of 9
Atmospheric Methane Emissions Correlate With Natural
Gas Consumption From Residential and Commercial
Sectors in Los Angeles
Liyin He
1
, Zhao
Cheng Zeng
1
, Thomas J. Pongetti
2
, Clare Wong
1,3
, Jianming Liang
4
,
Kevin R. Gurney
4
, Sally Newman
1,5
, Vineet Yadav
2
, Kristal Verhulst
2
,
Charles E. Miller
2
, Riley Duren
2
, Christian Frankenberg
1,2
, Paul O. Wennberg
1,6
,
Run
Lie Shia
1
, Yuk L. Yung
1,2
, and Stanley P. Sander
1,2
1
Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, CA, USA,
2
Jet Propulsion
Laboratory, California Institute of Technology, Pasadena, CA, USA,
3
Now at Of
fi
ce of Institutional Research, California
State University, Northridge, CA, USA,
4
School of Informatics, Computing, and Cyber Systems, Northern Arizona
University, Flagstaff, AZ, USA,
5
Now at Bay Area Air Quality Management District, San Francisco, CA, USA,
6
Division of
Engineering and Applied Science, California Institute of Technology, Pasadena, CA, USA
Abstract
Legislation in the State of California mandates reductions in emissions of short
lived climate
pollutants of 40% from 2013 levels by 2030 for CH
4
. Identi
fi
cation of the sector(s) responsible for these
emissions and their temporal and spatial variability is a key step in achieving these goals. Here, we
determine the emissions of CH
4
in Los Angeles from 2011
2017 using a mountaintop remote sensing
mapping spectrometer. We show that the pattern of CH
4
emissions contains both seasonal and nonseasonal
contributions. We
fi
nd that the seasonal component peaks in the winter and is correlated (
R
2
= 0.58) with
utility natural gas consumption from the residential and commercial sectors and not from the industrial and
gas
fi
red power plant sectors. The nonseasonal component is (22.9 ± 1.4) Gg CH
4
/month. If the seasonal
correlation is causal, about (1.4 ± 0.1)% of the commercial and residential natural gas consumption in Los
Angeles is released into the atmosphere.
Plain Language Summary
CH
4
is a desirable target for greenhouse gas emission reductions
because emission controls will have a rapid impact on radiative forcing. However, its emission budget is
highly uncertain and poorly quanti
fi
ed. This paper reports new results from a novel mountaintop remote
sensing spectrometer overlooking the Los Angeles basin. The study shows that the megacity's methane
emissions are strongly correlated with the consumption of natural gas by residential and commercial
consumers, with a leakage rate of (1.4 ± 0.1)%, while the nonseasonal component is (22.9 ± 1.4) Gg CH
4
/
month. By identifying a clear relationship between CH
4
emissions and natural gas consumption, our results
provide strong constraints on the pathways for fugitive CH
4
emissions from the natural gas distribution
system in Los Angeles.
1. Introduction
CH
4
accounts for about 25% of the change in radiative forcing total from increases in the well
mixed green-
house gases (GHGs) since the preindustrial era (Etminan et al., 2016). With an atmospheric lifetime of only
about 10 years, CH
4
is a desirable target for GHG emission reductions because emission controls will have a
relatively rapid impact on radiative forcing. However, emissions reduction strategies must be tailored to
address speci
fi
c sources, which include land
fi
lls, livestock, wastewater treatment, and fugitive emissions
from natural gas storage, distribution, and end use equipment.
Legislation in the State of California mandates reductions in emissions of short
lived climate pollutants of
40% from 2013 levels by 2030 for CH
4
(California Legislature, 2006). In the Los Angeles (LA) basin and other
urban areas, previous studies focused on methane source attribution have used a variety of methods includ-
ing C
2
H
6
as a tracer for fossil methane, and measurements of CH
4
isotopologues and VOCs to distinguish fos-
sil and biogenic sources (Hopkins et al., 2016; Kuwayama et al., 2019; Wennberg et al., 2012; Wunch et al.,
2016). These studies indicate that fugitive natural gas emissions account for 56
70% of the difference between
©2019. American Geophysical Union.
All Rights Reserved.
RESEARCH LETTER
10.1029/2019GL083400
Liyin He and Zhao
Cheng Zeng con-
tributed equally.
Key Points:
A mountaintop remote sensing
spectrometer is used to derive the
time series and spatial pattern of
methane emissions in LA basin
The methane emissions in the LA
basin are strongly correlated with
the consumption of natural gas by
residential and commercial
consumers
About (1.4 ± 0.1)% of the residential
and commercial natural gas
consumption in LA is released into
the atmosphere
Supporting Information:
Supporting Information S1
Correspondence to:
S. P. Sander,
ssander@jpl.nasa.gov
Citation:
He, L., Zeng, Z.
C., Pongetti, T., Wong,
C., Liang, J., Gurney, K. R., et al. (2019).
Atmospheric methane emissions
correlate with natural gas consumption
from residential and commercial
sectors in Los Angeles.
