Forecast Model Of Atmospheric Particulates And Its Simulation Of Spatial Distribution During Heating Period

With the rapid development of economy and the increase in material life day by day, both preserving ecological environment and improving the environmental quality have become the urgent needs of the whole society. Atmospheric particulates suspending in the atmosphere usually reduce the atmospheric visibility through extinction effect and even result in the serious air pollution-haze; epidemiological studies have indicated that atmospheric particulates have obviously harmful effects on human health and they can cause the extensive damages that often occur in respiratory system, cardiovascular system and reproductive system. Lanzhou is one of the most serious air-polluted cities in China even all over the world. In Lanzhou, particulate matter (PM) is primary atmospheric pollutant, especially during the heating period, the air pollution of particulate matter is very serious because of the sharp increase in the amount of coal and the special meteorological conditions, for example, the high frequency of inversion and calm wind. It is very important to understanding the concentration levels, the forecast model and the spatial distribution of particulate matter in Lanzhou city during the heating period.The diurnal variation of PM2.5and its levels during heating period from2007to2009in Lanzhou were understood by using statistic analysis methods. The forecast model of PM2.5was established with the meteorological data by using multiple stepwise regression analysis method and BP artificial neural network method respectively. In the end, concentration distribution of PM10in Baiyin city was predicted by using the WRF-AERMOD coupled system. The main results are shown as follows:(1) The air pollution of PM2.5was extremely serious. The exceeding standard rates of PM2.5in2007,2008and2009are59.2%,67.9%and68.8%respectively, the maximum exceed-limit multiples are2.88,3.17and3.60respectively. The curve of diurnal variation of PM2.5shows2peak values and2valley values. The concentrations of PM2.5in floating dust, sand storm and ash haze weather were much higher than the average value and the concentration of PM2.5in ash haze weather had the highest value.(2) The daily average concentration of PM2.5was negatively correlated with the daily average relative humidity and the relationship of the daily average concentration of PM2.5and the daily average wind speed could be stated with a quadratic function. The daily average concentration of PM2.5was uncorrelated with the daily mean temperature.(3) Compared with the forecast model which was established by multiple stepwise regression analysis method, that which was established by BP artificial neural network can simulate the change trend of PM2.5better. In term of the forecast-concentrations of PM2.5in special weather and forecast accuracy, the forecast model which was established by BP artificial neural network is the superior one.(4) AERMOD requires steady and horizontally homogeneous hourly surface and upper air meteorological observations. However, observations with such frequency are not available for AERMOD in Baiyin city. To overcome this limitation, we have developed a preprocessor for offline coupling of WRF with AERMOD, the planetary boundary layer and surface layer parameters required by AERMOD were computed using the Weather Research and Forecasting (WRF) Model.The results indicate that the coupled system of WRF-AERMOD can well simulate the concentration distribution of PM10. the simulated linear regression as against measured values yielded a degree of fitting of0.706with a slope of1.22(n=60, p<0.001), the mean bias error of3.91%, the root mean squared error of13.6%, the index of agreement of79.5%and the model efficiency of66.2%.