Study On Characteristics Of Ambient Air Pollution And Forecast Model In Nanning

In the process of analyzing and predicting dispersion changes of pollutants in city air, it is necessary to consider the local features. Mastery of the rules of pollution makes successful study on air quality forecast and pollution control. In international and domestic researches, statistical and numerical predictions are two main forecast methods.Analysis of the data on five pollutants (SO₂, NO₂, PM₁₀, CO and O₃) concentrations in Nanning from 2001 to 2005 showed that annual average of SO₂ was close to or over the standard limit, exceeding rate of daily average of PM₁₀ topped the others. Except that hourly concentration curve of O3 showed one-peak feature, the others showed two-peak. Concentration variations of pollutants were mainly influenced by emissions and atmospheric dispersion conditions. The analysis showed that SO₂ was mainly from point-source emission. Line-source emission had notable influence on NO₂ concentration. The source of PM₁₀ was more complicated. Under the influence of meteorological conditions, CO concentration was relatively higher during autumn and winter, and PM₁₀ concentration ascended in summer and autumn.Correlation analysis on pollutants concentration and meteorological elements indicated that pressure and temperature showed significant correlations with most of pollutants. Wind speed had weak influence on concentration variations of pollutants. Relative humidity, dew point, vapor pressure and rainfall were not correlative to pollutants concentration. Study on typical pollution cases revealed that light polluted days amounted to 68 in five years and mostly occurred in Jan., Oct., Dec, Nov. during autumn and winter. Dominant weather system was a ridge of high ground pressure and upper-air temperature inversion.Threshold auto-regressive statistical method is selected to conduct basic forecast. In this paper, three types of optimizing are tried on the basic model. Comparisons of the predicting and monitoring results show that using daily pressure as participant meteorological element, 360-day as sample length, logarithm as alternation of data make better forecasts. Under the meteorological and special pollution circumstances which model calculations fails to provide accuracy forecast, amendment forecast will be announced through consultation approach.Based on the above, considering the need of daily business activities in monitoring departments, the software of "Nanning City Air Quality Forecast Management System" is developed . It has been demonstrated an effective and good practice.