基于深度学习模型的广州市大气PM2.5和PM10浓度预测

精准预测大气污染颗粒物PM_2.5、PM₁₀浓度能为大气污染防治提供科学依据,但目前较多PM_2.5和PM₁₀浓度预测在缺少污染源排放清单和能见度数据时,预测精度不高。而目前深度学习模型应用于PM_2.5和PM₁₀浓度预测的研究还鲜见报道。基于广州市2015年6月1日—2018年1月10日的空气质量和气象监测历史数据,分别构建了随机森林模型(RF)、XGBoost模型2种传统的机器学习模型和长短时记忆网络(LSTM)、门控循环单元网络(GRU)2种深度学习模型,并对广州市的PM_2.5、PM₁₀日均浓度值进行预测。结果表明:在缺少污染源排放清单和能见度数据时,4种模型也能较好地预测PM_2.5、PM₁₀日均浓度。根据MSE、RMSE、MAPE、MAE和R2等评价指标,对4个模型的PM_2.5、PM₁₀预测效果进行测评,得出深度学习GRU模型预测效果均为最佳,RF模型的预测结果均为最差。相比目前研究及应用较多的RF模型、XGBoost模型、LSTM模型,基于深度学习的GRU模型能更好地预测PM_2.5、PM₁₀浓度。     

用固相萃取-高效液相色谱-三重四极杆串联质谱法测定海洋沉积物中13种氟喹诺酮类抗生素

为进一步探索海洋沉积物中氟喹诺酮类抗生素的污染情况,基于固相萃取-高效液相色谱-三重四极杆串联质谱(SPE-LC-MS/MS)技术,建立了海洋沉积物中 13 种FQs的测定方法;采用高效液相色谱-三重四极杆串联质谱多反应监测离子模式(MRM)对FQs进行分离检测。结果表明：在优化实验条件下,13种FQs的质量浓度为0.50~100 μg·L⁻¹,目标化合物峰面积与内标物质峰面积之比与质量浓度的线性关系良好(R²>0.99),方法检出限为0.003~0.03 μg·kg⁻¹;在加标量为1 μg·kg⁻¹和10 μg·kg⁻¹时,空白加标的平均回收率为73.5%~124.6%和67.5%~118.5%,相对标准偏差(RSD)为1.0%~9.7%(n=7);以海洋沉积物为基质,13种目标物的加标回收率为67.7%~142.4%,RSD小于10.2%(n=6);使用该方法对广州某湾区海洋沉积物中 13 种 FQs 的残留量进行了实地检测,培氟沙星质量分数最高,为1.6 μg·kg⁻¹,氧氟沙星、环丙沙星和恩诺沙星质量分数次之,为0.7 μg·kg⁻¹。该方法实现了对海洋沉积物中 13种 FQs 的同时检测,具有快速、准确等优点,适用于海洋沉积物中13种FQs的测定。本研究成果可为海洋生态环境保护提供数据基础及技术支撑。     

臭氧-生物活性炭净水工艺处理苯酚污染的应急效果及影响因素分析

考虑到突发性水污染中有毒有机物的高频现率,选用苯酚为有毒有机物代表,通过模拟突发苯酚污染的情况,研究了臭氧-生物活性炭(O3-BAC)工艺对苯酚污染的应急处理效果,并且分析了工程中实际环境因素如空床接触时间(EBCT)、pH值对该应急能力的影响.研究表明,在突发性苯酚污染的应急处理中,0.6 mg/L的臭氧可将初始苯酚浓度5倍于达标浓度的进水处理至饮用水达标浓度,即低于0.002 mg/L,15.3 min的空床接触时间是判定其是否为该应急过程的关键因素的分界点,为了取得较好的应急效果,可适当提高进水pH值.     

The sensitivity of the CHIMERE model to emissions reduction scenarios on air quality in Northern Italy

The sensitivity of the CHIMERE model to emission reduction scenarios on particulate matter PM2.5 and ozone (O₃) in Northern Italy is studied. The emissions of NO_x, PM2.5 SO₂, VOC or NH₃ were reduced by 50% for different source sectors for the Lombardy region, together with 5 additional scenarios to estimate the effect of local measures on improving the air quality for the Po valley area. Firstly, we evaluate the model performance by comparing calculated surface aerosol concentrations for the standard case (no emission reductions) with observations for January and June 2005. Calculated monthly mean PM10 concentrations are in general underestimated. For June, modelled PM10 concentrations slightly overestimate the measurements. Calculated monthly mean SO₄, NO₃^?, NH₄⁺ concentrations are in good agreement with the observations for January and June. Secondly, the model sensitivity of emission reduction scenarios on PM2.5 and O₃ calculated concentrations for the Po valley area is evaluated. The most effective scenarios to abate PM2.5 concentration are based on the SNAP2 (non-industrial combustion plants) and SNAP7 (road traffic) sectors, for which the NO_x and PM2.5 emissions are reduced by 50%. The number of days that the 2015 PM2.5 limit value of 25 μg m^?3 in Milan is exceeded by reducing primary PM2.5 and NO_x emissions for SNAP2 and 7 by 50%, does not change in January when compared to the standard case for the Milan area. It appears that 40% of the PM2.5 concentration in the greater Milan area is caused by the emissions surrounding the Lombardy region and from the model boundary conditions.This study also showed that a more effective pollutant reduction (emissions) per ton of pollutant reduced (concentrations) for the greater Milan area is obtained by reducing the primary PM2.5 emissions for SNAP7 by 50%. The most effective scenario on PM2.5 decrease for which precursor emissions are reduced is achieved by reducing SO₂ emissions by 50% for SNAP7.Our study showed that during summer time, the largest reductions in O₃ concentrations are achieved for SNAP7 emission reductions, when volatile organic compounds (VOCs) are reduced by 50%.     

