We investigated spatial and temporal patterns in the concentration and
composition of submicron particulate matter (PM
Organic aerosol (OA) contributes a significant fraction of the total ambient
particulate matter (PM) mass
Concentrations of PM and other pollutants are spatially variable in urban
areas and these spatial variations drive differences in human exposures. For
example, concentrations of ultrafine particles, NO, CO, and particulate black
carbon (BC) are enhanced near highways by a factor of 2–3 relative to areas
Studying these intra-urban PM variations can reveal the major sources
influencing local air quality and help inform mitigation strategies. In
particular, mobile sampling enables the deployment of high-time-resolution
measurements that can identify specific PM sources. For example,
Quantifying PM and OA spatial gradients in urban areas, and identifying the
sources driving those gradients, is important because more than half of the
world's population lives in urban areas
This study presents results of mobile measurements conducted in Oakland,
California. Oakland is a densely populated (
Several prior studies have focused on air quality in Oakland because of the
influence of ships and associated drayage trucks (trucks that transport cargo
between the port and warehouses) driving through the residential district.
The objective of this study is to determine which emission sources most
strongly impact spatial patterns in the local air quality of Oakland. We use
mobile sampling with aerosol mass spectrometry (AMS) to investigate spatial
gradients in the concentrations and chemical composition of OA across the three
distinct areas of Oakland: port, residential West Oakland, and downtown.
Additionally, using positive matrix factorization of AMS data
We conducted mobile sampling between 10 July and 2 August 2017 in Oakland,
CA,
using a mobile laboratory. Data were collected as part of the Center for Air,
Climate, and Energy Solutions (CACES) air quality observatory
We divided the sampling domain into three main areas: port, West Oakland, and
downtown. Owing to the relatively large size and road length density
(16.6 km of road per km
Figure
All instruments in the mobile laboratory were powered by a 110 V, 60 Hz
alternator coupled to the van's engine. A 0.5
We used a high-resolution time-of-flight aerosol mass spectrometer
We first adjusted the recorded time stamps on all instrument samples based on the predetermined instrumental response times. Response times were measured by releasing a tracer at the sample inlet and recording the time lag in response from instruments while the van was stationary. This adjustment was done to assign each data point to the time the sample entered the inlet (as opposed to the time the data point was recorded by the instrument). Additionally, a sampling duration offset was applied to the AMS data time stamps. This is because each AMS measurement is an average of mass spectra collected for 20 s and the time stamp is assigned at the end of the 20 s period. To ensure that the sample was spatially representative of the distance traveled by the van during the 20 s sampling interval, each AMS sample was advanced 10 s in time to assign the measured concentration to the middle of the 20 s sampling interval instead of the end. Next, upon alignment with GPS data, we assigned spatial coordinates to all instrument samples.
For spatial aggregation, we used a procedure similar to the “road length
snapping” procedure used by
The amount of time spent at a 200 m magnet can be longer than 20 s on days
when driving was paused at that magnet, e.g., for traffic lights, refueling
stops, etc. These samples can bias a magnet's representative concentration
when averaging is performed across multiple days. Conversely, as shown in the
zoom-in (inset d) in Fig.
We processed AMS data using SQUIRREL 1.57I and PIKA 1.16I routines in Igor
Pro 6.37
To identify sources of OA, we applied positive matrix factorization (PMF) to
the two-dimensional OA matrix (time series along rows, concentrations of
high-resolution organic ions up to
Over the course of mobile sampling, the urban background air quality can have
daily and diurnal variations due to meteorological changes (Fig. S2). These
variations can be accounted for with the help of concurrent stationary
measurements performed at an urban background location. As discussed in
Supplement Sect. S1, accounting for temporal trends only had a minor (
In order to compare observations of OA and its factors across areas
influenced by different emissions (port, West Oakland, and downtown), we
first determined the precision of these measurements by resampling the pool
of data occurring in these areas. The strength (i.e., number of elements) of
a bootstrapped dataset was the same as the strength of the dataset collected
in that area. For instance,
In Fig.
We now discuss the spatial patterns of OA in more detail. Figure
Median organic aerosol concentration at each magnet. The pie chart
shows the median contribution of AMS-measured non-refractory (organics: Org,
sulfate:
The lower panel of Fig.
Cumulative distributions, mean, and median of OA concentrations in Port, West Oakland, and downtown.
Downtown has a median OA concentration of 5.7
We identified three OA factors with distinct mass spectra using positive
matrix factorization (PMF) of AMS data: hydrocarbon-like OA (HOA), cooking OA
(COA), and less-oxidized oxygenated OA (LO-OOA). These factor profiles are
shown in Fig.
The HOA factor has an elevated signal at the series of
The COA factor has a distinct signal at
Compared to the HOA and COA factors, this factor is relatively more
oxygenated with a distinct peak at
Having made no thermodenuded measurements of OA volatility, we identify the
third oxygenated factor in our PMF solution as LO-OOA because (a) the mass
spectrum and elemental ratios are similar to those reported for LO-OOA (and
SV-OOA) elsewhere (Fig. S9), and (b) this factor is correlated with
the AMS-measured
Average contributions of COA, HOA, and LO-OOA to total OA are shown in
Fig.