Geophysical
Research Letters
,
46
. https://doi.org/
10.1029/2019GL083400
Received 19 APR 2019
Accepted 10 JUL 2019
Accepted article online 15 JUL 2019
HE ET AL.
1
annual top
down and bottom
up (annual excess) CH
4
emissions in LA (Hopkins et al., 2016; Peischl et al.,
2013; Wennberg et al., 2012).
Almost all previous studies were restricted in spatial and/or temporal coverage, and none were able to
determine which segments and operations of the natural gas distribution system were responsible for
the leakage. Wennberg et al. (2012) proposed that many small leaks downstream of the gas meters could
be responsible rather than the transport, storage, and distribution segments. Identifying and quantifying
the sources of these emissions is critical because methane budgets vary between urban areas, so under-
standing the emission pathways is essential for mitigation (Jeong et al., 2017; Lamb et al., 2016;
McKain et al., 2015).
Recently, we demonstrated a novel remote sensing technique to measure daytime CH
4
emissions in LA from
atop Mt Wilson, a mountaintop vantage point (~1,700
m elevation) with nearly unobstructed views of the
South Coast Air Basin (SOCAB; Wong et al., 2016). In this method, high
resolution near
infrared spectra
are recorded as solar radiation passes through the atmosphere and re
fl
ected by the land surface toward
the observatory. The instrument points to a series of locations in the SOCAB, providing maps every 90
min of trace gas slant column abundances with a spatial resolution of a few kilometers. Using the tracer
tracer correlation method combined with a highly resolved CO
2
emissions inventory, Wong et al. (2016)
showed that it was possible to identify seasonal peaks in spatially aggregated CH
4
emissions in LA. CH
4
emission peaks up to 37 Gg/month were consistently observed in the winter seasons, with a low of 27
Gg/month in the summer. These levels were revised upward in the present study as a result of changes to
the underlying CO
2
inventory. Overall, the measured SOCAB CH
4
emissions were 2
31% higher than the
scaled statewide bottom
up emissions estimated by the California Air Resources Board (CARB) from
2011
2013 averaged over the three years (CARB, 2011). This result is consistent with other studies that
obtained larger emissions than CARB estimates (Conley et al., 2016; Cui et al., 2017; Hedelius et al., 2018;
Hsu et al., 2010; Peischl et al., 2013; Wennberg et al., 2012; Wong et al., 2015; Wong et al., 2016; Wunch
et al., 2009; Wunch et al., 2016; Yadav et al., 2019). No seasonal variability has been observed in other inten-
sively monitored cities including Indianapolis and Boston although these cities differ signi
fi
cantly from LA
in topography, meteorology, infrastructure, and other factors that in
fl
uence methane emissions (Lamb et al.,
2016; McKain et al., 2015).
In the present study, we resolve seasonal and spatial variability of CH
4
emissions from 2011 to 2017 and
regress it against consumption data as an important step toward reconciling California's methane budget.
The goal of this paper is
fi
rst to leverage our powerful 2011
2017 data record to quantify seasonal to inter-
annual variability in LA CH
4
emissions. Second, we investigated whether the seasonality of SOCAB CH
4
emissions is related to natural gas consumption. Finally, we quanti
fi
ed the relative contribution of each sec-
tor (including residential, commercial, industrial, vehicle, and power plant) to the seasonality of SOCAB
CH
4
emissions.
2. Methodology and Data
2.1. Methodology
2.1.1. Observations From CLARS
This study uses GHG slant column abundance data acquired by a Jet Propulsion Laboratory
built Fourier
transform spectrometer (FTS) located at the California Laboratory for Atmospheric Remote Sensing
(CLARS) on Mt. Wilson overlooking the LA basin at an altitude of 1,673 m above sea level. The instrument
design, operating parameters, retrieval algorithms, and measurement approach and performance are dis-
cussed in detail in Fu et al. (2014) and Wong et al. (2015, 2016). The daytime measurements have been con-
tinuously acquired by CLARS since September 2011. Brie
fl
y, CLARS
FTS operates in two observation modes
(see supporting information Figure S1): Spectralon Viewing Observations (SVO) and Los Angeles Basin
Surveys (LABS). In SVO mode, the FTS points at a Spectralon® plate immediately adjacent to CLARS
FTS.
The spectrum is equivalent to a direct solar occultation spectrum in the absence of high clouds. Since
CLARS is above the planetary boundary layer (PBL; Ware et al., 2016), the SVO spectra are unaffected by
PBL pollution and therefore provide an approximation to background trace gas column densities. In
LABS mode, the FTS points sequentially at the 33 surface target sites (see supporting information Figure
S2), which span most of the LA Basin. In this mode the re
fl
ected radiance traces several kilometers in the
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
2
PBL and is therefore very sensitive to GHG emissions. There are
fi
ve to eight standard measurement cycles
per day. In each standard measurement cycle, the FTS observes 33 surface targets and acquires four inter-
spersed SVO measurements, pointing at each target for about 3 min. The sampling time of CLARS depends
on the length of the daytime. During summer, CLARS performs eight measurement cycles from ~6 a.m. to ~8
p.m., while during winter, CLARS operates
fi
ve measurement cycles from ~8 a.m. to ~5 p.m.
