Atmospheric Environment (1967)最新文献

Factor analysis receptor models attempt to estimate both the source composition and the source intensity from a series of observations. The factor analysis solution resulting from Principle Component Analysis (PCA) has no real physically interpretable meaning. Only an appropriate transformation enables a realistic interpretation. Any realistic transformation solution must obey certain natural and physical constraints, such as non-negative source elemental composition and non-negative source intensity, which are not explicitly examined in the existing receptor models. If these natural constraints are violated the results will be uninterpretable.

All observed data sets contain more or less information about the sources. This paper presents a receptor model, which extracts source information from the observed data set to deduce the source profiles, and respects the important natural constraints. This receptor model was tested with a simulated test data set, which was generated with the source profiles and intensities used in the Quail Roost II Workshop. It has also been applied to an ambient data set sampled in Berlin (West) during January and February 1984.

Runoff water samples were taken at the St. Rombout's Cathedral (Mechelen, Belgium), which was constructed with sandy limestones of Balegem and Gobertingen. Gypsum appeared to be the principal deterioration compound. The mean annual surface recession from the cathedral was calculated to be around 20 μm. Yearly several tons of stone material are flushed away from the cathedral with the rain water.

Eight short-term long-range transport models (MESOPUFF, MESOPLUME, MSPUFF, MESOPUFF II, MTDDIS, ARRPA, RADM and RTM-II) have been tested with field data from two data bases involving tracer releases. The Oklahoma data base involved two separate experiments with measurements taken at 100 and 600 km arcs downwind of a 3-h perfluorocarbon release. The Savannah River Plant data base encompassed 15 experiments with measurements taken over 2–5 days at distances of 28–144 km downwind from a 62 m stack release of Kr-85 gas.

Application of the American Meteorological Society statistics to the model/data comparisons showed that six of the eight models predicted within a factor of two of the observed concentrations for all of the following: points paired in space and time, points paired in space only, and for points unpaired in space and time. However, the ratio of the standard deviation of residuals to the average observed value showed improvement as more unpairing was done in the comparison of the models with the data. The statistical comparisons reveal a definite tendency of the models to overpredict plume concentrations. Supplemental graphical comparisons showed that plume concentration overprediction is accompanied with an underprediction of plume spreading, and that a definite time lag is often observed between the time of arrival of the observed plume and the time of arrival of the predicted plume.

The causes of model/data discrepancies can be largely traced to inadequate wind field modeling that leads to an incorrect temporal and spatial positioning of the plume, and the use of the Turner [Workbook of atmospheric dispersion estimates. U.S. Dept of H.E.W. Publication 999-AP-26 (1970)] curves to downwind distances beyond which they can accurately represent the scales of atmospheric turbulence. The use of multilayer wind field models and the use of the Heffter [J. appl. Met.4, 153–156 (1965)] formula for lateral plume dispersion close to the source appear to improve model accuracies.

The present paper deals with numerical effects occurring in the application of the implicit (“backward Euler”) method to solve the diffusion equation in the case of a point source (i.e. singular initial data). The numerical over-estimation of the concentration at the source level as well as conditions for an over- or under-estimation of the ground-level concentration are investigated. To improve the results, a specific filtering of the initial concentration distribution is suggested. All theoretical results are illustrated by numerical examples; for this, an approach of constructing analytical ‘reference’ solutions, for special profiles of the diffusion coefficient, is presented.

A Lagrangian statistical (Monte Carlo) model for airborne pollutant dispersion is presented. Its ability to simulate the atmospheric dispersion both in homogeneous and inhomogeneous turbulence by comparison with an analytical solution and with the Willis and Deardorff water tank experiments, respectively, has been stated in previous papers. In the present paper the model is used to simulate dispersion in the real atmospheric PBL. The numerical results obtained are verified against experimental data from the Karlsruhe Nuclear Research Center tracer experiments. The model is applied to the problem of predicting the ground level concentration of two different tracers simultaneously released from two heights (160 and 195 m) at the Karlsruhe meteorological tower. Convectively unstable and neutral conditions were prevailing during the two tracer experiments which have been simulated. Model performance was evaluated through two statistical indexes: relative mean bias and normalized mean square error. The cumulative frequency distribution of the point-by-point ratio between observed and predicted ground level concentrations (glcs) was also computed. The simulated concentrations agree very well with observations. The tracer data were also compared to the simulations of 10 Gaussian models. They differed one another for the choice of dispersion sigma curves and for the way to insert the wind speed and direction. None of them proved to perform better than our particle model in all the exercises.

The number distribution as a function of sodium mass has been determined for individual particles in ambient air at an urban location about 40 km inland from the Pacific ocean. The measurements were made in real-time by single particle mass spectrometry. The aerosol is ‘externally mixed’ with respect to sodium, that is, a small fraction of the total number of particles are highly enriched in Na. Size distributions calculated for Na mass equivalent spherical NaCl particles over the size range 0.1–0.4 μm show a consistent peak at 0.19 μm. Samples taken after foggy mornings had number densities for Na containing particles around 9 cm⁻³ compared with 0.9 cm⁻³ for samples taken after clear mornings. Comparison of the size distribution over this size range with marine particle size distributions indicates that sub-μm Na containing particles at the inland sampling site are of marine origin. Sodium mass density for particles smaller than 0.4 μm was found to be ∼ 0.02 μg m⁻³ and for particles of diameter ⩽2.5 μm, it was estimated to be ∼ 1.4 μg m⁻³.