This study investigated the impact of gaseous precursor emission reduction over key source regions on inorganic aerosols(IAs)and PM2.5 in Tianjin during a typical regional haze process with the NAQPMS(Nested Air Quality Prediction Modeling System).Based on high-precision simulations,the high IA concentration originating from the North China Plain(NCP)was transported to East China and subsequently returned to the NCP region in a clockwise pattern,resulting in two pollution periods in Tianjin.The online pollutant source-tagging method coupled with the NAQPMS was used to quantify the contribution of different source regions to IAs in Tianjin,and the NCP region was identified as the key source area,with a daily contribution of 57.6%-100%.Sensitivity experiments were performed to analyze the impact of reducing precursor emissions by 30%in the NCP region a day before a pollution event and during the pollution days in Tianjin.NH3 control notably reduced IA and PM2.5 concentrations in Tianjin by 30.8%and 13.3%,respectively,which were 16 and 26.6 times that of SO2 reduction and 7 and 6.4 times that of NOx reduction,respectively.SO2 reduction increased nitrate concentration by 3.5%in Tianjin because SO2 control increased gaseous NH3,and excessive NH3 in the atmosphere can neutralize HNO3 to nitrate production based on the thermodynamic effect.NOx reduction exhibited positive and negative effects on nitrate in different areas of the NCP,and transport of the two effects occurring in upstream areas caused the concentrations of nitrate,IAs,and PM2.5 in Tianjin to decrease first and then increase.This study highlights that transporting positive effects in key source areas can contribute to PM2.5 reduction while transporting negative effects harms the joint prevention and control of air pollution.
Hazardous/lethal compound temperature-humidity heatwaves with a wet bulb temperature(i.e.,≥33℃/35℃)can severely affect human health.The middle and lower reaches of the Yangtze River always experience high-frequency compound temperature-humidity heatwaves.This study investigates the population exposed to these heatwaves in the middle and lower reaches of the Yangtze River for the near-term(2021-2040),medium-term(2041-2060),and long-term(2081-2100)periods using the five-climate model outputs under seven SSPs(Shared Socioeconomic Pathways)—based scenarios(i.e.,SSP1-1.9,SSP1-2.6,SSP4-3.4,SSP2-4.5,SSP4-6.0,SSP3-7.0,and SSP5-8.5).These scenarios are obtained from the CMIP6(Coupled Model Intercomparison Project phase 6)in combination with the demographic characteristics under the Shared Socioeconomic Pathways(i.e.,SSP1-5).The results show that during the baseline period of 1995-2014,hazardous/lethal compound temperature-humidity heatwaves occurred at frequencies of approximately 6 d and 3 d in the middle and lower reaches of the Yangtze River,where the longest duration was approximately 10 d and 4 d,respectively.In the future,the frequency and duration of such heatwaves are projected to increase in the long-term period,where the frequency is expected to be approximately 12-39 d and 7-24 d,and the longest duration is as long as 30 d and 14 d,respectively.In the baseline period,the hazardous/lethal compound temperature-humidity heatwaves affected an area of approximately 74.8×104 and 22.3×104 km2 in the middle and lower reaches of the Yangtze River,exposing the populations of 170 million and 20 million people of the hazardous/lethal compound temperature-humidity heatwaves,respectively.The impact range and maximum exposed population in the 21st century were observed in the long-term period,accounting for approximately 83%-100%and 32%-98%of the study area,respectively.The exposed population was approximately 1.2-2.5 and 2.5-20.5 times higher than that during the base period in the middle and lower reaches of the Yangtze River,respectively.The population exposed to the lethal compound temperature-humidity heatwaves increased considerably by approximately 40 million-370 million.Spatially,the main areas affected by these heatwaves in the 21st century are Shanghai,northern Zhejiang,southern Jiangsu,southern Anhui,eastern Hunan,and eastern Jiangxi.Overall,forecasting,early warning,and risk prevention of lethal compound temperature-humidity heatwaves must be urgently improved.
