Measurement of atmospheric water vapor vertical column concentration and vertical distribution in Qingdao using multi-axis differential optical absorption spectroscopy
1.Key Laboratory of Environmental Optics and Technology, Anhui Institute of Optics and Fine Mechanics, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China 2.University of Science and Technology of China, Hefei 230026, China 3.CAS Center for Excellence in Urban Atmospheric Environment, Institute of Urban Environment, Chinese Academy of Sciences, Xiamen 361021, China 4.Institutes of Physical Science and Information Technology, Anhui University, Hefei 230601, China
Fund Project:Project supported by the National Key Research and Development Program of China (Grant No. 2018YFC0213201), the National Natural Science Foundation of China (Grant No. 41775029), and the Science-Technology Project of Science and Technology Commission of Shanghai Municipality, China (Grant No. 17DZ1203102)
Received Date:22 April 2020
Accepted Date:15 June 2020
Available Online:10 October 2020
Published Online:20 October 2020
Abstract:The method of retrieving the vertical column density (VCD) and the atmospheric vertical profile of water vapor in visible blue band (434.0–451.5 nm) were studied by using the multi-axis differential optical absorption spectroscopy (MAX-DOAS). First, the method of retrieving the VCD of water vapor was studied. Owing the the fact that the water vapor absorption cross section is of high resolution and it cannot be effectively measured by MAX-DOAS, a convolved cross section with the instrument slit function was used. In addition, the correction factor for water vapor saturation absorption was also used to obtain the true VCD. Second, the water vapor profile retrieved by applying the nonlinear optimal estimation of the trace gas retrieval method (PriAM) was studied, including the effects of aerosol state and the priori profile on the water vapor retrieval. Influence on the water vapor retrieval from the aerosol prior profile linear changes was unapparent. High aerosol state has a significant influence on the water vapor profile retrieval and it was still within the total error tolerance. This indicates that the PriAM is applicable in the water vapor profile retrieval. Using this method, a continuous observation experiment was carried out at the MAX-DOAS Aoshan regional station in Qingdao. The retrieved water vapor VCD results were compared with the daily average data of the European Centre for Medium-Range Weather Forecasts (ECMWF), and the R2 is 0.93. The comparison of the near-surface water vapor concentration of MAX-DOAS retrieval with the ECMWF and sounding data of the University of Wyoming shows that R2 is larger than 0.70 and 0.66, respectively. The two comparison results demonstrate that PriAM can retrieve the atmospheric water vapor VCD and profile accurately. The vertical distribution characteristics of water vapor in Qingdao was analyzed, and the profile results show that the concentration of water vapor in Qingdao was distributed mainly under 1.5 km in height. Keywords:multi-axis differential optical absorption spectroscopy (MAX-DOAS)/ atmospheric water vapor/ vertical column density/ vertical profile
其中$ {\sigma \left(\lambda \right)}_{\mathrm{h}} $为水汽高分辨率吸收截面, H为仪器狭缝函数. 图4示意了以HITEMP 2010为例的卷积过程. 为研究4种数据库下水汽吸收截面在434.0—451.5 nm波段反演水汽的差异, 取相同温压条件下(P = 1013 hPa, T = 293 K)的水汽吸收截面并与仪器狭缝函数卷积, 见图5(a). 选取一天的光谱数据(2019年3月9日), 将卷积后的4种水汽吸收参考截面和光谱数据进行DOAS拟合, 由于以20°仰角计算水汽VCD, 因此取20°仰角作为示例, 拟合残差见图5(b), 即均方根误差(root mean square, RMS). 图5表明, 4种数据库反演的水汽RMS之间差异很小, 不同数据库下水汽有效吸收参考截面对水汽的反演没有显著影响, 具有一致性. 本文使用HITEMP 2010数据库下水汽有效吸收参考截面. 图 4 水汽有效吸收参考截面获取过程 (a) HITEMP 2010水汽高分辨率吸收光谱; (b) 狭缝函数; (c) 水汽有效吸收参考截面 Figure4. Obtaining process of reference cross section for effective absorption of water vapor: (a) HITEMP 2010 high-resolution water vapor absorption spectrum; (b) slit function; (c) reference cross section for effective absorption of water vapor.
图 5 不同数据库下水汽有效吸收截面对比 (a) 4种数据库下水汽有效吸收参考截面; (b) 20°仰角下DOAS拟合残差对比 Figure5. Comparison of effective water vapor absorption cross sections under different databases: (a) Reference cross sections of effective water vapor absorption under four databases; (b) comparison of DOAS fitted residuals at 20° elevation.
其中$ {\sigma \left(\lambda \right)}_{\mathrm{s}{\rm{a}}\mathrm{t}\mathrm{u}} $为饱和吸收截面, 取7个水汽SCD浓度梯度, 范围为2 × 1023 molecules/cm2到1 × 1024 molecules/cm2. 饱和吸收前后的OD见图6(a), 图中表明当SCD小于6 × 1023 molecules/cm2时, OD和ODsatu的差异很小(ODsatu为饱和吸收后的光学厚度). 图6(b)是最大吸收峰442.6 nm处的OD差异, 可以发现, 饱和效应对于光学厚度OD的影响随着浓度的增大而增大. 当SCD为4 × 1023 molecules/cm2, 饱和吸收影响会使OD降低1.76%, SCD为6 × 1023 molecules/cm2, 饱和吸收影响会使OD降低2.61%. 本研究中, DOAS拟合结果SCD均小于4 × 1023 molecules/cm2, 因此饱和吸收效应对本文反演结果的影响较小. 图 6 在蓝光波段水汽饱和吸收对OD的影响 (a) 不同SCD下的OD差别; (b) 最大吸收峰442.6 nm处OD饱和校正前后的差别 Figure6. The effect of water vapor saturation absorption on the OD in the blue band: (a) OD difference under different SCD; (b) the difference before and after OD saturation correction at the maximum absorption peak at 442.6 nm.