Slant column densities (SCDs), the total number of molecules per unit area along the path, of trace gases
CO
2
,CH
4
, and O
2
are retrieved by
fi
tting spectral lines at 1.60, 1.67, and 1.27
μ
m, respectively (Fu et al.,
2014). The slant column averaged dry air mixing ratios of CO
2
(XCO
2
) and CH
4
(XCH
4
) are calculated as
follows:
XCO
2
¼
SCD
CO2
SCD
O2
×XO
2
(1)
XCH
4
¼
SCD
CH4
SCD
O2
×XO
2
(2)
where XO
2
is the mixing ratio of oxygen in air (0.2095).
2.1.2. Monthly Ratio of Excess XCH
4
to Excess XCO
2
As discussed by Wong et al. (2015), data
fi
ltering was applied to remove measurements with high aerosol
scattering impact, low signal
to
noise ratio, and high solar zenith angles. In addition, measurements were
removed when the spectra
fi
tting residuals exceed a prede
fi
ned threshold (Fu et al., 2014; Wong et al.,
2015). Excess XCH
4
(XCH
4
xs) and excess XCO
2
(XCO
2
xs) are obtained by subtracting the background
XCH
4
and XCO
2
acquired using SVO mode from those acquired using LABS mode, respectively (Wong
et al., 2016):
XCO
2xs
¼
X
CO
2LABS
XCO
2SVO
(3)
XCH
4xs
¼
XCH
4LABS
XCH
4SVO
(4)
Wong et al. (2016) used orthogonal distance regression analysis to quantify monthly XCH
4
xs/XCO
2
xs corre-
lation slopes for each surface target and then used the weighted average slope across all targets to represent
the monthly XCH
4
xs/XCO
2
xs ratio over the LA Basin. In the orthogonal distance regression analysis, we
found that data anomalies may bias the estimated correlation slope, which is important for the determina-
tion of emission ratios.
In this work, we derived an unbiased background of XCH
4
and XCO
2
along the same path as CLARS
target mode using XCO
2SVO
and XCH
4SVO
combined with National Oceanic and Atmospheric
Administration in situ
fl
ask measurement on Mt. Wilson (see supporting information S1). Also, we used
a different approach to derive monthly XCH
4
xs/XCO
2
xs, which is found to better capture the mean pat-
tern as well as the anomalies. To derive monthly XCH
4
xs/XCO
2
xs, daily XCH
4
xs/XCO
2
xs over the LA
Basin is
fi
rst obtained by averaging individual XCH
4
xs/XCO
2
xs measurements for each surface target.
Measurements with XCO
2
xs less than 2 ppm are removed to con
fi
ne the analysis to emissions within
the basin, as in Wunch et al. (2009). The time series of XCH
4
xs to XCO
2
xs ratio including all measure-
ments from CLARS
FTS over all surface targets in the LA basin from 2011 to 2017 is shown in support-
ing information Figure S3. Monthly XCH
4
xs/XCO
2
xs can then be computed as the average of the daily
ratios over a given month. The corresponding standard error for each month is also calculated.
Supporting information Figure S4 shows the seasonal pattern of XCH
4
xs/XCO
2
xs ratios and the
associated errors.
2.1.3. Deriving Top
Down Monthly CH
4
Emissions
Following Wong et al. (2015, 2016), we used the tracer
tracer method (Wunch et al., 2009) to derive the
monthly CH
4
emissions based on the estimated monthly XCH
4
xs/XCO
2
xs ratio in the LA basin:
E
CH4
j
top
down
monthly
¼
XCH
4xs
XCO
2xs
CLARS
monthly
×
E
CO2
inventory
monthly
×
MW
CH4
MW
CO2
(5)
where
E
CO2
j
inventory
monthly
is CO
2
inventory emissions in the LA Basin. We used the Hestia V2.5 data set in this
paper, assuming 10% uncertainty in the inventory data set Gurney et al. (2012) and Gurney et al. (2019)
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
3
(more details in supporting information S2);
MW
CH4
MW
CO2
is the ratio of the molecular mass of CH
4
and CO
2
. This
tracer
tracer inversion method is built on the strong correlations between CH
4
and CO
2
measured in the PBL
in source regions. The XCH
4
xs/XCO
2
xs ratio has been identi
fi
ed with local emission ratios for the two gases
(Wennberg et al., 2012; Wunch et al., 2009). Moreover, the effects of aerosol scattering on the XCH
4
xs and
XCO
2
xs largely cancel out and the impact on their ratio is assumed to be negligible (Zhang et al., 2015).
This tracer
tracer method in deriving CH
4
emissions is based on a number of assumptions as discussed in
detail by Wong et al. (2016).