Through the measurement results of the Beijing Nanjiao radiosonde in January-October 2021 and measurement data of the automatic station of the Alpine Skiing Venue in Yanqing during the Winter Olympic Games,the accuracies of temperature and humidity profiles retrieved by a ground-based MicroWave Radiometer(MWR)were assessed.The results revealed that the MWR-retrieved temperature profile exhibited a good correlation with the observations from the radiosonde and the automatic station,with relatively small errors and good reliability.However,the MWR-retrieved humidity profile exhibited poor correlation with the observations from the radiosonde and the automatic station(correlation coefficient=0.81).Comparison of the results from the MWR and those of different automatic stations and the radiosonde revealed that at levels of 0.15-0.6 km,the MWR-retrieved temperature profile exhibited a good correlation with the observations from the radiosonde and the automatic station.Furthermore,the Root-Mean-Square Error(RMSE)and Average Deviation(AD)of the MWR-retrieved temperature profile concerning the observation from the automatic station increased with increasing height and nevertheless those of the MWR-retrieved temperature profile with respect to the observation from the radiosonde decreased with increasing height at levels of 0.15-0.6 km.The correlation,RMSE,and AD of the MWR-retrieved relative humidity profile with respect to the observation from the radiosonde and the automatic station increased with increasing height.Based on the comparison between the MWR and the radiosonde,the authors observed a significant positive correlation between the MWR-retrieved temperature and radiosonde observations over the entire sounding height,which is higher in the lower atmosphere than in the upper atmosphere.In contrast,the correlation of the MWR-retrieved humidity profile with the observation from the radiosonde was significantly lower than that of the temperature profile and exhibited a negative correlation at levels of 2.75-3.5 km.The Mean Error(ME)of the MWR-retrieved temperature profile at each sounding height,except at 10 km,was within 2℃.Meanwhile,the RMSE and AD of the MWR-retrieved temperature profile were about 3.4℃ and 2.5℃,respectively,at levels of＜3 km.Furthermore,the RMSE and AD values of all other levels increased with increasing height.The RMSE and AD of the MWR-retrieved humidity profile were expectedly higher than those of the temperature profile at all levels.Meanwhile,the ME of the MWR-retrieved humidity profile was large at most levels,with a maximum of 23.67%.Precipitation caused the error in the MWR-retrieved temperature profile to increase at the most levels for the duration of 0800 LST and 2000 LST,in which the RMSE and AD under precipitation were expectedly higher than those under the no-rain condition at levels＞0.5 km.However,the RMSE and AD of the MWR-retrieved humidity profile during 0800 LST and 2000 LST in rainy days(at most levels in the lower atmosphere)were expectedly lower than those during clear days,in which the RMSE and AD at 2000 LST degraded dramatically.
Based on the snow depth dataset over China and CN05-gridded precipitation data over China,the relationship between the snow depth in the Tibetan Plateau(TP)and summer precipitation in Yunnan is investigated through singular vector decomposition and correlation analysis.Results show that the positive snow depth anomalies in the central and western TP during winter and spring can enhance the summer precipitation in Yunnan,particularly in the Jinsha River basin and the southwest of Yunnan,and the correlations between the snow depth in the TP and the summer precipitation in Yunnan may be independent of the influence of the El Niño-Southern Oscillation.The possible impact mechanism has been investigated through diagnostic analyses using the fifth generation ECMWF reanalysis datasets(ERA5).The extreme snow depth in the key region of the TP leads to a low surface air temperature in the central and western parts of the TP and nearby areas in spring.This is conducive to the late onset of the South Asian summer monsoon and leads to a weak South Asian summer monsoon and the associated monsoon depression,along with the abnormal westerly wind to the south of the TP.Moreover,the cold surface temperature associated with the extreme snow depth in the TP can initiate the wave train at 200 hPa,which propagates from the western part of the TP through Mongolia to Northeast Asia along the westerly jet stream.Furthermore,an anomalous cyclonic circulation can be observed in Northeast Asia,which is conducive to the southward movement of cold air in the middle and high latitudes,leading to increased rainfall in Yunnan.Meanwhile,a wave train at 850 hPa,which spreads from the southwest side of the plateau to the South China Sea,can be observed,leading to an anomalous anticyclonic circulation in the South China Sea.An anomalous low-level shear over Yunnan develops due to the westerly wind on the southern side of the plateau and the southwesterly wind on the northwest side of the anticyclonic circulation in the South China Sea,which is favorable for increased precipitation in Yunnan.Meanwhile,cold air flows southward to Yunnan and converges with warm and humid air,which also contributes to the increased summer precipitation in Yunnan.
