The portable microAeth® MA200 (MA200) is widely applied for measuring black carbon in human exposure profiling and mobile air quality monitoring. Due to it being relatively new on the market, the field lacks a refined assessment of the instrument's performance under various settings and data post-processing approaches. This study assessed the mobile real-time performance of the MA200 to determine a suitable noise reduction algorithm in an urban area, Augsburg, Germany. Noise reduction and negative value mitigation were explored via different data post-processing methods (i.e., local polynomial regression (LPR), optimized noise reduction averaging (ONA), and centred moving average (CMA)) under common sampling interval times (i.e., 5, 10, and 30 s). After noise reduction, the treated data were evaluated and compared by (1) the amount of useful information attributed to retention of microenvironmental characteristics, (2) the relative number of negative values remaining, (3) the reduction and retention of peak samples, and (4) the amount of useful signal retained after correction for local background conditions. Our results identify CMA as a useful tool for isolating the central trends of raw black carbon concentration data in real time while reducing nonsensical negative values and the occurrence and magnitudes of peak samples that affect visual assessment of the data without substantially affecting bias. Correction for local background concentrations improved the CMA treatment by bringing nuanced microenvironmental changes into view. This analysis employs a number of different post-processing methods for black carbon data, providing comparative insights for researchers looking for black carbon data smoothing approaches, specifically in a mobile monitoring framework and data collected using the microAeth® series of Aethalometer.
Black carbon particulate matter with size ranging from 0.01 to 1
Instrument manufacturers in the USA have recently developed a new instrument for measuring black carbon concentrations in a variety of exposure-related contexts, including personal exposure assessment, ambient and vertical profiling, and indoor emissions concentration measurements, among others. This instrument, the microAeth® MA200 (MA200; AethLabs, San Francisco, CA, USA), continuously collects aerosol particles on a filter and measures the optical attenuation (ATN) at five wavelengths (880, 625, 528, 470, and 375 nm) with a data collection time base as frequent as 1 Hz. This instrument supports the DualSpot® loading compensation method, which corrects the optical loading effect (Virkkula et al., 2007) and provides additional information about aerosol optical properties. However, the raw data recorded by the MA200 at high frequencies (e.g., 1 Hz) can exhibit noise that obscures nuanced signals surrounding the central tendency of the data, increasing the difficulty of analysis in mobile settings or during rapidly changing microenvironmental characteristics. These negative values usually contain valid information required for noise reduction or smoothing, and so simply removing them may result in bias. Noise reduction of the raw data without direct removal of negative values is thereby recommended to enhance data quality and temporal resolution (Liu et al., 2020). In addition, when the sampling equipment traverses from a highly polluted to a marginally polluted area, such as a park, the instrument produces strong negative values due to the measurement principle of the instrument and the strength of the pollution gradient between microenvironments. Therefore, the raw black carbon concentrations collected by MA200 need to be post-processed to ensure that researchers can adequately analyse the spatio-temporal distribution of black carbon.
Some progress has been made in the study of black carbon monitoring (Apte et al., 2011; Dons et al., 2012; Cao et al., 2020); however, noise reduction algorithms have not been fully assessed for the new generation of micro-Aethalometer and for mobile monitoring contexts. In previous studies, Hagler et al. (2011) and Cheng and Lin (2013) evaluated optimized noise reduction averaging (ONA) for post-processing mobile monitoring data. Due to the high spatial heterogeneity of black carbon, the ONA algorithm may ignore important microenvironmental effects and lead researchers to perhaps incorrectly conclude that the resolution of microenvironmental source information cannot be determined from their data.
