基于环流分型法的地面臭氧预测模型

Study on circulation classification based surface ozone concentration prediction model

利用2011~2016年地面臭氧观测数据和同期地面气象要素观测及大尺度再分析资料,选取日最大8h臭氧浓度指标,分析臭氧浓度,局地气象要素以及大尺度环流因子的关系,并将Lamb-Jenkinson客观环流分析方法与逐步回归模型结合,建立臭氧浓度逐日预报模型.结果表明,杭州地面臭氧浓度呈季节性变化特征,春夏两季杭州臭氧平均浓度较高,为臭氧超标易发时段,其中5月臭氧浓度超标频次最高.地面臭氧浓度受局地气象要素影响显著,其中总辐射和日最高气温与臭氧浓度呈显著正相关,相对湿度和降水则呈负相关.在客观分型得到的10种环流型中,杭州全年受反气旋环流控制的概率最高,占26.5%,受西北气流环流控制的概率最低,仅占0.6%.在南风型环流形势下,杭州臭氧浓度超标频率最高,达23.8%,北风型环流形势下的臭氧浓度超标频率最低,为3.7%.基于季节环流分型的地面臭氧预报模型对预报效果改进明显,2016年模型预报值与观测臭氧浓度值相关系数达到0.87.模型尤其提高了高浓度臭氧事件预报准确性,2016年共24次臭氧超标事件,模型成功预报15次,TS评分达到52%.

In this study,a new surface ozone concentration level simulation model was developed by combining the Lamb-Jenkinson objective circulation classification method and the stepwise linear regression model,and successfully applied in Hangzhou.The influence of local meteorological factors and large-scale circulation factors,which are from large scale reanalysis data between2011and2016,to the ozone concentration level are analysed first.Our results showed a strong seasonal trend of the surface ozone concentration in Hangzhou.Ozone concentration level was much higher during spring and summer,and frequently exceeded the national standard which caused ozone pollution events,especially in May.Local meteorological conditions had a significant impact on the ozone concentration level.Both solar radiation and daily maximum temperature are positively correlated to ozone concentration level,but the relative humidity and precipitation are negatively correlated to it.In addition,we also classified the local cyclonic circulation observations using Lamb-Jenkinson objective circulation classification method,and identified that different cyclonic circulation types had different impact to the ozone concentration.The anti-cyclonic circulation was the dominant type and accounted for26.5%of all the cyclonic circulation in Hangzhou,while the circulation from northwest had the least frequency of0.6%.Ozone pollution events happened most frequently(23.8%)under the south airflow controlled circulation condition,but had the lowest frequency under north airflow controlled circulation condition(3.7%).Our results proved that the performance of the ozone concentration level simulation model will improve greatly by considering the seasonal circulation types and the meteorological factors using the stepwise regression method,which could increase the correlation between model prediction and observations up to0.87,and our new model also improves the simulation of the extreme high ozone concentration pollution events significantly,it successfully predicted15ozone pollution events out of24events that happened in2016,with a high TS score of52%.