Based on the use of multiple linear regression models,the air quality of Jiyuan City,a typical industrial city,was analyzed at different stages from October 2021 to March 2022.The impact characteristics of the change of meteorological factors on PM2.5,PM2.5 pollution level,and the difference in the growth rate of pollutant concentration before and after the start of winter heating(15 Nov 2021)were studied to explore the characteristics of PM2.5 pollution in cities along Taihang Mountain in autumn and winter.Results reveal that in the first stage(1 Oct-14 Nov 2021),26.1%of PM2.5 hourly concentration change was determined using meteorological factors,and the correlation between any single factor and PM2.5 was＜36%.In the second stage(15 Nov-31 Dec 2021),72.4%of the hourly concentration change of PM2.5 was determined using meteorological factors.Wind direction,relative humidity,and visibility had considerable influences on PM2.5,and the correlation between PM2.5 and relative humidity and visibility was the highest(61.5%correlation with relative humidity and 73.1%correlation with visibility).In the third stage(1-31 Jan 2022),53.2%of hourly concentration change of PM2.5 was determined using meteorological factors,and the relative humidity and wind speed had no considerable effect on the hourly change of PM2.5,which is associated with the considerable effect of long-range migration and detention of pollution clusters in this stage.In the fourth stage(1 Feb-31 Mar 2022),32.2%of PM2.5 hourly concentration change was determined using meteorological factors.Wind speed had no considerable effect on PM2.5 hourly change because it was affected by sand and dust.During the pollution period in autumn and winter in Jiyuan City,the main components of particulate matter were NO3-,NH4+,OC,and SO42-,with the proportion of secondary inorganic ions(SO42-,NO3-,and NH4+)being＞65.7%,and the secondary pollution was severe.Thus,the growth rates of particle component concentrations show that,with the increase in pollution,the growth rates of NO3-,S,EC,and Cl-decrease while those of SO42-,OC,K+,and NH4+ exhibit a"slow-fast"trend.
To investigate the effect of Volatile Organic Compounds(VOCs)on ozone(O3)formation during summertime—which is conducive to O3 pollution—the chemical composition characteristics of VOCs and their sources were studied using high-resolution online monitoring data obtained in an urban site of Hohhot during the summer of 2021.Furthermore,the sensitivity of O3 pollution days and the control strategy of its precursors were further analyzed using an observation-based model(OBM).Results revealed that the averaged total mixing ratio of VOCs was 21.10±9.38 ppb(1 ppb=10-9),with Oxygenated Volatile Organic Compounds(OVOCs)being the most abundant group(36.3%),followed by alkanes(23.8%),halogenated hydrocarbons(16.8%),alkynes(10.4%),aromatic hydrocarbons(6.6%),and alkenes(6.1%).According to the Positive Matrix Factorization(PMF)source analysis,the primary sources of VOCs in Hohhot are diesel tail-gas sources,gasoline tail-gas sources,solvent sources,natural gas and combustion sources,biological emission sources,and liquefied petroleum gas sources,with contribution rates of 19.8%,18.2%,17.6%,16.3%,15.4%,and 12.7%,respectively.According to the Relative Incremental Reactivity(RIR)and Empirical Kinetic Modeling Approach(EKMA)analysis,O3 sensitivity was in the VOCs-limited regime during the O3 pollution days in Hohhot,with greater RIR values from alkenes and aromatic hydrocarbons.Simulating precursor reduction scenarios from the various VOCs sources resolved by PMF revealed that reducing VOCs from motor vehicle-related sources is most beneficial to the control of O3 during summertime.
The Arctic-boreal zone north of 50°N is one of the two major fire zones in the world.Fires in this region affect the local vegetation succession as well as the regional and global carbon cycle and climate.Previous studies have focused on fires in small regions or over a specific land cover type and individual extreme fire events;however,the spatiotemporal variability of the Arctic-boreal fires remains unclear.In this study,we comprehensively analyzed the spatiotemporal variability of the burned area in the Arctic-boreal zone from 2001 to 2016 based on three satellite-based global fire products:GFED4.1s,MODIS C6,and FireCCI51.Results show that during 2001-2016,the average burned area in this zone was 7.47±0.72 Mha/a,and large values for the burned area are mainly observed in locations including Alaska,central Canada,and south and central-east Siberia.The burned area exhibits large interannual variability,with similar magnitude and spatial pattern to the multiyear average.The annual burned area for forests demonstrates an upward trend in the entire Arctic-boreal region and Arctic-boreal North America,while that for croplands exhibits a downward trend in Arctic-boreal Europe.Arctic-boreal fires mainly occur during spring and summer seasons.However,the primary land cover types that burned are different.On a multiyear average,fires occurred mainly in savannas and forests in summer over the Arctic-boreal North America,in croplands in spring and forests in summers over the Arctic-boreal Europe,as well as in forests and shrublands in summers and croplands and forests in spring in the Arctic-boreal Asia.In years where the burned area is extremely large,the land cover types that burned in various regions are similar to the multiyear average results.