Previous AMS-PMF studies have reported the presence of both LO-OOA and MO-OOA
factors in ambient OA. The absence of an MO-OOA factor in Oakland can be
explained by the hypothesis that air masses arriving in Oakland are oceanic.
These air masses are expected to contain very low OA concentrations, even
though most of this OA is highly oxidized MO-OOA
PMF decomposes measured OA concentrations using a linear combination of
contributions from static factors. The amount of observed mass that cannot be
explained by the reconstructed factor contributions is binned into residual
mass. Residuals of factorization are shown in Fig. S11. The ratio of scaled
residuals,
A four-factor solution was also examined (Fig. S12). While the mass spectra and
fractional contributions of both HOA and COA remain unchanged from the
three-factor solution, the LO-OOA factor from the three-factor solution was further
deconvolved into a more oxygenated MO-OOA factor and a fourth less oxygenated
factor that bore no similarity to the typical LO-OOA factor spectra reported
in the literature. We discarded this four-factor solution because (a) given that
fresh OA factors (HOA and COA) and OOA factors form a continuum of
atmospheric oxygenation, we do not expect the presence of the fresh OA and
MO-OOA factors while an LO-OOA factor is absent, (b) we did not find a strong
PMF-independent tracer correlation (e.g., with AMS-measured particulate
In this section, we further analyze the spatial and temporal patterns of OA
and its factors. All times are presented in local time (Pacific Daylight
Time, UTC minus 7 h). We begin by examining the primary–secondary split of
OA and how this split varies across space and time. Understanding variability
in the primary fraction of OA is important because we know from recent
findings that in close proximity to sources such as highways
Overall, the OA mass in Oakland is split into primary and secondary factors
roughly evenly (Fig.
Figure
Figure
In Fig.
Diurnal shift of the source of primary OA from morning rush hour traffic to midday cooking activities. Dots show individual measurements and diamonds show hourly medians over the entire sampling campaign.
During the morning period (08:00 to 11:00 LT), the median OA concentration in
downtown is 40 % (
Cumulative distribution functions of total OA and the three factors identified in this study resolved by area (port, West Oakland, and downtown) and local time (LT) of day: morning (08:00 to 11:00), midday (11:00 to 14:00 LT), and afternoon (14:00 to 18:00 LT). The colored box around each set of CDFs indicates the period of the day those data represent.
During the morning period, the median HOA concentration in downtown is 52 %
(
Overall, HOA is highest in downtown in the morning, despite the fact that all
of Oakland has roughly equal proximity to highways. While we do not have
detailed traffic data for Oakland, it is reasonable to assume that downtown
receives a large influx of commuters during the morning rush hour and thus
would be expected to have the highest HOA concentrations. Downtown also has a
higher road length density compared to West Oakland and port and, as a
result, can accommodate a larger traffic volume per km
The median COA concentration in downtown is 55 % (
Overall, COA is consistently highest in downtown, which is not surprising
given the large number of restaurants. The spatial distribution
of COA in West Oakland and port is uniformly low in the morning, as
demonstrated by the steepness of the distribution functions in Fig.
The median LO-OOA concentration in downtown is 36 % (
LO-OOA concentrations are consistently higher in downtown compared to port
and West Oakland. This finding is unexpected because LO-OOA is secondary; the
null hypothesis for secondary species is that concentrations would be
spatially uniform. Multiple lines of evidence contribute to the conclusion
that LO-OOA is indeed higher in downtown than the port and West Oakland.
First, as shown in Figs.
The increasing concentrations of LO-OOA with increasing inland distance
(Fig.
The enhanced photochemical production of SOA could be due to several factors.
First, the pool of reactive SOA precursor vapors is likely enhanced in
downtown relative to other areas. We show above in Fig.
Spatial variations in measured LO-OOA on select intra-polygon
transit drives on different days, colored by three different diurnal local time
(LT) periods. Drives that reasonably spanned the longitudinal extent of the
sampling domain with minimal latitudinal displacement were picked for this
plot. Markers are individual LO-OOA samples and lines are fits for each
transect drive. Fits that have a positive slope with proximity to downtown
are shown as solid lines. The only fit that has a negative slope is shown as
a dashed line.