In this study, we aim to determine a suitable noise reduction algorithm for the MA200 Aethalometer, starting with ONA and moving on to two additional smoothing techniques offered by AethLabs in their suite of free online data post-processing (i.e., noise reduction) tools: the local polynomial regression (LPR) and centred moving average (CMA) algorithms. The interpretation accuracy of data analysed and reported upon in black carbon mobile monitoring studies can be increased by assessing the relative performance of these post-processing methods to each other and to ONA. The quality of each noise reduction approach was assessed on data collected in an urban environment and post-processed with ONA, LPR, and CMA. Assessment criteria included (1) retention of detailed information attributed to microenvironmental characteristics, (2) relative number of negative values remaining, (3) reduction and retention of peak samples, and (4) retention of detailed information on microenvironmental characteristics after background correction.
The MA200 measures optical ATN from black carbon on a filter across five
optical wavelength ranges: infrared, red, green, blue, and ultraviolet (880, 625,
528, 470, and 375 nm, respectively). A common black carbon metric called
“equivalent black carbon” (eBC) is assessed via the 880 nm channel, i.e., data from the IR BCc channel. The
detection limit of the MA200 is reported at 30 ng/m
In this study, seven MA200 portable black carbon monitors (serial numbers MA200-0051, MA200-0053, MA200-0059, MA200-0060, MA200-0155, MA200-0153, MA200-159) were used simultaneously to measure black carbon levels in the city centre under different interval times (5, 10, and 30 s). To evaluate the relative performance of MA200, this study analysed black carbon data collected from multiple MA200 devices, identified individually by serial numbers. The instruments were prepared and adjusted in our laboratory before each walk, consisting of “zero” calibration checks and the examination of the MA200 filter cassette, battery, GPS, and memory. Flow calibrations were adjusted with a factory-calibrated flowmeter (Alicat Scientific Inc., Tucson, AZ, USA).
Comparative measurements of the MA200 and a stationary Aethalometer (AE33,
Magee Scientific, Berkeley, USA) taken approximately 30 to 60 min between
walks showed a good agreement (Pearson's
Measurements of black carbon by different MA200 devices.
The MA200 instrument is able to measure black carbon in 1, 5, 10, 30, 60, and 300 s interval times. The 1 s time base exhibits the most
challenging interpretation because of low signal-to-noise ratio, especially
at low concentrations, which is similar to other optical black carbon
monitors (Hagler et al., 2011). Therefore, 1 s measurement resolution may
be most useful when sampling in high-concentration environments, performing
direct emissions testing and requiring high time resolution for the
application. However, the eBC average concentration is low in the city
centre of Augsburg, Germany, (measured at 2.62
To account for the different land-use types of the microenvironments, a fixed walking route within the centre of the city was determined. Wherever possible, the mobile measurements were carried out on the right side of the road, simulating people's common habits (driving and walking on the right side in Germany). All walks along the route were conducted on weekdays, with clear skies and calm winds to avoid misrepresentation of typical urban exposure conditions. The route started from Augsburg University of Applied Sciences (UAS) and continued approximately 14 km for 3 h average walking time, passing through different types of land use to ensure that different microenvironments were represented in the entire area and to ensure the validity of the results (Fig. S1). Meanwhile, as performed in our previous study (Liu et al., 2021), we divided the monitoring route into four microenvironment groups in Augsburg, including high traffic flow (H_Traffic, average 500–1000 vehicles per hour), medium traffic flow (M_Traffic, average 200–500 vehicles per hour), low traffic flow (L_Traffic, average 1–200 vehicles per hour), and park area (N_Traffic, average 0 vehicles per hour), according to the actual traffic density examined during the daytime and determined from the traffic flow observed by street views.
Briefly, the study consisted of the following phases: (1) collecting raw black carbon data using the sampling instruments (MA200); (2) smoothing the acquired raw black carbon data under different post-processing methods (i.e., noise reduction); (3) comparing the noise reduction data based on the detail of microenvironmental characteristic and number of negative values; (4) following the peak samples identification by the coefficient of variation (COV); (5) following the background estimation and correction by thin-plate regression spline (TPRS); and (6), finally, selecting the best noise reduction approach.
