嵌套网格空气质量预报模式系统的发展与应用

主要综述中国科学院大气物理研究所自主开发的嵌套网格空气质量预报模式系统(NAQPMS,Nested Air Quality Prediction Modeling System)的历史发展与应用情况.模式发展伊始为欧拉污染物输送实用模型,利用其研究东亚硫氧化物的跨国输送问题,得出中国对于周边国家的输送量不大的结论; 在系统中嵌入适合东亚的起沙机制模块,用来模拟沙尘发生、输送及沉降等过程,估算亚洲大陆沙尘气溶胶对海洋地区的输送与沉降通量,为研究海洋生物地球化学循环提供基础数据; 利用该系统研究沙尘及其土壤粒子对酸雨的中和作用,发现沙尘输送对东亚酸雨的分布影响很大; 发展城市尺度高分辨率气象和空气质量预报技术,使模式水平分辨率达到500 m,并应用于台北高浓度臭氧和PM10的模拟; 研究和集成区域及城市尺度大气污染预报理论和模拟技术,研制成目前的嵌套网格空气质量预报模式系统,以探讨不同尺度各种污染(如沙尘暴、城市光化学烟雾、酸雨、高浓度悬浮颗粒物等)的变化规律.在模式系统中初步建立资料同化模块,开展大气化学成分及沙尘输送模拟的资料同化研究.系统已经在北京、上海、深圳、郑州等城市环境监测中心实施空气质量的实时预报.未来,系统将集成到全球环境大气输送模式(GEATM),以实现从城市群到全球具有双向耦合功能的模式系统.     

沙尘发生期遥感反演地表温度与实测沙尘数据之间的比较分析

以2002年3月份沙尘天气为例,用卫星遥感反演地表温度(LST)时间变化趋势与同一地区野外观测站点接收的沙尘干物质含量/人体可吸人物质含量(TSP/PM10)数据的时间变化趋势进行对比分析,两者有较好的对应关系.卫星观测的时间分辨率高、范围广,所提供的地表相关信息可为沙尘源头的治理及减轻或减少沙尘天气发生提供基础数据,也为沙尘暴预测提供了重要的科学依据.     

沙尘天气定量分级方法研究与应用

利用已有的关于能见度与沙尘浓度统计反演关系的研究成果,对沙尘天气进行了定量分级研究.利用2004、2005年春季3～5月沙尘天气地面气象观测资料,采用沙尘天气的强度分类与反演的沙尘浓度分级的统计方法,建立了不同沙尘天气(扬沙、沙尘暴和强沙尘暴)对应的沙尘浓度等级,并通过对2006年沙尘天气的实例分析,验证了该种分级方法的可用性.通过由能见度反演的观测结果与沙尘数值预报模式的对比分析,探讨了沙尘数值预报业务模式产品的天气学释用方法,建立了不同强度沙尘天气与沙尘数值模式输出的浓度之间的定量分级关系.     

影响三大洲的北非强沙尘暴动力机制WRF-chem模拟与分析北大核心

基于WRF-chem模式对北非2018年3月下旬的典型强沙尘暴过程进行模拟,分析了此次强沙尘发生季节、持续时间、局地特征以及传输路径的关键动力系统与动力机制。鉴于起沙是沙尘暴发生的关键点之一,并且起沙主要取决于风力和下垫面沙源性质,本文测试了三种起沙参数化方案的影响,并将模拟结果与卫星MODIS监测及其再分析资料MERRA-2进行了对比,又经系列统计方法检验。结果显示,宏观思路的起沙方案GOCART比AFWA和UoC两种起沙方案更适合此次大尺度强沙尘暴数值模拟(锋面跨度接近60个经度)。综合沙尘暴关键系统的动力机制分析和数值模拟结果显示,强沙尘暴关键系统为深厚的西风槽、沙尘冷锋锋面和锋后的地面高压反气旋。北非中部深厚的西风槽为后倾槽,该系统稳定,造成沙尘暴持续时间长。沙尘暴锋后反气旋中的下沉气流抑制了扬沙向高层扩散,造成低层能见度恶劣。沙尘锋区结合了动力、热动力以及湿热动力不稳定,因此锋区风力大,地面沙尘驱动力强。而西风槽和强大反气旋依托环流形势,提供了沙尘传输到三大洲的长途输送力。     

Model-Experimental Analysis of Variational Assimilation of Dust Concentration with GRAPES Sand-Dust Numerical Forecasting System in a Sand-Dust Storm

