Anthropogenic carbon dioxide (CO₂) emissions mainly come from cities and their surrounding areas. Thus, continuous measuring of CO₂ in urban areas is of great significance to studying human CO₂ emissions. We developed a compact, precise, and self-calibrated in-situ CO₂/H₂O sensor based on TDLAS (tunable diode laser absorption spectroscopy), WMS (wavelength modulation spectroscopy), and VCSEL (vertical-cavity surface-emitting laser). Multi-harmonic detection is utilized to improve the precision of both measurements to 0.02 ppm for CO₂ and 1.0 ppm for H₂O. Using the developed sensor, we measured CO₂ concentrations continuously in the city center of Munich, Germany, from February 2018 to January 2019. Urban CO₂ concentrations are strongly affected by several factors, including vegetation photosynthesis and respiration (VPR), planetary boundary layer (PBL) height, and anthropogenic activities. In order to further understand the anthropogenic contribution in terms of CO₂ sources, the HySPLIT (Hybrid Single-Particle Lagrangian Integrated Trajectory) model was applied to calculate six-hour backward trajectories. We analyzed the winter CO₂ with the trajectory clustering, PSCF (potential source contribution function), and CWT (concentration weighted trajectory) methods, and found that local emissions have a great impact on urban CO₂ concentration, with main emission sources in the north and southeast directions of the measurement site. In situations with an uneven trajectory distribution, PSCF proves somewhat superior in predicting the potential emission sources compared to CWT.

This study estimates the influence of anthropogenic emission reductions on the concentration of particulate matter with a diameter smaller than 2.5 µm (PM_2.5) during the 2020 lockdown period in German metropolitan areas. After accounting for meteorological effects, PM_2.5 concentrations during the spring 2020 lockdown period were 5 % lower compared to the same time period in 2019. However, during the 2020 pre-lockdown period (winter), PM_2.5 concentrations with meteorology accounted for were 19 % lower than in 2019. Meanwhile, NO₂ concentrations with meteorology accounted for dropped by 23 % during the 2020 lockdown period compared to an only 9 % drop for the 2020 pre-lockdown period, both compared to 2019. SO₂ and CO concentrations with meteorology accounted for show no significant changes during the 2020 lockdown period compared to 2019. GEOS-Chem (GC) simulations with a COVID-19 emission reduction scenario based on the observations (23 % reduction in anthropogenic NO_x emission with unchanged anthropogenic volatile organic compounds (VOCs) and SO₂) are consistent with the small reductions of PM_2.5 during the lockdown and are used to identify the underlying drivers for this. Due to being in a NO_x-saturated ozone production regime, GC OH radical and O₃ concentrations increased (15 % and 9 %, respectively) during the lockdown compared to a business-as-usual (BAU, no lockdown) scenario. O_x (equal to NO₂+O₃) analysis implies that the increase in ozone at nighttime is solely due to reduced NO titration. The increased O₃ results in increased NO₃ radical concentrations, primarily during the night, despite the large reductions in NO₂. Thus, the oxidative capacity of the atmosphere is increased in all three important oxidants, OH, O₃, and NO₃. PM nitrate formation from gas-phase nitric acid (HNO₃) is decreased during the lockdown as the increased OH concentration cannot compensate for the strong reductions in NO₂, resulting in decreased daytime HNO₃ formation from the OH + NO₂ reaction. However, nighttime formation of PM nitrate from N₂O₅ hydrolysis is relatively unchanged. This results from the fact that increased nighttime O₃ results in significantly increased NO₃, which roughly balances the effect of the strong NO₂ reductions on N₂O₅ formation. Ultimately, the only small observed decrease in lockdown PM_2.5 concentrations can be explained by the large contribution of nighttime PM nitrate formation, generally enhanced sulfate formation, and slightly decreased ammonium. This study also suggests that high PM_2.5 episodes in early spring are linked to high atmospheric ammonia concentrations combined with favorable meteorological conditions of low temperature and low boundary layer height. Northwest Germany is a hot-spot of NH₃ emissions, primarily emitted from livestock farming and intensive agricultural activities (fertilizer application), with high NH₃ concentrations in the early spring and summer months. Based on our findings, we suggest that appropriate NO_x and VOC emission controls are required to limit ozone, and that should also help reduce PM_2.5. Regulation of NH₃ emissions, primarily from agricultural sectors, could result in significant reductions in PM_2.5 pollution.