Based on statistical analysis and multiple sensitivity experiments using the Weather Research and Forecasting model,it is found that the relationship between preceding soil moisture anomalies and summer heat waves in North China is affected by the strength of the West Pacific Subtropical High(WPSH).When the WPSH is strong,the southerly winds on its west side carry a large amount of water vapor from the tropics to the southern parts of North China and increase the precipitation in the region,which is not conducive to maintain the preceding soil moisture dry anomaly,thereby restricting the contribution of the preceding soil moisture anomaly to the summer heat waves.In contrast,when the WPSH is weak,the soil moisture dry anomalies in North China can last for a longer duration and lead to heat waves.The WPSH intensity is associated with the Sea Surface Temperature(SST)in the tropical central-eastern Pacific.When the tropical Pacific SST exhibits positive anomaly in summer,the WPSH is relatively weak,and the precipitation in North China is low,which is conducive for maintaining dry soil conditions in North China.Under such scenarios,the preceding soil moisture anomaly can be employed as the prediction signal of heat waves in North China.
The Noah-MP(Noah land surface model with Multi-Parameterizations)provides numerous complex parameterization schemes for land surface physical processes.However,a Noah-MP LSM(Land Surface Mode)simulation is often limited by a large calculational requirement and an insufficient computational ability.Therefore,determining an applicable parameterization scheme through full combination scheme experiment is difficult.To scientifically reduce the number of experiments,an orthogonal test method was introduced in this study,and five primary factors affecting surface temperature simulations were selected—dynamic vegetation,canopy stomatal resistance,soil moisture factor for stomatal resistance,surface layer drag coefficient,and radiative transfer.To determine an applicable combination of parameterization schemes for studying parametric scheme optimization,nine experiments were designed using the CLDAS-V2.0(China Meteorological Administration Land Surface Data Assimilation System)to drive the Noah-MP LSM for surface temperature simulation in Southeast China.Results show that the choice of a parameterization scheme had a substantial impact on the simulation of a woodland area as well as the simulation in July and August.The sensitivity of the land surface physical process and the optimal parameterization scheme were simultaneously affected by the underlying surface and season.In most cases,dynamic vegetation and canopy stomatal resistance substantially affected the simulation and were relatively sensitive physical factors.Through a spatiotemporal analysis of the optimal combinations of parameterization schemes for different underlying surfaces in different seasons,a simulation verification showed that the optimal combination for surface temperature simulation in southeast China was the following:Dynamic vegetation,Ball-Berry-type canopy stomatal resistance,Noah-type soil moisture factor,M-O-type surface heat exchange coefficient,and GAPFVEG type radiative transfer schemes.
利用浙江省地面观测数据和新一代静止气象卫星数据,通过逻辑回归(LR)、线性判别(LDA)、K近邻算法(KNN)、决策树(CART)、高斯贝叶斯(NB)和支持向量机(SVM)6种机器学习算法针对浙江省金丽温高速公路进行低能见度识别建模,并运用多种评估方法评估模型结果,显示SVM算法模型效果较好,且针对小于1000 m的能见度天气有较好的识别.进一步结合地面观测数据和卫星数据建立识别模型,发现效果优于单一来源的数据建模,一般以KNN算法建模效果较好,且在对浓雾、强浓雾的识别中,结合地面和卫星数据的模型识别效果更好.针对单一数据利用SVM算法,结合地面和卫星数据选择KNN算法再对金丽温高速公路的大雾过程进行识别,显示新一代静止气象卫星数据的模拟效果不差于地面观测数据模拟效果,且能够识别夜间和凌晨的雾,对地面观测识别可作为有效补充,将对省内没有地面气象观测的低能见度识别和短临预测有一定辅助参考作用.
