlation increase. Benefited from the physical knowledge, the neural network obtains stable performance with lower variance, as well as higher robustness against negative background conditions.

Modelling the occupational exposure of workers to certain hazardous chemicals

Hazardous substances are considered to be liquids, gases or solids which are of risk for the health and safety of workers and they may be found in almost all workplaces, including in SME’s. Hazardous substances include chemical agents, but also biological ones, such as bacteria, viruses, mould etc. Most chemicals used in the industry present risk for the workers, depending on their physico-chemical properties. Directive 2012/18/EU is the legal act which regulates the chemical substances field within the Member States, regulation dealing with the control of hazards involving dangerous substances which may lead to major accidents. In Romania, Law 59 dated April 2016 on controlling the hazard of major accidents caused by hazardous substances evaluates “hazardous chemicals”, describing the risk for people and regulating concentrations of such substances. By using the commercially available Phast consequence modelling package, within this study were modelled leaks of several chemical substances used in the industry, in order to estimate their hazardous influence extent. The current article is a significant work on modelling of discharge and atmospheric dispersion of hazardous substances using state-of-the-art consequence modelling software. Emergency Response Planning Guidelines (ERPG) are used as reference exposure levels within the present study. Output data of computational modelling are significantly influenced by input parameters. In this regard, the effects of the latter for ensuring robustness of the simulation and for identifying improvements have proven to be necessary.