2.2. Relative Importance of Different Natural Gas Consumption Sectors in CH
4
Emissions
To determine the relative importance of different natural gas consumption sectors to the monthly CH
4
emissions, we applied a multiple linear regression model (Grömping, 2006) with the averaging over order-
ings method proposed by Lindeman (1980). The relative importance is obtained by decomposing the
model
explained variance into a
fi
xed number of components associated with the contribution of each
predictor. In practice, the R package of
relaimpo
(Grömping, 2006) is used to quantify the relative
importance of each predictor in the regression model. This method has been widely used in environmen-
tal studies (Arrigo et al., 2015; Jones et al., 2014; Wu et al., 2013). In our case, the predictors are the nat-
ural gas consumption from several individual sectors, including residential, commercial, industrial,
vehicle, and electric power. The dependent variable is the total monthly CH
4
emissions inferred from
CLARS data.
2.3. Data
2.3.1. Monthly Natural Gas Consumption Data Set
Natural gas usage data are the sum of natural gas usage data from residential, commercial, industrial, vehi-
cle, and power plant sectors in the SOCAB. The residential, commercial, and industrial data are available
publicly on Southern California Gas Company (SoCalGas) database (SoCalGas, 2018). Power plant data
are provided by the California Energy Commission online database (California Energy Commission,
2018). The time series of the natural gas consumption of individual sectors are shown in supporting informa-
tion Figure S7.
2.3.2. Monthly Average Surface Air Temperature
The monthly average surface air temperature data in Los Angeles Downtown/USC, CA (171 meters above
sea level) are obtained from the stational data inventory in the National Oceanic and Atmospheric
Administration/National Weather Service Cooperative Observer Network (https://wrcc.dri.edu/cgi
bin/
cliMAIN.pl?ca5115). All months from September 2011 to December 2016 are used. The surface air tempera-
ture in this study is the temperature of the free air conditions surrounding the station at a height between 1
and 2 m above ground level. The air should be freely exposed to sunlight and wind. It is not close to or
Figure 1.
Monthly CH
4
emissions in the Los Angeles basin from September 2011 to December 2017. The monthly average
over the measurement period is shown in black. Uncertainties are estimated by error propagation using the uncertainties
of monthly patterns of XCH
4
xs
XCO2xs (supporting information Figure S4) and the assumption of 10% uncertainties in
Hestia CO
2
inventory data (supporting information Figure S9). SOCAB = South Coast Air Basin.
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
4
shielded by trees, buildings, or other obstructions. The temperature data from the observatory are averaged
for every 15 s and then averaged to the daily and monthly data.
3. Results
Figure 1 shows the monthly CH
4
emissions in the LA basin from September 2011 to December 2017 along
with the monthly average for the observing period overlaid year by year. LA CH
4
emissions exhibit a consis-
tent seasonal pattern, ranging from a minimum of ~27 Gg/month in June
July to a maximum of ~45
Gg/month in December
January. We de
fi
ne the observed difference
between measured winter and summer CH
4
emissions as the
seasonal
excess
to distinguish it from the annual excess emissions de
fi
ned above.
A spike in emissions in November 2015 coincides with the period of max-
imum emissions from a very large natural gas storage well blowout at
Aliso Canyon that impacted the entire LA basin (Conley et al., 2016).
Figure 2 shows the data from Figure 1 represented as a continuous time
series, illustrating the prominent winter emissions maxima.
As discussed above, multiple previous studies have identi
fi
ed fugitive
emissions from natural gas infrastructure as a likely contributor to the
observed SOCAB annual excess CH
4
emissions. Figure 2 compares our
CH
4
emissions data and monthly natural gas consumption in the
SOCAB from the residential, commercial, and industrial sectors as pro-
vided by the utility company (see section 2.3). The natural gas consump-
tion data are based on metered customer usage. While the usage data do
not include contributions from transmission line leaks, compressor sta-
tions, blowdowns,
fl
aring events, and other sources upstream of customer
meters, these sources may correlate with metered usage. Figure 2 shows
that the observed December
January peaks in monthly CH
4
emissions
closely track natural gas consumption.
Figure 3 shows the weighted linear least squares
fi
t between derived
monthly CH
4
emissions and natural gas consumption in the SOCAB.
The black line is the linear regression weighted by the uncertainties in
the derived CH
4
emissions and uncertainties in the consumption data
Figure 2.
SOCAB CH
4
emissions expressed as a continuous time series (black line, left axis). Monthly natural gas consumption data in the Los Angeles basin from
the residential, commercial, and industrial sectors (red dashed line, right axis). The natural gas consumption from the power plant sector does not s
how signi
fi
cant
seasonal variability (supporting information Figure S7). Emissions data from November
December 2015 are impacted by the Aliso Canyon natural gas storage well
blowout. SOCAB = South Coast Air Basin.
Figure 3.
Correlation between derived monthly California Laboratory for
Atmospheric Remote Sensing methane emissions and monthly natural gas
consumption (residential, commercial and industrial) in the SOCAB from
September 2011 to August 2017. Points are color coded by season illustrating
the progressive increase in emissions from summer (red) to winter (blue).