Second, concentrations of the hydroxyl (OH) radical may also be higher
downtown. The OH radical is the dominant daytime oxidizing agent, especially
for reduced compounds emitted from motor vehicles. It has been shown that in
urban, polluted environments (street canyons), high
A third factor may also contribute to higher precursor concentrations and
therefore additional LO-OOA in downtown. The presence of tall buildings in
downtown can create higher surface roughness, which in turn can reduce
pollutant dispersion and promote internal recirculation
The combined impact of vehicle emissions on OH (via HONO) and gas-phase
precursor concentrations in downtown would be expected to be largest in the
morning, since these are co-emitted with HOA. Indeed, the largest enhancement
of LO-OOA in downtown occurs in the morning hours at the same time as the
largest enhancement of HOA. From morning to midday, LO-OOA concentrations
become more spatially uniform; median LO-OOA concentrations in West Oakland
and port increase by 23 % (0.6
We further investigated the enhanced photochemical activity in downtown by
analyzing mobile measurements of particulate sulfate (
Ships associated with the port are the major source of
In this subsection, we compare the influence of emissions from diesel trucks
against that from gasoline-powered vehicles. We do this by comparing
concentrations and ratios of OA, HOA, BC, and CO. Traditionally, diesel
vehicles have significantly higher emissions of HOA and BC than gasoline
vehicles
Fine-scale maps of
Based on typical emissions from gasoline and diesel vehicles, we would expect
that (a) all areas with heavy traffic should have elevated concentrations of
OA, HOA, BC, and CO, and (b) diesel-dominated areas will have higher BC
Figure
All three ratios suggest less diesel influence in West Oakland and downtown
than in port. BC
The HOA
The overall picture painted by Fig.
Since the vehicle fleet appears to be significantly different between
downtown and port, we attempted to derive separate HOA factors for these two
areas as a means to directly quantify gasoline versus diesel emissions using
the AMS. We isolated port OA data from downtown OA data and factorized them
separately using PMF. The HOA factors identified for port and downtown are nearly
identical (
In this section, we investigate the measured elemental ratios (H
The Van Krevelen plane. Gray points represent all OA measurements in
this study. Diamonds represent OA factors identified in this study. For
reference, the placement of OA factors from other ambient measurements is shown:
Pittsburgh
Figure
Figure
Finally, this VK analysis also reinforces the choice of a three-factor PMF
solution in this study. In Fig. S15, the three- and four-factor PMF solutions are
compared on the VK plane, along with other reference data as already
described in Fig.
Having one of the largest shipping ports in the US, the air quality in Oakland has
been historically impacted by shipping-related activities such as the presence of
ships burning high-sulfur fuel and drayage trucks driving through the
directly adjacent residential neighborhood
In addition to the port, Oakland also has a central business district
(“downtown”) that has activities such as domestic vehicular traffic and
cooking, similar to other urban areas. Urban downtowns are also a prominent
source of organic aerosol
The objective of this study was to examine the spatial and temporal patterns
in pollutants impacting the air quality in Oakland through mobile sampling in
an instrumented van. Organic aerosol (OA) contributes the largest fraction
( Organic aerosol is the dominant component of PM In downtown, concentrations of primary OA are higher than secondary OA. The
dominant source of these primary OA emissions shifts diurnally between
cooking and vehicles: pre-10:00 LT fresh emissions are from vehicles, but
cooking emissions contribute dominantly to OA after lunchtime. While it is challenging to mathematically apportion traffic-emitted OA
between drayage trucks and cars, we use ratios of OA and black carbon (BC;
particulate matter typically emitted from diesel combustion in trucks) to CO
and show that drayage truck emissions have an important effect on the
concentrations of OA and BC at the port, especially when truck traffic
typically peaks in the afternoon. However, cars seem to be the dominant
source of traffic emissions in downtown and West Oakland. Secondary OA (SOA) also exhibits spatial variability similar to primary OA.
LO-OOA concentrations are higher in downtown, likely because a combination of
various factors (poor ventilation of air masses in street canyons, higher
emissions of SOA precursors, higher OH concentrations) results in downtown
being a microenvironment with high photochemical activity. The overall chemical composition of OA in Oakland is more chemically reduced
relative to that reported in studies in several other locations.
HOA in Oakland is particularly more reduced than other locations. This reflects the
importance of primary emissions in Oakland. Further, by comparing
measurements from other studies, we show that the chemical composition of HOA
has likely become more reduced over the last decade. Lastly, external mixing
of air masses with contrasting pollutant concentrations plays an important
role in the processing of these chemically reduced emissions.
These findings have important implications for population exposure studies
because urban downtowns tend to have a concentrated presence of workplaces,
resulting in people being exposed to these elevated pollutant concentrations
for
Raw data may be accessed from our online repository
The supplement related to this article is available online at:
RUS, ESR, and PG collected data. RUS performed data analysis with input from all coauthors on the interpretation of results. RUS and AAP wrote the paper with significant input from ALR. ALR, JSA, and AAP designed the research.
The authors declare that they have no conflict of interest.
This work has not been formally reviewed by the funding agencies. The views expressed in this document are solely those of the authors and do not necessarily reflect those of the funding agencies. EPA does not endorse any products or commercial services mentioned in this publication.
This research was funded by Environmental Defense Fund (EDF) and NSF grant number AGS1543786. This publication was developed under assistance agreement no. RD83587301 awarded by the U.S. Environmental Protection Agency. We thank Sarah Seraj (UT Austin) for her assistance with data collection and Thomas Kirchstetter, Chelsea Preble, and Julien Caubel (Lawrence Berkeley National Lab) for assistance with calibration instrumentation. Edited by: Rupert Holzinger Reviewed by: two anonymous referees