采用加拿大环境部研发的RHtests均一化系统,结合台站详细的历史沿革信息,对1951～2019年黑龙江省83个气象台站逐日平均、最高和最低气温进行了均一性检验和订正,与均一化逐日气温数据集(CHTM)进行了对比研制,重点评估了最近10年观测仪器换型对资料均一性的影响;基于均一化气温日值数据,统计了黑龙江省极端气温指数:持续冷日日数(CSDI)、霜冻日数(FD)、冰冻日数(ID)和气温日较差(DTR).结果表明:1951～2019年,日平均气温、最高气温和最低气温分别存在40个、20个和57个断点,台站迁移、仪器变化和自动观测业务软件升级是造成黑龙江省气温序列非均一的主要原因,订正后的气温序列空间一致性更高,平均气温和最低气温的变化趋势分别由0.27℃/10 a和0.25℃/10 a上升为0.29℃/10 a和0.27℃/10 a.黑龙江省CSDI、FD和ID等极端气温指数均呈明显下降趋势,1998年中国平均气温出现历史上第二次最热记录,黑龙江省FD指数的最低值也出现在1998年.黑龙江省平均最高气温升温趋势比平均最低气温升温趋势略小,造成DTR呈现下降趋势.
生物质燃烧向大气中排放大量痕量气体和颗粒物,源排放清单是深入研究生物质燃烧环境气候效应的重要基础数据.利用全球火排放数据库GFED(Global Fire Emissions Database)、NCAR全球火排放清单FINN(Fire INventory from NCAR)和中国露天生物质燃烧排放清单 MEIC(Multi-resolution Emission Inventory for China),对2008～2017年中国地区生物质燃烧源排放的空间分布、季节和年际变化特征以及不同清单间的异同进行分析研究.3个清单都显示生物质燃烧释放的黑碳(BC)、有机碳(OC)、空气动力学粒径小于2.5 μm的颗粒物(PM2.5)和一氧化碳(CO)在中国东北、长江和黄河下游之间地区和中国南方的排放量较高,与我国的主要农作物产地和森林地区分布一致.FINN清单排放量在华南地区与西南地区比其他两个清单高,而GFED清单排放量在长三角地区比其他两个清单排放量高.中国地区平均生物质燃烧排放量在春季出现峰值,而在不同的生物质燃烧地区峰值出现的季节不同,与各地农作物播种、收获时节和农耕习惯不同有关.2008～2017年,中国地区年平均生物质燃烧排放量的峰值主要出现在2014年,但各地区峰值出现的年份明显不同,东北、华中/东、华南和西南地区分别在2015年、2013年、2008年和2010年排放量达到最大.对于BC、OC和PM2.5,GFED和MEIC清单中的排放量比较接近,而FINN中的排放量是GFED和MEIC中的2～3倍;3个清单中CO的排放量比较接近.2014年生物质燃烧源排放与人为源排放的对比分析表明,所有物种中,生物质燃烧排放的OC和PM2.5相对于人为源排放量占比最大,3个清单中占比分别为9％～24％和5％～16％,说明生物质燃烧排放的OC和一次PM2.5是中国气溶胶的重要来源.
基于局地气候区分类,选取了北京门头沟地区4种局地气候区(高层开阔、高层密集、中层密集、稀疏建筑)作为研究对象,利用高分辨率大涡模拟方法,数值研究温度层结效应对不同局地气候区风和湍流特性的影响.2019年11月7日晴天小风个例模拟结果表明:1)温度层结对湍涡的形状和范围有显著影响.在近地面水平剖面上,稳定层结下涡旋数量较中性层结情况减少,但涡旋的纵向延伸范围可增大67％;不稳定层结下涡旋数量较中性层结增加,涡旋的纵向范围可缩小60％.在垂直剖面上,相较于中性层结,稳定层结下环流结构减弱且涡旋的纵向范围可缩小40％,不稳定层结下环流结构增强且涡旋的纵向范围可增大20％,该现象在高层密集型地块最为明显.2)4种局地气候区的风速的高值区主要位于平行于盛行风方向的建筑物两侧及屋顶附近,热力作用对总风速有增益作用,近地面风速较入流风速可增加1.27～2.18倍.3)4种局地气候区湍动能的高值区主要位于建筑物底部拐角处和屋顶,不稳定层结下近地面的湍动能是中性层结的1.2～1.5倍,而稳定层结下是中性层结的0.5～0.8倍,即不稳定层结条件下浮力引起的热力湍流增强混合效率,而稳定层结条件下湍流运动受到抑制.4)相较于其他局地气候区,高层密集区域的建筑物底部风速较大,在不稳定层结下易形成较强的狭管效应,其街区峡谷最大风速是中层密集的1.5倍.