The correlation that includes power plant consumption is shown in sup-
porting information Figure S12. SOCAB = South Coast Air Basin
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
5
are assumed to be ±10%. The regression slope is 0.0140 ± 0.0014, and
the
y
intercept is 22.9 ± 1.1 Gg/month with
R
2
= 0.58. We interpret
the slope as the fraction of the post
meter natural gas consumption
emitted into the atmosphere (from both appliance use and static
leaks) while the
y
intercept gives the CH
4
emissions extrapolated to
zero metered consumption, that is, associated with non
metered
emissions. The latter would include emissions from land
fi
lls, waste-
water treatment, local geological sources, and natural gas transmis-
sion lines and mains. These sources may have their own seasonality
that this simple two
parameter model cannot capture. The fraction
of emissions to consumption derived here, (1.4 ± 0.14)%, is somewhat
smaller than the range 2.5
6% estimated previously (Wennberg
et al., 2012).
The correlation between natural gas consumption and CH
4
emissions
may be due to increased wintertime demand by appliances for space
heating, water heating, cooking, and other purposes that involve heat
generation. To gain further insight into the source sectors responsible
for this correlation, we use data on natural gas consumption classi-
fi
ed by end use in California from the local utility, SoCalGas, and
the California Energy Commission. Monthly data are available for
fi
ve sectors: residential, commercial, industrial, vehicle fuel, and elec-
tric power (see section 2.3 and supporting information Figure S7).
Residential and commercial consumption both peak in the winter
months, industrial consumption shows small peaks in the summer,
while electric power consumption peaks strongly in the late summer
(August
September). Consumption by the transportation sector is
only a few percent of the total and is not considered. Peaks in indus-
trial consumption are less pronounced and out of phase with
residential/commercial usage. A multivariate correlation analysis
shows that U.S. Energy Information Administration data for natural
gas consumption from the sum of the residential and commercial sec-
tors correlates well (correlation coef
fi
cient,
R
2
= 0.89), while the cor-
relations
between
industrial
consumption
and
residential/commercial consumption are less evident (
R
2
= 0.23 and
0.29, respectively).
To quantify the sectoral contributions, the regression equation is given by
E
CH4
top
down
monthly
¼
a
0
þ
a
1
NG
residential
monthly
þ
NG
commercial
monthly

þ
a
2
×
NG
industrial
monthly
(6)
where
E
CH4
=
total monthly excess CH
4
emissions inferred from CLARS data (Gg)
NG
i
= reported monthly sectoral natural gas consumption (Gg)
a
i
= regression coef
fi
cients
The best
fi
t regression coef
fi
cients are 27.37, 0.0156, and 0.0094 for
a
0
,
a
1
, and
a
2
, respectively. About 40.4%
of the variance between the model and observations is explained by the sum of residential and commercial
consumption, and 9.1% is explained by industrial consumption. The results from the regression modeling
indicate that there is a strong connection between CH
4
emissions into the atmosphere and
residential/commercial natural gas consumption based on time series analysis of both data sets. Taking
a
0
as the background excess methane emission in the LA basin, we see that the seasonal component results
in a doubling of the total emissions relative to the background. Note that the pattern of emissions must have
a seasonal component in order to explain the observations. Quiescent emissions (persistent leaks) from
equipment and plumbing cannot explain the strong seasonal signal.
Figure 4.
(upper panel) The correlation between monthly natural gas consump-
tion and inverse monthly mean temperature at USC/LA Downtown. (lower
panel) The correlation between California Laboratory for Atmospheric Remote
Sensing inferred monthly CH
4
emissions and inverse monthly mean temperature
at USC/LA Downtown. Increased natural gas consumption for space and water
heating at lower ambient temperatures may provide the link to higher observed
CH
4
emissions. SOCAB = South Coast Air Basin.
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
6
There have been few long
term studies in the SOCAB of CH
4
emissions from the most important sources
(natural gas infrastructure, postmeter equipment, land
fi
lls, and wastewater treatment plants), providing
weak evidence for seasonal variability from these sources (Wong et al., 2016). Only postmeter consumption
mimics the observed CH
4
emissions pattern (Wong et al., 2016). Figure 4 shows clear correlations between
the inverse of the ambient temperature measured near downtown LA and both natural gas consumption and
CH
4
emission rates. Reduced surface air temperatures drive air and water heating demands, resulting in the
expected increase in observed CH
4
emissions with decreasing surface temperature (see section 2.3). These
observations reinforce the connections between ambient temperature, heating demand and fugitive natural
gas emissions. Figure 4 also provides some insight for considering the temperature as an important variable
linking natural gas consumptions and CH
4
emissions.
4. Discussion
Since there are no national air quality standards for methane, very little work has been done to characterize
the methane emission factors from natural gas
fi
red appliances such as furnaces, water heaters, stoves,
ovens, swimming pool and spa heaters and similar equipment. Currently, the only available emission factor
for CH
4
from natural gas
fi
red furnaces is 5 g/GJ for both commercial and residential furnaces (U.S.
Environmental Protection Agency, 2018). From Figure 2, SOCAB winter natural gas consumption surpasses
1,000 Gg/month. Using the Environmental Protection Agency emission factor, this would result in about
0.29
Gg/month seasonal excess methane emissions, which is far less than the observed value of
~20 Gg/month.