为了提高沙尘模式的预报准确率,通过在区域天气数值模式GRAPEs（G10bal／RegionalAs-similationandPrEdictionSystem）的三维变分同化系统中增加沙尘浓度这一控制变量的方法建立了GRAPES-3DVAR-DusT沙尘同化系统。利用中国北方8个观测站提供的沙尘PM10数据和沙尘模式（GRAPEs-CUAcE／Dust系统）提供的背景场,对2008年2月29日-3月1日发生在中国北方的一次沙尘暴天气进行了控制试验和一次同化与间断同化的敏感性试验,结果表明：（1）引入该同化系统后,一次同化和间断同化试验模拟的地面沙尘浓度分布较未同化的控制试验结果更接近卫星监测,而间断同化的结果又好于一次同化;（2）一次同化试验与控制试验对单站PM10浓度的演变预报较差;（3）间断同化试验较准确地再现了单站PM10浓度的连续演变;（4）间断同化试验效果整体上优于一次同化试验。总体而言,引入沙尘同化系统在一定程度上可以提高沙尘模式对沙尘天气的预报准确率。     

气象预报模式参数化方案对重污染过程PM2.5浓度预报效果的影响

针对北京市2016年12月16～21日的重污染过程,基于嵌套网格空气质量模式预报系统(NAQPMS),面向气象驱动模式WRF中7类物理过程的参数化方案,通过单扰动和组合扰动方式构建了51组不同的WRF模式运行配置,对比分析不同方案配置下NAQPMS对这次重污染过程细颗粒物(PM2.5)浓度预报的性能.结果表明:在重污染...     

春季沙尘天气对内蒙古相关城市颗粒物污染影响研究

沙尘天气是造成我国北方春季区域性沙尘型重污染的主要原因,然而目前对此研究并不多见。因此,本文利用中国环保网2014年1月1日-2016年12月31日内蒙古11个城市环境监测站的颗粒物浓度的逐日和逐时资料,首先分析近3年该地区颗粒物污染浓度的年变化特征,然后对比这3年沙尘天气发生的次数及时段,探究颗粒物污染的年变化特征及其与沙尘天气之关系。统计结果表明,近3年春季内蒙古沙尘天气的发生次数是逐年增加的,中西部是沙尘天气频发区,与之相对应,西部颗粒物浓度的年变化高于东部,且造成内蒙古主要城市PM10浓度春季出现全年的最高值,表明沙尘天气频繁发生对当地粗颗粒物污染有显著的影响。对比内蒙古全年3个时间段的PM10浓度值,其排序是：春季沙尘期间＞春季非沙尘期间＞其他季节;即春季沙尘期间PM10浓度比非沙尘期间高69%,比其他季节高101%。有所不同的是,3个时间段平均PM2.5浓度排序则为：春季沙尘期间＞其他季节＞春季非沙尘期间;春季沙尘期间PM2.5的平均浓度比其他季节高16%,比春季非沙尘期间高29%;可见,春季沙尘天气对相关城市PM10浓度的影响明显大于对PM2.5浓度的影响。最后对内蒙古地区典型沙尘暴和扬沙个例进行细致研究, 发现沙尘暴个例中PM10浓度的增加倍数在2.3~15.1之间,而扬沙过程PM10浓度的增加倍数在0.8~5.6之间,两者相比可看出,沙尘暴过程对颗粒物污染的影响显著大于扬沙过程。     

2010—2012年我国西北地区沙尘个例气溶胶特征分析

利用2010—2012年间中国西北地区敦煌、民勤和塔中3个站点的CE-318太阳光度计观测资料,反演获得了气溶胶440 nm波段的大气气溶胶光学厚度（AOD）及440—870 nm波长指数（Alpha）,同时结合Moderate Resolution Imaging Spectroradiometer(MODIS)卫星L1B产品及环境颗粒物监测仪Tapered Element Oscillating Microbalance(TEOM)观测的PM10数据,挑选出2010—2012年间沙尘天气特征明显的6个日期,并对这6天的气溶胶光学特性、PM10浓度变化特征及沙尘气溶胶来源进行了分析。研究结果表明：MODIS卫星图有明显沙尘天气过境时,当天的AOD值较高,Alpha值则较低,且AOD和Alpha表现出相反的变化趋势。这表明在这3个站点沙尘气溶胶占主导,PM10浓度变化与AOD变化趋势有较好的正相关性。Hybrid Single Particle Lagrangian Integrated Trajectory（HYSPLIT）后向轨迹分析表明,气团大多起源于塔克拉玛干沙漠或干旱、半干旱区。     