采用WRF模式模拟了云南省哀牢山区域2020年6月13～14日一次降水过程.通过不同高度的地形敏感性试验对比分析,讨论了哀牢山地形对强降水时空分布的影响及可能的物理机制.研究结果发现:1)不同高度的地形敏感性试验表明,地形高度对低涡切变线的位置有影响.2)地形升高后,中低层的假相当位温线更密集且梯度较大,水汽与不稳定能量迅速堆积,伴随的强上升运动可能会提前触发强对流天气;地形高度降低后,则假相当位温线平直且疏散;哀牢山局地抬升作用与不稳定能量较小,且不足以触发中小尺度强对流天气.3)在WSM6微物理方案下,地形高度的变化亦对云微物理过程有明显的影响.地形升高后强迫抬升作用加强,使中高层的冰晶与雪混合物在空中停留的时间更长而扩展范围逐步增大;从而产生次级环流的下沉气流,中低层云水和雨滴碰并增强,造成云水混合比减小而雨水混合比增加.
基于 ECMWF(European Centre for Medium-Range Weather Forecasts)ERA5(Reanalysis version 5)、CFSR(Climate Forecast System Reanalysis)、MERRA2(Modern-Era Retrospective analysis for Research and Applications version2)3种再分析数据,13个海洋观测站和3个测风塔的观测数据,研究了中国近海风资源时空特征,并讨论了3种再分析数据在中国近海风资源评估中的适用性.结果表明,再分析数据在中国东海和南海的弱风频率比渤海、黄海高,且ERA5在所有海域小于6m/s的弱风累积概率比CFSR(MERRA2)高39.0％(44.9％)、43.6％(47.5％)、60.7％(41.6％)和 47.9％(38.2％).ERA5、CFSR 和 MERRA2 在中国近海的有效风时空间分布相似,量级都介于84％～95％;3种再分析平均风能密度自北向南呈"低高低"空间分布,其中台湾海峡是WPD大值中心(超过4000W/m2).风能稳定性方面,ERA5和CFSR的日变异呈"南弱北强"特征,而MERRA2日变异系数介于1.03～4.适用性分析表明,ERA5整体性能优于CFSR和MERRA2,但MERRA2在再现渤海、南海的风能日波动,CFSR在刻画黄海的有效风时、风能密度和东海、黄海的变异系数时具有一定优势.由此说明不同再分析数据对中国近海风资源不同指标的适用性各有优劣,应根据需要及数据条件,针对不同海域采用不同类的再分析数据开展风资源评估研究及工作.
基于1979～2022年NCEP/DOE逐月再分析资料和NOAA海面温度资料,通过合成、相关分析等统计方法,以海面温度(Sea Surface Temperature,SST)异常影响为切入点,对比分析了 2022年与La Niña年强迫作用下的我国夏季季节内环流差异,并在此基础上进一步探讨了异常高温与同期热带SST之间存在的可能联系.结果表明:1)2022年夏季,我国中东部高温区具有明显空间变化特征,6月位于华中地区、7月位于西南地区、8月影响整个长江流域.2)西太平洋副热带高压和南亚高压的同时异常加强,以及两者重叠打通并形成少见的北半球副热带高压带,是造成2022年夏季我国中东部异常高温过程的直接原因.3)在持续2年的较强La Niña背景下,2022年东亚夏季环流并未完全表现出对La Niña冷SST的响应,南海及菲律宾以东对流异常偏弱、东亚—太平洋遥相关型不显著均与La Niña年环流典型特征有较大出入.夏季同期热带SST异常对高温过程的形成具有一定贡献,热带西印度洋和热带中太平洋的冷SST异常分别有利于我国长江流域和华中地区出现高温酷暑,其中热带中太平洋冷SST异常可能是西太平洋副热带高压加强的重要原因,而热带西印度洋冷SST对南亚高压的增强有贡献作用.