There are a number of factors that may close the gap between top
down and bottom
up estimates of seasonal
excess methane emissions. Far more research needs to be conducted on emission factors from gas
fi
red
appliances and industrial combustors under different operating conditions (start
up, operation, and shut
down). While increased demand for space heating is clearly associated with lower ambient temperatures
in the winter, water heating demand also increases because of decreases in supply water temperature. For
example, in the mild, Mediterranean climate of Pasadena, California, measurements from 2001
2016 at
six locations showed an average difference of 12 °C in supply water temperature from winter to summer
(Kimbrough, 2017). Signi
fi
cantly larger seasonal temperature variability would be expected in colder cli-
mates. There is increasing evidence that the probability density functions for CH
4
emissions have a long tail,
characterized by a small number of emitters with very large emissions, perhaps due to malfunctioning equip-
ment or improper operating conditions (Zavala
Araiza et al., 2015). This will require a concerted measure-
ment campaign examining large numbers of emitters (thousands) under actual operating conditions
targeting residential, commercial, and industrial sectors (Fischer et al., 2018).
5. Conclusions
In conclusion, using mountaintop remote sensing with coverage over the greater LA basin, we observe sea-
sonal excess methane emissions that correlate very well (
R
2
= 0.58) with combined commercial and residen-
tial natural gas consumption. From the covariance we observe that the emissions arise from two terms: one
that is seasonally invariant (22.9 ± 1.1 Gg/month) and another that peaks in the colder months of the year
corresponding to (1.4 ± 0.14)% of residential plus commercial natural gas consumption. Other natural gas
consumption sectors (industrial, power plant, and transportation) either have no clear seasonal relationship
that matches the observed emissions or are too small. The available emission factor data for residential and
commercial natural gas
fi
red combustion sources fail to explain the observed emissions. Indeed, far more
work needs to be done to measure the seasonally varying probability distribution functions of emitters under
actual operating conditions.
References
Arrigo, K. R., van Dijken, G. L., & Strong, A. L. (2015). Environmental controls of marine productivity hot spots around Antarctica.
Journal
of Geophysical Research: Oceans
,
120
, 5545
5565. https://doi.org/10.1002/2015JC010888
California Air Resources Board (2011). Greenhouse gas emission inventory
Query tool for years 2000
2011. Retrieved from https://www.
arb.ca.gov/app/ghg/2000_2011/ghg_sector.php
California Energy Commission (2018). Online database for power plant statistical information. (Accessed July 23, 2018). https://www.
energy.ca.gov/almanac/electricity_data/web_qfer/Power_Plant_Statistical_Information.php
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
7
Acknowledgments
This research was supported by NIST,
CARB, and NASA. We gratefully
acknowledge discussions with M.
Fischer, G. Heath, J. Hedelius, M.
Weitz, and V. Camobreco. L.H. thanks
the Resnick Sustainability Institute at
Caltech for fellowship support. We
thank A. Andrews and E. Dlugokencky
for providing the NOAA
fl
ask
measurements at Mt. Wilson
Observatory (National Oceanic and
Atmospheric Administration, 2019).
CLARS
FTS data are available from the
NASA Megacities Project (https://
megacities.jpl.nasa.gov). The research
was carried out at the Jet Propulsion
Laboratory, California Institute of
Technology, under a contract with the
National Aeronautics and Space
Administration. L. H., Z. C. Z., T. P.,
C. W., and S. S. carried out the data
acquisition and analysis, J. L. and K. G.
provided the Hestia inventory, S. S., L.
H., and Z. C. Z. wrote the paper, and all
authors contributed to the analysis and
discussion of the results. The authors
declare no competing
fi
nancial
interests.
California Legislature (2006). Assembly Bill No. 32. http://www.leginfo.ca.gov/pub/05
06/bill/asm/ab_0001
0050/ab_32_bill_20060927_
chaptered.pdf
Conley, S., Franco, G., Faloona, I., Blake, D. R., Peischl, J., & Ryerson, T. B. (2016). Methane emissions from the 2015 Aliso Canyon blowout
in Los Angeles, CA.
Science
,
351
(6279), 1317
1320. https://doi.org/10.1126/science.aaf2348
Cui, Y. Y., Brioude, J., Angevine, W. M., Peischl, J., McKeen, S. A., Kim, S. W., et al. (2017). Top
down estimate of methane emissions in
California using a mesoscale inverse modeling technique: The San Joaquin Valley.
Journal of Geophysical Research: Atmospheres
,
122
,
3686
3699. https://doi.org/10.1002/2016JD026398
Etminan, M., Myhre, G., Highwood, E. J., & Shine, K. P. (2016). Radiative forcing of carbon dioxide, methane, and nitrous oxide: A sig-
ni
fi
cant revision of the methane radiative forcing.
Geophysical Research Letters
,
43
, 12,614
12,623. https://doi.org/10.1002/
2016GL071930
Fischer, M. L., Chan, W. R., Delp, W., Jeong, S., Rapp, V., & Zhu, Z. (2018). An estimate of natural gas methane emissions from California
Homes.