In order to reduce the noise of concentration data obtained using high time
resolutions, post-processing algorithms can be used. AethLabs offers tools
for applying several noise reduction algorithms (ONA, LPR, and CMA) to
MA-series device data on its website (
ONA is based on the time series of three parameters in the original
observation data, namely the observation time, the original eBC
concentration, and the amount of change in optical ATN over time, as
specifically described by Hagler et al. (2011). Briefly, a
The LPR algorithm is a non-parametric tool similar to a moving average, but it operates on polynomial regression rather than simple averaging (Masry, 1996; Breidt and Opsomer, 2000; Kai et al., 2010). In LPR, the number of points to smooth across must be manually identified. This value should be chosen to balance effective smoothing of the measured values and the sensitivity required to provide spatial resolution in mobile measurements (e.g., the distance over which the average was taken). The distance resolution was chosen to be approximately 100 m. Assuming the sampling speed is 1.3 m/s, when the interval time is 5, 10, and 30 s, the smoothing numbers of points are 15, 7, and 3, respectively.
The CMA algorithm is a smoothing technique used to make the long-term trends of a time series clearer (Easton and McColl, 1997). Unlike a simple moving average, CMA has no shift or group delay in the data processing, as it incorporates data from both before and after the data point that is being smoothed. The smoothing number of points was determined as previously described in the LPR algorithm, assuming a sampling speed of 1.3 m/s.
After post-processing the data, the characteristic change of the treated data is used as criterion to select the best method. In this regard, when the treated data provide more detailed microenvironmental characteristics, the data reflect the actual situation of air pollutants and facilitate the identification of pollution sources. However, if microenvironmental trends are less pronounced, it may hinder the identification of the pollution source. Therefore, more detailed microenvironmental features result in more accurate information. In addition, the number of remaining negative values is determined as another criterion to propose the best method. Specifically, the method with the smallest proportion of the negative values is selected as the best method. The proportion of negative values (NV) remaining was calculated as the number of negative values divided by the total sample size.
An earlier study by Brantley et al. (2014) compared several methods for
identifying and eliminating peak samples in mobile air pollution
measurements. These include identifying samples outside of a threshold based
on a median produced using road segmentation, an
To calculate the reduction of peak samples (RP), the number of peak samples was calculated before and after post-processing the data, and the difference value was obtained. Then the change in the number of peak samples was divided by the total number of peak samples before post-processing the data. After noise reduction, we compared the reduction and the number of peak samples to further evaluate post-processing methods. In short, if the reduction of peak samples is high, the treated data have a high peak noise reduction without removing the numbers of peak samples. Therefore, the method with high reduction of peak samples and retaining the number of peak samples after post-processing is considered the better method.
The ability of a processing method to adequately remove the estimated background concentration was used to evaluate which method provides the most useful information related to microenvironmental effects. A noise reduction method that appears to better facilitate background estimation and correction (as described below calculated from noise-reduced data via a defined background estimation and evaluation approach) is assessed to select a better post-processing method.
Background correction methods include the single sample standardization method, the sliding minimum method, the linear regression post-processing method, and the spline (of minimum) regression post-processing method. Brantley et al. (2014) suggest that a thin-plate regression spline (TPRS) method can reliably evaluate the background value of mobile measurements and be used to examine the “useful” information in the noise-reduced data (i.e., non-spurious non-background pollution trends). Briefly, the TPRS approach includes three steps: first, the noise reduction data of the pollutant were processed by a 30 s moving average; second, the results of the 30 s moving average were sequentially processed by the specified time window (i.e., 5 and 10 min), and the position of the minimum sample of pollutant concentration was identified in each window; and, finally, thin-plate spline regression was used to fit the sample of minimum pollutant concentration obtained in the previous step, then the background concentration at each time point was obtained.
The average eBC concentrations of raw, ONA-processed, LPR-processed, and CMA-processed data (measurements 1–10) monitored by all instruments were compared in this study (Table S2). The results show that the three post-processing methods accounted of approximately 1 % bias from the average of raw concentrations (except measurement 5, ONA-processed data at 5 s). This indicates that the average concentration under each post-processing method did not affect the average concentration of the raw unprocessed data.