GRAPES沙尘数值预报系统对一次沙尘暴沙尘浓度同化模拟试验分析

为了提高沙尘模式的预报准确率,通过在区域天气数值模式GRAPES（Global/Regional Assimilation and PrEdiction System）的三维变分同化系统中增加沙尘浓度这一控制变量的方法建立了GRAPES₃DVAR_DUST沙尘同化系统。利用中国北方8个观测站提供的沙尘PM10数据和沙尘模式（GRAPES_CUACE/Dust系统）提供的背景场,对2008年2月29日-3月1日发生在中国北方的一次沙尘暴天气进行了控制试验和一次同化与间断同化的敏感性试验,结果表明:（1）引入该同化系统后,一次同化和间断同化试验模拟的地面沙尘浓度分布较未同化的控制试验结果更接近卫星监测,而间断同化的结果又好于一次同化;（2）一次同化试验与控制试验对单站PM10浓度的演变预报较差;（3）间断同化试验较准确地再现了单站PM10浓度的连续演变;（4）间断同化试验效果整体上优于一次同化试验。总体而言,引入沙尘同化系统在一定程度上可以提高沙尘模式对沙尘天气的预报准确率。     

WRF/Chem模式起沙权重因子对沙尘模拟结果的影响

利用耦合了起沙模块的大气化学全耦合WRF/Chem模式,对2014年4月23—25日中国西北一次典型沙尘天气过程进行模拟,分析Shao2004起沙参数化方案(简称“Shao04方案”)垂直沙尘通量公式中权重因子γ对沙尘时空分布特征的影响,并与气象卫星遥感监测沙尘范围以及站点颗粒物质量浓度进行对比分析,确定了较适用于中国西北地区Shao04方案的权重因子γ的取值。结果表明:(1)γ对沙尘质量浓度模拟范围和质量浓度中心值有影响,对垂直沙尘通量中心值的大小有影响;(2)不同γ取值都能很好地模拟出沙尘天气PM10和PM2. 5质量浓度的趋势,但只有当γ=1时,即耦合Shao2011起沙参数化方案(简称“Shao11方案”)的WRF/Chem模式能够较准确地模拟出中国西北沙尘过程中PM10和PM2. 5质量浓度的变化。     

RegCM-Dust模式对东亚一次沙尘暴天气过程模拟的效果检验

为检验区域气候模式与沙尘模式耦合模式RegCM Dust的性能,以2006年东亚地区一次沙尘暴过程为例,将模拟结果与观测资料进行对比,以检验模式对沙尘天气过程的模拟能力。结果表明：模式对沙尘暴过程的地面风场特征模拟效果较好,总体上重现了大风区的分布;地面沙尘浓度和沙尘光学厚度模拟结果与观测分布总体吻合。模式虽然是区域气候模式与沙尘模式耦合模式,但由于其内核是建立在中尺度数值模式MM4基础上,因此对天气过程尤其是沙尘天气过程具备较好的把握能力,对于沙尘天气过程预测具有良好的应用前景。     

北京地区一次空气重污染过程的目标观测分析

针对北京市2016年12月16~21日的空气重污染过程进行了回报试验,探讨了该次事件预报的目标观测敏感区。使用新一代高分辨率中尺度气象模式（Weather Research Forecasting,WRF）和嵌套网格空气质量模式（Nested Air Quality Prediction Model System,NAQPMS）,针对初始气象场的不确定性,通过4套初始场资料识别了影响北京地区细颗粒物（PM2.5）预报水平的目标观测敏感变量及其敏感区。结果表明:当综合考虑初始气象场的风场、温度、比湿不确定性的影响时,发现改善黑龙江区域上述气象要素的初始场精度,对北京地区PM2.5预报不确定的减小最显著;当分别考察风场、温度、比湿的不确定性的影响时,发现初始风场精度的改善,尤其是黑龙江区域风场精度的改善,能够更大程度地减小北京地区PM2.5的预报误差,对北京东南地区的PM2.5预报误差的减小甚至可达到40%以上。因此,优先对黑龙江区域的气象场,尤其是该区域的风场进行目标观测,并将其同化到预报模式的初始场中,将会有效提高初始气象场的质量,进而大大减小北京地区PM2.5浓度的预报误差,提高北京地区空气质量的预报技巧。初始风场代表了北京地区该次空气重污染事件预报的目标观测变量,而黑龙江地区则是该目标观测的敏感区域。     

