中文摘要：

基于中尺度气象模式（MM5）、排放源模式（sMOKE）和大气化学模式（CMAQ），耦合本地排放源清单，建立了珠三角区域空气质量数值模式预报系统。该套模式系统对清洁、污染过程的一次、二次污染物都有较好的预报能力。冬季，能见度和PM。。的预报值和观测值的相关系数最高达0．76和0．78。O3和NOx的预报值和观测值的相关系数为0．64和0．61。随着预报时效的增加，预报效果没有明显降低。夏季，在2个典型光化学过程中，O3的预报值和观测值的相关系数分别为0．68和0．81，NO3分别为0．57和0．64。在相关的边界层气象因子中，地表通风系数和能见度相关度最高，相关系数达0．71。混合层高度、地面风速、比湿和能见度的相关系数分别为0．55、0．50、和-0．37。在灰霾过程中，PM25占PM10的质量浓度为78％。在气溶胶质量权重中，硫酸盐所占的比重最高，占PM25权重达到33％，元素碳为18％，有机碳为14％，铵盐为5％，硝酸盐仅为1％。灰霾过程的二次气溶胶的质量权重比清洁过程的大。

英文摘要：

Air quality （photochemical smog and haze） numerical model forecast system was established by coupling MMS, SMOKE and CMAQ models which used the sources emission inventories of Pearl River Delta. The study showed that the system has a good forecasting ability for a variety of pollutants on different pollution processes. In winter, the correlation coefficients of visibility was up to 0.76 between forecasted and observed values, while PM25, 03 and NOx was 0.78, 0.64 and 0.61. As the forecasting time increased, the accuracy of forecast did not significantly reduce. In summer, the correlation coefficients of Oa were up to 0.68 and 0.81 between forecasted and observed values in two photochemical processes, while NOx was 0.57 and 0.64. Of the related boundary layer meteorological factors, surface ventilation coefficient had the highest correlation of 0.71 with visibility, while mixing layer height, surface wind speed and humidity were 0.55, 0.50 and-0.37 respectively. In haze process, PM25 contributed 78% of PM10 mass concentration. Of all the aerosol mass, the sulfate contributed the highest fraction （33%） of PM25 concentration, while black carbon, organic carbon, ammonium and nitrate contributed 18%, 14%, 5% and 1% respectively. The mass weight of secondary aerosol was more than those generated in clear process.