As shown in Fig. 1, three MA200 devices were used at the time bases of 5, 10, and 30 s. The proportion of negative values in the raw data collected under different time bases were 42.1 %, 37.6 %, and 30.5 % for 5, 10, and 30 s, respectively (Fig. 1a, Table 2, Fig. S4a). Following this, the raw data were processed using ONA, LPR, and CMA (Fig. 1b, c, and d).
The temporal fluctuations of the black carbon levels
measured with the MA200 at sampling time bases of 5, 10, and 30 s during
a typical sampling period (about 190 min);
The proportion of negative value (NV) and average reduction of peak (RP) samples under the different post-processing methods (values are shown in percent; NV [%]: proportion of negative values remaining; RP [%]: average reduction of peak samples. Symbol “–” means no data; measurements 1–10).
In the 5 s time base, the eBC values changed very rapidly (Fig. 1a), and the
ONA processing of the data resulted in only one value (which was negative)
(Fig. 1b). Thus, the microenvironmental characteristics of the eBC
concentration were not reproduced. We found that all
In the 10 s interval time base, negative values were not found after ONA processing, suggesting that a reasonable smoothing effect is obtained at low black carbon concentration. The microenvironmental characteristics presented strong changes compared to the raw data, leaving less detailed information on air pollution. After post-processing with LPR and CMA, the microenvironmental characteristics revealed more detailed information on air pollution, with 30.2 % of negative values for LPR and 25.3 % for CMA. In the 30 s interval time base, the negative values comprised 0 % of the post-processed data for ONA, 25.5 % for LPR, and 22.4 % for CMA. The 30 s interval dataset presented the lowest proportion of negative values before and after post-processing, due to the longer interval time of sampling. However, the longer 30 s measurement period results in more distance covered during each measurement, given the mobile nature of the sampling device. Thus, 30 s black carbon measurements may be too long to detect local concentration peaks in urban contexts that supported another study (Kerckhoffs et al., 2016).
The ONA algorithm showed a strong tendency to remove negative values and,
depending on the
The processing of peak sample is a pivotal evaluation index for the measurement of time-averaged roadside air quality. Passing vehicles, for example, may bias estimates of typical local concentrations due to their contribution to the dataset of peak concentrations that may substantially relate to arithmetic averages. Therefore, after noise reduction, we compare the reduction and the retained number of peak samples to further evaluate the post-processing methods.
Identification of the spatial
In the interval time 5 s, the average reduction of peak samples (RP) for the LPR and CMA algorithms was 72.0 % and 87.4 %, respectively (as discussed above, the ONA method could not be used). In this interval time, the reduction of peak samples was relatively high, indicating that when monitoring black carbon at low concentrations and high sample frequencies, drastic noise may occur in the raw data, and higher noise reduction may affect the actual values. Therefore, a suitable interval time should be considered when monitoring low eBC concentrations. In the interval time 10 s, the average reduction of peak samples for CMA (47.7 %) is higher than ONA (5.54 %) and LPR (22.7 %). In the interval time 30 s, CMA presented the greatest average reduction of peak samples (39.1 %) compared to ONA (6.24 %) and LPR (0.62 %) (Table 2, Fig. S4b). The retention of peak samples remaining after post-processing was also assessed using the COV method (measurements 1–10). The result showed that all three algorithms retained all peak samples before and after post-processing. In this regard, CMA retained all peak samples despite the highest reduction in their magnitude. Therefore, CMA highlights microenvironmental trends while preserving the identity of peak samples, facilitating the identification of local pollution sources, and may thus be a better post-processing method than ONA or LPR (Table 2, Fig. S4b).