Quality considerations of European PM emission inventories

This paper reviews information on emission inventories of particulate matter (PM) in Europe. A large body of scientific literature is available to cover many different aspects. Studies focus on specific sources or source sectors (road transport as well as off-road machinery, domestic heating, industry, agriculture, and natural sources), among which especially road transport emissions are clearly best established. Emission inventories comprising all sources are available for specific European regions, often pointing out regional differences, but also for the entire continent. Still these inventories often are not able to satisfy the needs. Due to PM specific circumstances such as the large number of sources, very different release pathways and differences of the individual particles in terms of chemical composition or size, it is very difficult to appropriately handle measurement conditions to arrive at adequate emission factors, especially when emission points cannot be defined clearly. Information on fugitive emissions (caused by wind shear, material transfer processes or other mechanical forces from non-point sources) is sparse, except for road traffic emissions where recent data seems to converge. The problem of data gaps concerns activities in industry (quarries), agriculture, but also natural particles like sea salt and wind-blown dust. Comparing complete inventories to independent efforts in assessing emissions, e.g. atmospheric measurements combined with source apportionment, allows to better understand and quantify the reliability of inventory data. Methodological improvements and harmonization currently under way in Europe will focus efforts and allow for more reliable PM inventories in the near future.     

湿法烟气脱硫系统的脱汞性能现场实验

分析湿法烟气脱硫系统的脱汞性能,对控制燃煤电厂的汞污染具有重要意义。利用安大略水法和吸附管法分别对某600 MW电厂湿法脱硫系统的进出口的烟气进行了采样,测量了烟气中各形态汞浓度,并分析了该系统对烟气总汞、气态氧化态汞的脱除效果以及对气态单质汞的影响。研究结果表明,安大略水法和吸附管法均能较为准确地测定湿法脱硫系统进出口烟气中的汞含量,测得入口和出口的氧化汞与平均值的相对误差的绝对值分别为3.5%和1.3%;入口和出口的单质汞与平均值相对误差的绝对值分别为16.6%和3.3%。其中吸附管法操作相对简单。通过湿法烟气脱硫系统后,烟气中氧化态汞的浓度可下降87.5%,其中约67.5%的氧化态汞被湿法脱硫系统脱除,约20%的氧化态汞在脱硫浆液的还原作用下被还原为单质汞,导致脱硫系统出口的单质汞浓度高于入口。     

基于抗侧刚度比的防屈曲支撑组合框架优化研究

针对某26层设置了防屈曲支撑的钢管混凝土组合框架的benchmark模型,以振型分解反应谱法作为基础算法,编制了基于遗传算法的优化程序,通过改变结构各层的抗侧刚度比,对设置防屈曲支撑的钢管混凝土组合框架结构进行了防屈曲支撑截面面积的优化设计,对以变化抗侧刚度比设计的结构和以固定抗侧刚度比设计的结构进行了弹性阶段和弹塑性阶段的地震响应对比分析。结果表明:以变化抗侧刚度比设计的结构能在不减弱抗震性能的同时,减少48%的防屈曲支撑用量。在罕遇地震作用下,以变化抗侧刚度比设计的防屈曲支撑的耗能比例达54.98%,充分发挥防屈曲支撑的塑性变形能力消耗地震能量,起到保护框架梁、柱的作用。     

填海工程对海水入侵影响的数值模拟研究

在系统分析研究区地质及水文地质条件的基础上,运用地下水模型FEFLOW建立了研究区垂向二维非均质流体的对流-弥散数学模型,并通过水位和水质的动态观测资料对模型进行了识别和校正.运用识别后的模型模拟了有填海和无填海两种工况下的海水入侵过程.结果表明:地下水开采是海水入侵的主要原因;在开采地下水条件下,研究区有无填海工程对海水入侵的影响不大,研究区填海工程对海水入侵有一定的阻挡作用,但作用不是很显著.     

生态修复后某矿区下游农田土壤金属污染特征与风险评价

为研究广东省某矿区开展生态修复多年后下游农田土壤的金属污染状况,选取该矿区下游某村周边农田土壤及灌溉水渠作为研究对象,对该区域采集了40个土壤表层样本和8个水体样本,利用Arcgis软件对农田土壤样品中As、Cu、Cd、Pb、Zn、Mn和Fe₂O₃的质量分数进行克里金空间插值,解析该区域农田土壤金属的空间分布特征;采用综合污染指数法和潜在生态风险指数法对该区域耕作层土壤中As、Cu、Cd、Pb、Zn和Mn进行风险评价。结果表明,40个土壤样品中As、Cd、Cu、Zn和Pb的超标率分别为77.5%、70%、87.5%、27.5%和67.5%,说明调查区域农田土壤污染属于多金属复合污染,且对农作物的生产和安全产生巨大的威胁。部分土壤样品中As、Pb和Cd含量超过了中国农用地土壤污染风险管制值,需采取严格管控措施。通过分析土壤金属的空间分布,发现土壤金属含量超标点位主要位于灌溉口与受污染河流周边,且含量与离灌溉口距离成反比。结合目前灌溉水样中的金属均未超标的情况,得出该区域农田土壤污染是由该矿区生态环境修复前所产生的含金属灌溉水导致土壤中金属的积累...