A novel gas recognition and concentration detection algorithm consisting of a dynamic wavelet convolutional neural network (DWCNN) and a many-to-many long short-term memory-recurrent neural network (LSTM-RNN), respectively, is proposed as a replacement for the traditional data processing algorithm in artificial olfaction. The proposed DWCNN gas recognition algorithm does not require gas signal preprocessing, and it directly converts the raw time-domain gas signal data to 64 * 64 2-D gray images as the input layer of a convolutional neural network (CNN). The experiments show that the recognition accuracy of CO, H ₂ , and the gas mixture of CO and H ₂ is nearly 100%, and the many-to-many LSTM-RNN algorithm requires only a few labeled data from the steady-state values of the gas sensor array signals. In addition, comparisons with other neural network multilayer perceptrons (MLPs), gated recurrent unit (GRU) algorithms, and conventional algorithms, such as the Bayesian ridge, support vector machines (SVMs), decision tree, k-nearest neighbor (KNN), random forest, AdaBoost, gradient-boosting decision tree (GBDT), and bagging, revealed that the algorithm can obtain a higher concentration detection accuracy, which was evaluated using two different kernel functions for the kernel principal component analysis (KPCA) dimensionality reduction: polynomial and rbf. The experimental results demonstrated that the proposed many-to-many LSTM gas concentration detection model outperformed the abovementioned algorithms and can more accurately estimate the concentration of different gases while using less labeled data.

In many cities around the world the overall air quality is improving, but at the same time nitrogen dioxide (NO₂) trends show stagnating values and in many cases could not be reduced below air quality standards recommended by the World Health Organization (WHO). Many large cities have built monitoring stations to continuously measure different air pollutants. While most stations follow defined rules in terms of measurement height and distance to traffic emissions, the question remains of how representative are those point measurements for the city-wide air quality. The question of the spatial coverage of a point measurement is important because it defines the area of influence and coverage of monitoring networks, determines how to assimilate monitoring data into model simulations or compare to satellite data with a coarser resolution, and is essential to assess the impact of the acquired data on public health. In order to answer this question, we combined different measurement data sets consisting of path-averaging remote sensing data and in situ point measurements in stationary and mobile setups from a measurement campaign that took place in Munich, Germany, in June and July 2016. We developed an algorithm to strip temporal from spatial patterns in order to construct a consistent NO₂ pollution map for Munich. Continuous long-path differential optical absorption spectroscopy (LP DOAS) measurements were complemented with mobile cavity-enhanced (CE) DOAS, chemiluminescence (CL) and cavity attenuated phase shift (CAPS) instruments and were compared to monitoring stations and satellite data. In order to generate a consistent composite map, the LP DOAS diurnal cycle has been used to normalize for the time of the day dependency of the source patterns, so that spatial and temporal patterns can be analyzed separately. The resulting concentration map visualizes pollution hot spots at traffic junctions and tunnel exits in Munich, providing insights into the strong spatial variations. On the other hand, this database is beneficial to the urban planning and the design of control measures of environment pollution. Directly comparing on-street mobile measurements in the vicinity of monitoring stations resulted in a difference of 48 %. For the extrapolation of the monitoring station data to street level, we determined the influence of the measuring height and distance to the street. We found that a measuring height of 4 m, at which the Munich monitoring stations measure, results in 16 % lower average concentrations than a measuring height of 1.5 m, which is the height of the inlet of our mobile measurements and a typical pedestrian breathing height. The horizontal distance of most stations to the center of the street of about 6 m also results in an average reduction of 13 % compared to street level concentration. A difference of 21 % in the NO₂ concentrations remained, which could be an indication that city-wide measurements are needed for capturing the full range and variability of concentrations for assessing pollutant exposure and air quality in cities.