Numerical modeling of a strong dust event over the south China region in March 2010

The mesoscale model WRF-Chem was used to simulate a severe dust storm event that occurred in March 2010. The storm affected a vast area of East Asia, including the south China region and Hong Kong. This southern region is rarely affected by dust weather. The performance of the WRF-Chem was evaluated by observational data such as the National Center for Atmospheric Research reanalysis data for atmospheric circulation, PM10 concentration from various ground stations, and satellite images of Moderate Resolution Imaging Spectroradiometer and Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations. The dependence of the model’s performance on certain important parameterizations was examined in this study. For this particular dust storm event, the model results suggest that the simulation is not very sensitive to certain key physical parameterizations such as threshold wind speed of dust emission and the choice of land surface model. In general, the WRF-Chem is capable of capturing the key physical processes for this severe dust event. The analysis of the dust transport fluxes suggests that the dust transport to the south China region is mainly from the north, although there is a mountainous region in the northern part of the south China region.     

An evaluation of air quality modeling over the Pearl River Delta during November 2006

In this paper, we evaluate the performance of several air quality models using the Pearl River Delta (PRD) region, including the Nested Air Quality Prediction Modeling System (NAQPMS), the Community Multiscale Air Quality (CMAQ) model, and the Comprehensive Air Quality Model with extensions (CAMx). All three model runs are based on the same meteorological fields generated by the Fifth-Generation Pennsylvania State University/National Center for Atmospheric Research (PSU/NCAR) Mesoscale Model (MM5) and the same emission inventories. The emission data are processed by the Sparse Matrix Operator Kernel Emissions (SMOKE) model, with the inventories generated from the Transport and Chemical Evolution over the Pacific/Intercontinental Chemical Transport Experiment Phase B (TRACE-P/INTEX-B) and local emission inventory data. The results show that: (1) the meteorological simulation of the MM5 model is reasonable compared with the observations at the regional background and urban stations. (2) The models have different advantages at different stations. The CAMx model has the best performance for SO₂ simulation, with the lowest mean normalized bias (MNB) and mean normalized error (MNE) at most of the Guangzhou stations, while the CMAQ model has the lowest normalized mean square error (NMSE) value for SO₂ simulation at most of the other PRD urban stations. The NAQPMS model has the best performance in the NO₂ simulation at most of the Guangzhou stations. (3) The model performance at the Guangzhou stations is better than that at the other stations, and the emissions may be underestimated in the other PRD cities. (4) The PM₁₀ simulation has the best model measures of FAC2 (fraction of predictions within a factor of two of the observations) (average 53–56%) and NMSE (0.904–1.015), while the SO₂ simulation has the best concentration distribution compared with the observations, according to the quantile–quantile (Q–Q) plots.     

PM10 data assimilation over south Korea to Asian dust forecasting model with the optimal interpolation method

A data assimilation (DA) system using ground PM₁₀ observation for Asian Dust Aerosol Model version 2 (ADAM2), which is the operational dust forecasting model of Korea Meteorological Administration (KMA), has been developed with the optimal interpolation (OI) method. The observations are provided by the PM₁₀ network operated by KMA. Three DA experiments are performed to simulate a dust event observed in Korea from 1 March to 31 May 2009 with different assimilation cycles of 24 (DA24), 12 (DA12), and 06 hours (DA06). 48-hour forecasts from the adjusted Initial Condition (IC) of dust concentration are compared with control simulation (CTL) and observation from independent stations. It is found that CTL simulates spatial patterns of dust emitted and transported associated with a developing low pressure system over the dust source regions quite well, compared with satellite measurement. However, it appears that there is considerable uncertainty in estimating the concentration of dust. With IC adjustment, the model simulates improved dust concentration, showing considerably reduced RMSE, particularly for the prediction within 12 hours of forecast. At the same time, it is shown that the time interval of DA affects the predictability of ADAM2, so that DA06 appears to have better predictability within a 12-hour simulation, reducing RMSE by 50% compared with CTL. This suggests that assimilating PM₁₀ to the dust prediction model using OI has the potential to predict air quality in Korea when the cycle of assimilation is sufficiently short.     