Environmental Science & Technology
,
52
(17), 10,205
10,213. https://doi.org/10.1021/acs.est.8b03217
Fu, D., Pongetti, T. J., Blavier, J. F., Crawford, T. J., Manatt, K. S., Toon, G. C., et al. (2014). Near
infrared remote sensing of Los Angeles
trace gas distributions from a mountaintop site.
Atmospheric Measurement Techniques
,
7
(3), 713
729. https://doi.org/10.5194/amt
7
713
2014
Grömping, U. (2006). Relative importance for linear regression in R: The package relaimpo.
Journal of Statistical Software
,
17
(1), 1
27.
https://doi.org/10.18637/jss.v017.i01
Gurney, K. R., Patarasuk, R., Liang, J., Song, Y., O'Keeffe, O., Rao, P., et al. (2019). The hestia fossil fuel CO
2
emissions data product for the
Los Angeles Megacity (Hestia
LA).
Earth System Science Data Discussions
, February, 1
38. https://doi.org/10.5194/essd
2018
162
Gurney, K. R., Razlivanov, I., Song, Y., Zhou, Y., Benes, B., & Abdul
Massih, M. (2012). Quanti
fi
cation of fossil fuel CO
2
emissions on the
building/street scale for a large US city.
Environmental Science & Technology
,
46
(21), 12,194
12,202. https://doi.org/10.1021/es3011282
Hedelius, J. K., Liu, J., Oda, T., Maksyutov, S., Roehl, C. M., Iraci, L. T., et al. (2018). Southern California megacity CO
2
,CH
4
, and CO
fl
ux
estimates using ground
and space
based remote sensing and a Lagrangian model.
Atmospheric Chemistry and Physics
,
18
(22),
16,271
16,291. https://doi.org/10.5194/acp
18
16271
2018
Hopkins, F. M., Kort, E. A., Bush, S. E., Ehleringer, J. R., Lai, C. T., Blake, D. R., & Randerson, J. T. (2016). Spatial patterns and source
attribution of urban methane in the Los Angeles Basin.
Journal of Geophysical Research: Atmospheres
,
121
, 2490
2507. https://doi.org/
10.1002/2015JD024429
Hsu, Y. K., VanCuren, T., Park, S., Jakober, C., Herner, J., FitzGibbon, M., et al. (2010). Methane emissions inventory veri
fi
cation in
southern California.
Atmospheric Environment
,
44
(1), 1
7. https://doi.org/10.1016/j.atmosenv.2009.10.002
Jeong, S., Cui, X., Blake, D. R., Miller, B., Montzka, S. A., Andrews, A., et al. (2017). Estimating methane emissions from biological and
fossil
fuel sources in the San Francisco Bay Area.
Geophysical Research Letters
,
44
, 486
495. https://doi.org/10.1002/2016GL071794
Jones, A. K., Jones, D. L., & Cross, P. (2014). The carbon footprint of lamb: Sources of variation and opportunities for mitigation.
Agricultural Systems
,
123
,97
107. https://doi.org/10.1016/J.AGSY.2013.09.006
Kimbrough, D. E. (2017). Local climate change in Pasadena, Calif., and the impact on stream
fl
ow.
Journal
American Water Works
Association
,
109
(10), E416
E425. https://doi.org/10.5942/jawwa.2017.109.0114
Kuwayama, T., Charrier
Klobas, J. G., Chen, Y., Vizenor, N. M., Blake, D. R., Pongetti, T., Conley, S. A., Sander, S. P., Croes, B., &
Herner, J. D. (2019). Source Apportionment of Ambient Methane Enhancements in Los Angeles, California, To Evaluate Emission
Inventory Estimates.
Environmental Science & Technology
,
53
(6), 2961
2970. https://doi.org/10.1021/acs.est.8b02307
Lamb, B. K., Cambaliza, M. O., Davis, K. J., Edburg, S. L., Ferrara, T. W., Floerchinger, C., et al. (2016). Direct and indirect measurements
and modeling of methane emissions in Indianapolis, Indiana.
Environmental Science & Technology
,
50
(16), 8910
8917. https://doi.org/
10.1021/acs.est.6b01198
Lindeman, R. H. (1980). Introduction to bivariate and multivariate analysis (No. 04; QA278, L553.).
McKain, K., Down, A., Raciti, S. M., Budney, J., Hutyra, L. R., Floerchinger, C., et al. (2015). Methane emissions from natural gas infra-
structure and use in the urban region of Boston, Massachusetts.
Proceedings of the National Academy of Sciences
,
112
(7), 1941
1946.
https://doi.org/10.1073/pnas.1416261112
National Oceanic and Atmospheric Administration (2019). Earth System Research Laboratory Global Monitoring Division (ESRL GMD)
projects at Mt. Wilson Observatory. (Accessed April 8, 2019. http://www.esrl.noaa.gov/gmd/dv/site/MWO.html
Peischl, J., Ryerson, T. B., Brioude, J., Aikin, K. C., Andrews, A. E., Atlas, E., et al. (2013). Quantifying sources of methane using light
alkanes in the Los Angeles basin, California.