Conventional gas recognition algorithms used in artificial olfaction rely on manually selected steady-state features of the sensor array signal and ignore the dynamic process signal response features of the sensor array; hence, a large amount of dynamic feature information, which can improve the gas recognition accuracy, is lost. To address this problem, a wavelet transform coefficient map-capsule network (WTCM-CapsNet) model for gas recognition is presented in this paper. It directly uses a raw signal from a sensor array without any signal-processing steps and consists of a wavelet coefficient map and a capsule network (CapsNet) model with two submodels. The WTCM-CapsNet model directly performs a five-scale wavelet transform on the sensor array signal and extracts 4,096 wavelet coefficients to form a 2D ${64}^{ \boldsymbol {\ast }}{64}$ wavelet coefficient image as the CapsNet input. Compared with conventional gas recognition algorithms, such as the convolutional neural network (CNN) with 99.3% gas recognition accuracy, the K-nearest neighbor (KNN) with 95.74% gas recognition accuracy, support vector machine (SVM) with 96.45% gas recognition accuracy, random forest with 95.74% gas recognition accuracy, naive Bayes with 92.91% gas recognition accuracy and BP neural network (BPNN) algorithms with 97.87% gas recognition accuracy, which extract the steady-state values of the time-domain signal and the wavelet coefficient maps of the whole dynamic response signal in the time-frequency domain as input features, CapsNet uses an iterative routing-by-agreement mechanism that can effectively improve the gas recognition accuracy to approximately 100%, as demonstrated by the dynamic response signal of the gas sensor array collected by our fabricated artificial olfactory system.

This study estimates the influence of anthropogenic emission reductions on the concentration of particulate matter with a diameter smaller than 2.5 μm (PM_2.5) during the 2020 lockdown period in German metropolitan areas. After accounting for meteorological effects, PM_2.5 concentrations during the spring 2020 lockdown period were 5 % lower compared to the same time period in 2019. However, during the 2020 pre-lockdown period (winter), meteorology accounted for PM_2.5 concentrations were 19 % lower than in 2019. Meanwhile, meteorology accounted for NO₂ concentrations dropped by 23 % during the 2020 lockdown period compared to an only 9 % drop for the 2020 pre-lockdown period, both compared to 2019. Meteorology accounted for SO₂ and CO concentrations show no significant changes during the 2020 lockdown period compared to 2019. GEOS-Chem (GC) simulation with a COVID-19 emission reduction scenario based on the observations (23 % reduction in NO_X emission with unchanged VOC and SO₂) are consistent with the small reductions of PM_2.5 during the lockdown and are used to identify the underlying drivers for this. Due to being in a NO_X saturated ozone production regime, GC OH radical and O₃ concentrations increased (15 and 9 %, respectively) during the lockdown compared to a Business As Usual (no lockdown) scenario. The increased O₃ results in increased NO₃ radical concentrations, primarily during the night, despite the large reductions in NO₂. Thus, the oxidative capacity of the atmosphere is increased in all three important oxidants, OH, O₃, and NO₃. PM nitrate formation from gas-phase nitric acid (HNO₃) is decreased during the lockdown as the increased OH concentration cannot compensate for the strong reductions in NO₂ resulting in decreased day-time HNO₃ formation from the OH + NO₂ reaction. However, night-time formation of PM nitrate from N₂O₅ hydrolysis is relatively unchanged. This results from the fact that increased night-time O₃ results in significantly increased NO₃ which roughly balances the effect of the strong NO₂ reductions on N₂O₅ formation. Ultimately, the only small observed decrease in lockdown PM_2.5 concentrations can be explained by the large contribution of night-time PM nitrate formation, generally enhanced sulfate formation and slightly decreased ammonium. This study also suggests that high PM_2.5 episodes in early spring are linked to high atmospheric ammonia concentrations combined with favorable meteorological conditions of low temperature and low boundary layer height. North-West Germany is a hot-spot of NH₃ emissions, primarily emitted from livestock farming and intensive agricultural activities (fertilizer application), with high NH₃ concentrations in the early spring and summer months. Based on our findings, we suggest that appropriate NO_X and VOC emission controls are required to limit ozone, and that should also help reduce PM_2.5. Regulation of NH₃ emissions, primarily from agricultural sectors, could result in significant reductions in PM_2.5 pollution.