To further characterize the distribution of peak sample concentration under CMA, we performed an intensive graphical analysis on a single data stream (measurement 4; Fig. 2). As shown in Fig. 2, eBC values along the main roads and intersections were higher than other locations, presumably due in large part to stop-and-go traffic and cars in close proximity to the mobile monitor (Fig. 2). It can be seen from Fig. 2a that the peak samples of black carbon were mainly found in four locations, represented by red triangles. Vehicle counts and traffic in these locations vary depending on the time of measurement. The highest eBC values were repeatedly found in the streets with moderately high traffic volumes and dense coverage with relatively high buildings (street canyon situation), indicating that heterogeneity in air pollution concentrations in Augsburg and similar settings is largely caused by a combination of effects from traffic and topography (Buonanno et al., 2011). To determine whether peak samples are due to local sources or instrumental artefacts and to provide further evidence that traffic and topography effects are primary contributors to spatial heterogeneity in pollution concentrations, we compared the data measurements of the three co-located MA200 units during measurements 5, 6, and 7. The results showed that there were no major differences in the hotspot areas (an indicator of considerable peak samples) identified by the measurements of the three instruments (Fig. S5).
Local air pollution can be highly affected by long-range and regional transport. The timing and magnitude of such transport events vary in space and time and are highly dependent upon the stochasticity of meteorology. As a result, local background concentration changes may vary, affecting the comparability of measurements made at the same location at different times (Brantley et al., 2014). For this reason, reliable comparison of time-variable mobile measurements across a city (and thus reliable pinpointing of hotspots and pinpointing of key local sources) requires effective methods to estimate, isolate, and remove the effects of fluctuations in background concentration. Our analysis indicates that the effectiveness of background correction is affected by the noise reduction method chosen during post-processing.
After post-processing, the data were evaluated using the TPRS method. We
calculated the 5 and 10 min background concentrations under different
post-processing approaches. As shown in Fig. 3a and b, the background
concentration after LPR processing has both the largest proportion of
negative values and the most negative values (i.e., negative values of the
greatest absolute magnitude), resulting in estimates of background-corrected
concentrations that are greater than actual monitored concentrations.
Background concentrations calculated after ONA and CMA post-processing
presented fewer and lower negative values than LPR but were not
convincingly different from each other. Therefore, to further compare the
ONA and CMA algorithms, we also compared concentrations after background
correction (Fig. 3c and d). As shown in Fig. 3c and d, when the
concentration is lower than 1
Background concentration of black carbon under different
time series for
In order to verify the CMA applicability and its advantages, this study further analysed the eBC concentrations measured by a fixed background monitoring station at the University of Applied Sciences (UAS) (Fig. S6) (Cyrys et al., 2006). The background value under the 5 min window exhibits wave-like characteristics, and the fitting curve in the 10 min window is relatively smooth. However, the TPRS-based background value often does not fluctuate greatly over short periods, and the black carbon background value curve under the 5 min window does not conform to the “actual” urban background situation as estimated using the fixed-site monitor data, which are assumed to primarily represent the fluctuations in background concentrations. Moreover, by comparing the curve produced by the spline of 10 min minimums with the eBC background concentration (“Background-UAS”, Fig. S6), it can be found that the background correction method based on the time series can well characterize the time-varying characteristics of background pollution in each experiment, suggesting that, of the two options, 10 min showed the better window for fitting the background value curve of black carbon.
Under the TPRS method, the background concentration of eBC can be fitted at any sampling time. The TPRS-estimated background contribution of the observed eBC concentration averaged 37.8 % of the total measured concentration. However, when the contribution of background concentration to a single measurement was examined, a large fluctuation (10.4 %–71.3 %) was observed, which may be closely related to sizable changes in the meteorological conditions, traffic conditions along the road (and over time at the same point in the road), and urban street canyon effects in each measurement. Therefore, based on the comparison of background correction, CMA showed better applications for estimating the background concentration and location source contribution.
The proposed decision tree for mobile monitoring data from the microAeth® MA200.