Journal of Geophysical Research: Atmospheres
,
118
, 4974
4990. https://doi.org/10.1002/
jgrd.50413
Southern California Gas Company (2018). Online gas energy usage database (2018). (Accessed June 20, 2018). https://energydatarequest.
socalgas.com/
U.S. Environmental Protection Agency (2018). Inventory of U.S. greenhouse gas emissions and sinks. Report No. EPA 430
R
18
003,
Annex 2013, Washington, DC. Retrieved from https://www.epa.gov/ghgemissions/inventory
us
greenhouse
gas
emissions
and
sinks
1990
2016
Ware, J., Kort, E. A., DeCola, P., & Duren, R. (2016). Aerosol lidar observations of atmospheric mixing in Los Angeles: Climatology and
implications for greenhouse gas observations.
Journal of Geophysical Research: Atmospheres
,
121
, 9862
9878. https://doi.org/10.1002/
2016JD024953
Wennberg, P. O., Mui, W., Wunch, D., Kort, E. A., Blake, D. R., Atlas, E. L., et al. (2012). On the sources of methane to the Los Angeles
atmosphere.
Environmental Science & Technology
,
46
(17), 9282
9289. https://doi.org/10.1021/es301138y
Wong, C. K., Pongetti, T. J., Oda, T., Rao, P., Gurney, K. R., Newman, S., et al. (2016). Monthly trends of methane emissions in Los Angeles
from 2011 to 2015 inferred by CLARS
FTS observations.
Atmospheric Chemistry and Physics
,
16
(20), 13,121
13,130. https://doi.org/
10.5194/acp
16
13121
2016
Wong, K. W., Fu, D., Pongetti, T. J., Newman, S., Kort, E. A., Duren, R., et al. (2015). Mapping CH
4
:CO
2
ratios in Los Angeles with CLARS
FTS from Mount Wilson, California.
Atmospheric Chemistry and Physics
,
15
(1), 241
252. https://doi.org/10.5194/acp
15
241
2015
Wu, Z., He, H. S., Liang, Y., Cai, L., & Lewis, B. J. (2013). Determining relative contributions of vegetation and topography to burn severity
from LANDSAT imagery.
Environmental Management
,
52
(4), 821
836. https://doi.org/10.1007/s00267
013
0128
3
Wunch, D., Toon, G. C., Hedelius, J. K., Vizenor, N., Roehl, C. M., Saad, K. M., et al. (2016). Quantifying the loss of processed natural gas
within California's South Coast Air Basin using long
term measurements of ethane and methane.
Atmospheric Chemistry and Physics
,
16
(22), 14,091
14,105. https://doi.org/10.5194/acp
16
14091
2016
Wunch, D., Wennberg, P. O., Toon, G. C., Keppel
Aleks, G., & Yavin, Y. G. (2009). Emissions of greenhouse gases from a North American
megacity.
Geophysical Research Letters
,
36
, L15810. https://doi.org/10.1029/2009GL039825
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
8
Yadav, V., Duren, R., Mueller, K., Verhulst, K. R., Nehrkorn, T., Kim, J., et al. (2019). Spatio
temporally resolved methane
fl
uxes from the
Los Angeles Megacity.
Journal of Geophysical Research: Atmospheres
,
124
, 5117
5130. https://doi.org/10.1029/2018JD029224
Zavala
Araiza, D., Lyon, D., Alvarez, R. A., Palacios, V., Harriss, R., Lan, X., et al. (2015). Toward a functional de
fi
nition of methane super
emitters: Application to natural gas production sites.
Environmental Science & Technology
,
49
(13), 8167
8174. https://doi.org/10.1021/
acs.est.5b00133
Zhang, Q., Natraj, V., Li, K.-F., Shia, R.-L., Fu, D., Pongetti, T. J., et al. (2015). Accounting for aerosol scattering in the CLARS retrieval of
column averaged CO
2
mixing ratios.
Journal of Geophysical Research: Atmospheres
,
120
, 7205
7218. https://doi.org/10.1002/
2015jd023499
References From the Supporting Information
Tanaka, T., Yates, E., Iraci, L. T., Johnson, M. S., Gore, W., Tadi
ć
, J. M., et al. (2016). Two
year comparison of airborne measurements of
CO
2
and CH
4
with GOSAT at Railroad Valley, Nevada.
IEEE Transactions on Geoscience and Remote Sensing
,
54
(8), 4367
4375. https://
doi.org/10.1109/TGRS.2016.2539973
U.S. Energy Information Administration (2018). Online database for electricity. Accessed May 20, 2018. https://www.eia.gov/electricity/
data/state/
Western Regional Climate Center. Climate Historical Summaries. (2019). Online database for Los Angeles DWTN USC campus (Accessed
April 8, 2019). https://wrcc.dri.edu/cgi
bin/cliMAIN.pl?ca5115
10.1029/2019GL083400
Geophysical Research Letters
HE ET AL.
9