由于气溶胶影响光的传输路径, 在PriAM两步反演算法中气溶胶状态会对水汽反演结果造成影响, 不同气溶胶先验廓线的形状和大小也会对反演结果造成影响. 为了量化这些影响, 我们从即墨区环境监测站(120.47° E, 36.38° N, 海拔高度22 m)近地面数据库中选取监测期间污染最为严重一天(3月6日, PM2.5 = 109 μg/m3)和污染最轻一天(3月22日, PM2.5 = 16 μg/m3)的数据, 分别定义为高气溶胶状态和低气溶胶状态, 进行对比研究. 在标准指数型气溶胶先验廓线BP (baseline priori, 本文反演水汽所用的气溶胶先验廓线)的基础上改变先验廓线的大小和形状(指数型和玻尔兹曼型), 增加4条气溶胶先验廓线TP1, TP2, TP3, TP4 (test priori 1, test prior 2, test prior 3, test prior 4)进行测试, 研究了其对水汽廓线反演结果的影响, 整个过程水汽先验廓线均用标准指数型. $ {H}_{\mathrm{m}} $为灵敏度高度上限, 0—$ {H}_{\mathrm{m}} $即为多轴DOAS反演气体的灵敏度范围[11-13]. 平均核是一个矩阵, 用来表征反演对于不同高度大气状态的敏感度, 将每层平均核的最大值连接起来便形成平均核的包络线, 代表了敏感度随高度的变化. $ {d}_{\mathrm{s}} $为自由度, 它的值由平均核对角线上的值相加, 用来表征反演的高度分辨率. 两种气溶胶状态下反演的垂直廓线结果、廓线反演总误差$ {S}_{\mathrm{t}} $, $ {H}_{\mathrm{m}} $, $ {d}_{\mathrm{s}} $和平均核的包络线如图10所示. 图10表明, 两种气溶胶状态下4种TP与BP反演水汽结果的最大差异都在最低层50 m, 3月22日低气溶胶状态下BP在50 m廓线反演总误差$ {S}_{\mathrm{t}} $为0.45 g/kg, 4种TP与BP反演结果差异值分别为0.011, 0.017, –0.007和0.011 g/kg; 3月6日高气溶胶状态下BP在50 m廓线反演总误差$ {S}_{\mathrm{t}} $为0.53 g/kg, 4种TP与BP反演结果差异值分别为–0.28, 0.11, –0.50和–0.44 g/kg, 受高气溶胶状态影响, 反演结果差异比低气溶胶状态大, 但均在BP的反演总误差$ {S}_{\mathrm{t}} $范围内, 表明不同气溶胶状态下反演算法都能很好地重建水汽廓线. 从$ {H}_{\mathrm{m}} $, $ {d}_{\mathrm{s}} $以及平均核的包络线来看, 几种不同气溶胶先验廓线反演结果都比较接近, 可见气溶胶廓线类型对水汽垂直廓线反演结果影响较小. 通过图10还可以发现, 气溶胶状态影响水汽廓线的$ {H}_{\mathrm{m}} $和$ {d}_{\mathrm{s}} $, 3月6日高气溶胶状态下的$ {H}_{m} $和$ {d}_{\mathrm{s}} $均小于3月22日低气溶胶状态下的$ {H}_{m} $和$ {d}_{\mathrm{s}} $, 表明水汽廓线的$ {H}_{\mathrm{m}} $和$ {d}_{\mathrm{s}} $将随着气溶胶消光的增加而降低, 这是因为在高气溶胶条件下光子在空中多次散射影响光的传输路径, 从而会降低廓线的灵敏度高度和自由度. 图 10 气溶胶状态及线型对水汽廓线反演结果的影响 (a) 5种气溶胶先验廓线; (b) 3月6日5种气溶胶先验廓线下反演水汽结果及误差; (c) 3月22日5种气溶胶先验廓线下反演水汽结果及误差; (d) 指数型水汽先验廓线; (e) 3月6日平均核的包络线; (f) 3月22日平均核的包络线 Figure10. Effects of aerosol state and line type on the retrieval results of water vapor profile: (a) Five aerosol prior profiles; (b) the results and errors of water vapor retrieval under the five aerosol prior profiles on March 6; (c) the results and errors of water vapor retrieval under the five aerosol prior profiles on March 22; (d) the exponential water vapor prior profile; (e) the envelope of the average kernel on March 6; (f) the envelope of the average kernel on March 22.
图 12 MAX-DOAS不同高度廓线数据与ECMWF和探空数据的相关性分析 Figure12. Correlation analysis of MAX-DOAS profile data at different heights with ECMWF and sounding data.
图13是采用上述方法反演的MAX-DOAS鳌山区域站监测期间其中10 d的水汽廓线图(0—4 km)示例. 近地面最低是50 m, 然后是200 m, 200 m以上垂直分辨率为200 m, 仪器每完成一次仰角扫描循环将会获得一条垂直廓线. 从图13可知, 在探测时段内, 青岛市鳌山区域站水汽主要集中在1.5 km以下, 且底层浓度较大, 随着高度的升高浓度逐渐降低. 受边界层影响, 2 km以下水汽日间变化比较显著, 2 km以上则变化均匀. 图 13 基于MAX-DOAS反演的水汽0?4 km垂直分布廓线 Figure13. Vertical distribution profile of water vapor 0?4 km based on MAX-DOAS retrieval.