To verify the generalizability of our assessment, we performed another three measurement runs in Munich (measurements 8, 9, and 10). Raw data were post-processed for noise reduction using CMA (Fig. S7). The results showed that the following method is equally applicable in a city like Munich as in our study site in Augsburg, which are two cities that differ in location and environmental characteristics (e.g., population, economy, traffic density, etc.). After treated using CMA, the peak samples can be identified in different interval times (Fig. S8), and the estimated background concentrations showed few negative values (Fig. S9). Further research into the transferability of our results to a more diverse set of contexts is still needed.
The MA200 is widely used to measure human exposure to black carbon and for
mobile air quality monitoring. In this study the MA200 devices were applied in
mobile measurements in an urban area (Augsburg), and the sensitivity of the
final analysis to various data post-processing methods was investigated. In
contrast to our findings, Hagler et al. (2011) suggested the use of the ONA
algorithm to post-process Aethalometer data from microAeth AE51, portable
AE42, and rack-mounted AE21 Aethalometer instruments (Magee Scientific, Berkeley, CA,
USA). In their analysis, ONA demonstrated a strong noise reduction in all
datasets and retained spatio-temporal variation. ONA also reduced the
occurrence of negative data values in low concentration sampling
environments. However, for the microAeth® series of black
carbon monitoring instruments, our study showed that ONA under reasonable
In addition, our analysis highlights that the selection of an appropriate data post-processing method is crucial to the proper assessment and interpretation of exposure-relevant microenvironmental contributors to pollution concentrations in urban areas. This analysis is important when estimating exposures that occur during transit, where spatio-temporal variability in pollution concentrations is vast, like in commuter traffic (Snyder et al., 2013). Due to the typically low-but-heterogeneous nature of eBC concentrations in many areas like Augsburg, noisy measurement with the MA200 under high-frequency sampling may obscure actual trends in measured values. This study demonstrated that post-processing MA200 data using CMA can reliably extract the actual signals from such noise and, alternatively, that post-processing via ONA and LPR could be less reliable. Future researchers and agencies may find a distillation of our results in the form of the flow diagram in Scheme 1 useful in determining how to reliably assess spatio-temporal variability of MA200 measurements for black carbon in different microenvironments.
A mobile monitoring campaign was conducted in the city centre of Augsburg, Germany, to determine a suitable noise reduction algorithm for the MA200 Aethalometer. Our results showed that, at the interval times of 5, 10, and 30 s, CMA post-processing effectively removed spurious negative concentrations without major bias and reliably highlighted effects from local sources, effectively increasing spatio-temporal resolution in mobile measurements. Evaluation of the effects of each method on peak sample reduction and the estimation of background concentrations further support the reliability of CMA. Further analysis is needed to understand how well these findings apply in different seasons; across different diurnal patterns; and in more rural, more urban, and non-German locations.
The data are available upon request by contacting the first author of the paper (xiansheng.liu@helmholtz-muenchen.de).
The supplement related to this article is available online at:
XL was responsible for data curation, methodology and software analysis, and writing original draft. HH was responsible for methodology, software analysis, and writing original draft. XZ was responsible for funding acquisition and project administration. LDH was responsible for discussion and review and editing. JSK was responsible for investigation and supervision. JB and GJ were responsible for methodology. AHAW and BSH were responsible for writing, review and editing. RZ was responsible for investigation.
L. Drew Hill is an employee of AethLabs (San Francisco, CA, USA) and Andrew H. A. White was, at the time of contribution, an intern at AethLabs. These affiliations did not affect the conclusions of the paper.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We gratefully thank Erik Hopp (AethLabs, San Francisco, CA, USA) for the implementation and maintenance of the AethLabs data processing web tool.
The work is funded by the National Key Research and Development Program of China (grant no. 2020YFB1806500), Support Project of High-level Teachers in Beijing Municipal Universities in the Period of 13th Five-year Plan (grant no. CIT&TCD201904037), the National Natural Science Foundation of China (grant no. 81800854), and the Germany Federal Ministry of Transport and Digital Infrastructure (BMVI) as part of SmartAQnet (grant no. 19F2003B).
This paper was edited by Alfred Wiedensohler and reviewed by two anonymous referees.