以大气垂直累积液态水含量的预报问题为例,使用UNet网络结构作为基础结构构建时空预报模型,对比了采用两类预报策略的模型的预报效果,预报策略包含一个迭代预报策略(Recursive Forecast Strategy,RFS)以及两个直接预报策略(Direct Forecast Strategies,DFSs).研究结果表明,两个直接预报模型对整体预报时段的预报效果明显优于迭代预报模型,直接预报模型的RMSE比迭代预报模型低19％.随着预报时次的增加,迭代预报模型的预报误差累积速度比两个直接预报模型快.在两个直接预报模型中,多时次输出模型(Direct Forecast Model Multi-Steps,DFS-M)的预报表现更加稳健,在整体预报时段上预报效果优于单时次输出模型(Direct Forecast Model Single-Step,DFS-S),但DFS-S模型对几个前期时次的预报效果较好.本研究利用深度学习可解释性技术中的深度学习重要特征分析方法(Deep Learning Important FeaTures,DeepLIFT)分析DFS-M和DFS-S模型各个输入时次对于模型预报的相对重要性.研究结果表明,DFS-M和DFS-S模型80％的输入重要性都集中在最后两个输入时次上,较早期输入时次的重要性随着预报时次的增加而呈现上升趋势.由于各输出时次间存在一定的统计相关性,受输出时次相关性约束的DFS-M模型的输入时次重要性变化比DFS-S模型更加稳定.通过将DFS-M和DFS-S模型对于不同时次的预报进行结合,可以得到效果更加均衡的预报.本研究可以为基于深度学习的天气气候预报方法的选择提供新的思路.
云南省地处低纬高原地区,毗邻东南亚,空气污染物质除受本地排放外也受东南亚地区跨境输送的影响.本文收集整理了 2017～2021年云南省16个州市40个国控站观测数据,分析云南省污染特征和变化趋势;并利用后向轨迹(HYSPLIT)聚类分析和浓度权重轨迹(CWT)方法分析了 PM2.5的主要区域来源.结果表明,云南省PM2.5年均浓度呈下降趋势,下降率为0.91±0.23μgm-3a-1.在季节变化上,春季浓度最高,全省平均为31.92±9.08 μgm-3,夏季浓度最低,为13.50±2.69 pg m-3.春季东南亚地区生物质燃烧导致云南省PM2.5的污染最严重,最大贡献值超过40 µg m-3,此外广西西南部也是云南省春季高潜在源区之一.在日变化上,PM2.5浓度呈现"双峰型"特征,最大值出现在09:00(北京时间,下同)至12:00和21:00至01:00,主要是人为活动与较低的边界层高度和风速等气象因素共同作用的结果.研究显示,云南省春季PM2.5浓度多源于跨境输送,这将为云南省空气污染物治理提供新的启示.
利用ENVISAT卫星搭载的迈克尔逊干涉仪和Aqua卫星搭载的AIRS探测仪观测到的大气NH3浓度数据以及全球大气化学—气候模式EMAC模拟的NH3浓度结果,分析了 2008～2011年6～9月亚洲地区大气NH3的空间分布特征.结果显示,夏季时近地面NH3浓度最高值出现在印度北部,同时紧邻印度北部的孟加拉湾存在深对流,凭借青藏高原的高海拔地势,此深对流可以将寿命较短的NH3输送到上对流层和下平流层(Upper Troposphere and Lower Stratosphere,UTLS),所以在青藏高原上空出现了 NH3的向上输送柱,即青藏高原是NH3向上输送的主要通道.亚洲夏季风反气旋的位置主导着NH3在UTLS区域的空间分布,反气旋内持续存在NH3高浓度中心,NH3高浓度中心位置与反气旋中心位置对应良好,会出现一个或两个NH3高浓度中心,说明反气旋内环流形式的变化对反气旋内NH3分布特征有重要影响.
近十多年来云南地区多次发生极端的季节性连续干旱,尤其是夏季、秋季的连续干旱会加重传统冬、春季节干旱的危害,给当地经济和社会活动造成了严重的影响.本文基于1970～2019年云南省气象局120站降水数据及NCEP海平面气压场数据,从降水持续性异常指数出发,分析了云南夏、秋季节连续干旱事件发生的演变特征,进而探究了该区域夏秋连续干旱事件与前期4月海平面气压的可能联系,以期能够在极端干旱的预报中提供帮助.研究结果表明:(1)云南夏秋降水持续性异常的主要特征为全省一致的偏多或偏少;(2)云南夏秋连续干旱事件发生前期表现为3个区域海平面气压异常的组合,分别为北大西洋、北印度洋和中太平洋海平面气压异常;(3)由前期4月上述关键信号构建的组合信号指数,能够较好地表征云南夏季、秋季的降水连续异常,是具有意义的潜在前期预报信号.