Numerical simulations of Asian dust events: A Lagrangian Dust Model and its applications

An uni-modal Lagrangian Dust Model (LDM) was developed to simulate the dust concentrations and source-receptor (SR) relationships for recent Asian dust events that occurred over the Korean Peninsula. The following dust sources were used for the S-R calculation in this study: S-I) Gurbantunggut desert, S-II) Taklamakan desert, S-III) Tibetan Plateau, S-IV) Mu Us Desert, S-V) Manchuria, and S-VI) Nei Mongol and Gobi Desert. The following two 8-day dust simulation periods were selected for two case studies: (Period A) March 15–22, 2011, and (Period B) April 27–May 4, 2011. During two periods there were highly dense dust onsets observed over a wide area in Korea. Meteorological fields were generated using the WRF (Weather Research and Forecasting) meteorological model, and Lagrangian turbulent properties and dust emission were estimated using FLEXPART model and ADAM2 (Asian Dust Aerosol Model 2), respectively. The simulated dust concentrations are compared with point measurements and Eulerian model outputs. Statistical techniques were also employed to determine the accuracy and uncertainty associated with the model results. The results showed that the LDM compared favorably well with observations for some sites; however, for most sites the model overestimated the observations. Analysis of S-R relationships showed that 38–50% of dust particles originated from Nei Mongol and the Gobi Desert, and 16–25% of dust particles originated from Manchuria, accounting for most of the dust particles in Korea. Because there is no nudging or other artificial forcing included in the LDM, higher error indicators (e.g., root mean square error, absolute gross error) were found for some sites. However, the LDM was able to satisfactorily simulate the maximum timing and starting time of dust events for most sites. Compared with the Eulerian model, ADAM2, the results of LDM found pattern correlations (PCs) equal to 0.78-0.83 and indices of agreement (IOAs) greater than 0.6, suggesting that LDM is capable of estimation of dust concentrations with the quantitative information on the S-R relationships that can be easily obtained by LDM.     

A simulation of Asian dust events in March 2010 by using the ADAM2 model

Asian dust events occurred in Asia during March 2010 were simulated using the Asian Dust Aerosol Model 2 (ADAM2). The performance of the model for simulations of surface dust concentrations and dust event occurrences was tested at several monitoring sites located in the dust source region and the downstream region of Korea. The observed and modeled dust event occurrences at each monitoring site were defined with the hourly observed and modeled dust concentrations that were used to evaluate the performance of the model by constructing a contingency table for the dust event occurrence. It was found that the model simulated quite well the starting and ending times of dust events with their peak dust concentrations for most dust events occurred both in the dust source region and the downstream region of Korea. However, the model failed to simulate a few dust events observed in both regions mainly due to the inaccurate simulations of the meteorological fields. Inaccurate simulations of wind speeds have caused for the model to simulate dust events poorly in the dust source region whereas poor simulations of precipitation of the fifth-generation mesoscale model (MM5) model have led to miss dust events in the downstream region of Korea. The contingency table made with the hourly data for the dust event occurrence made it possible to evaluate the ADAM2 model for the simulation of the dust event occurrence. It was found that the model has the probabilistic simulation capability for dust events of about 78% with the hit rate of more than 83% and the false alarm rate of about 27% for the dust events occurred during March in 2010. The probabilistic capability of the model could be much improved by improving the meteorological model (MM5 model).     

通用地球系统模式对亚洲夏季风降水的模拟能力评估

通过与观测/再分析资料和参加第五次耦合模式比较计划(CMIP5)的模式模拟结果进行对比,评估了通用地球系统模式(CESM,1.0.3版本)对亚洲夏季风降水的模拟能力。结果表明:CESM能够合理地模拟出亚洲夏季风降水的气候平均态,但与其他CMIP5模式模拟结果类似,对中国东南地区降水模拟偏少,而对中国西部高原地区降水模拟偏多;CESM可以再现亚洲季风区降水冬弱夏强、雨带北进南退的季节变化特征,其模拟偏差具有区域性和季节性差异;从EOF分析结果来看,CESM能够模拟出亚洲夏季风降水的时空变化特征,且能较好地抓住亚洲夏季风降水与厄尔尼诺-南方涛动(El Ni?o-Southern Oscillation,简称ENSO)的相关关系。总的说来,CESM对亚洲夏季风降水的模拟是合理的,模拟水平与4个最好的CMIP5模式相当。