1. Detecting regional and point sources of methane from space
Daniel J. Jacob
with Bram Maasakkers, Daniel Varon

2. Challenges and blessings in detecting methane from space
Emissions are widely distributed – mix of area and point sources
Prior knowledge of emission patterns is limited
Some sources are transient (“super-emitters”)
Enhancements from sources are small relative to background
BLESSINGS:
Methane is relatively easy to measure from space
Lack of sensitivity to vertical gradient in SWIR (full column measurement)
Emissions have no diurnal variation and most are steady

3. Example of the Barnett Shale
Regional: Q = 72 tons CH4 h-1 in 300x300 km2 area
Super-emitter point sources: Q = 1.0 tons CH4 h-1 (may be transient)
Lyon et al. [2015]

4. A simple mass balance approach to assess detectability
Regional/area source
Regional or point source Q [kg h-1]
Background
column
mole fraction Xo
Xo + X
wind speed U
Instrument precision: Q
W = source region dimension
Point source
Measurement requirements:
for detection: σ < X/2
for quantification:σ < X/5 Xo
Xo + X
wind speed U Q
W = satellite pixel dimension

5. Averaging time, single-pass detection
Time needed for detecting/quantifying a source characterized by ΔX:
tR = observation return time (time between passes)
F = fraction of successful retrievals (0.17 for GOSAT, 0.09 for SCIAMACHY)
N = number of views per pass
q = 2 for detection, 5 for quantification
Minimum point source size Qmin for single-pass detection:

6. SWIR instruments for observing methane
Agency
Data period
Pixel size [km2]
Coverage
Precision
Low Earth Orbit
Solar backscatter
SCIAMACHY
ESA
3060
6 days
1.5 %
GOSAT
JAXA
2009-
1010
3 days (sparse)
0.7 %
TROPOMI
2016-
77
1 day
0.6%
GHGSat
GHGSat, Inc.
0.05x0.05
targets
1-5%
GOSAT-2
2018-
10x10
0.4%
CarbonSat
proposed
22
5-10 days
Active (lidar)
MERLIN
DLR/CNES
2020-
5050
along track
1.0%
Geostationary
GEO-CAPE
NASA
44
hourly
geoCARB
45
2-8 hours

7. Observability of regional/point sources from space
Regional example is the Barnett Shale
Instrument
Regional source quantification
(Q =72 tons h-1 over 300300 km2)
Single-pass point source detection threshold
(Qmin , tons h-1)
SCIAMACHY
1 year averaging time 7
GOSAT
TROPOMI
single pass 4
GHGSat NA
0.2
GOSAT-2
4 months averaging time
CarbonSat
0.8
GEO-CAPE/geoCARB
1 hour
“Super-emitter” point sources observed in Barnett Shale are ~ 1.0 tons h-1

8. PDFs of 0.1ox0.1o and point sources of methane in the US
100
100 10 10
TROPOMI (single pass) 1
Emission [tons h-1]
TROPOMI (year) 1
0.1
GHGSat (single pass)
0.01
0.1
Cumulative probability distribution function

9. How about observing downwind plume from point source?
Time-averaged
Gaussian plume
(top view)
Increase # of pixels but weaker signal
Integrate across plume cross-section
More work is needed to understand information available y
Instantaneous
plume
pixels
For pixel sizes < 1km, plume structure affects the retrieval (“plume shadow”): IB Io
Air mass factor for column retrieval depends on plume structure
plume
(front view)

10. Building an OSSE testbed for GHGSat
23×23 m2 pixels, 12×12 km2 imaging grid
Investigate ability to quantify point source emissions from plume data including
Instrument noise, optics smoothing function
Plume shadow effect
Uncertainty on plume transport, dispersion
Gaussian plume as would be
observed by GHGSat (noise to be added)
Varon et al.,
In progress

11. Building a North American methane monitoring system
2016 satellite launches: TROPOMI global daily mapping with 77 km2 pixels
GHGSat targeted sampling with 5050 m2 pixels
Integrate satellite data with surface, aircraft observations
INTEX-A
SEAC4RS
CalNex
EPA national inventory
Improved understanding of emissions to serve climate policy

12. Working with IBM (ARPA-E project): monitoring of emissons from oil/gas fields
How can we best combine
surface and satellite data
to monitor emissions at device level.
detect super-emitters?
IBM surface monitors
Oil/gas production field