ATMOSPHERIC CHEMISTRYNocturnal survival of isoprene linked to formation of
upper tropospheric organic aerosol
Paul I. Palmer1,2*†, Margaret R. Marvin1,2†, Richard Siddans3,4, Brian J. Kerridge3,4, David P. Moore5,6
Isoprene is emitted mainly by terrestrial vegetation and is the dominant volatile organic compound
(VOC) in EarthÕs atmosphere. It plays key roles in determining the oxidizing capacity of the troposphere
and the formation of organic aerosol. Daytime infrared satellite observations of isoprene reported
here broadly agree with emission inventories, but we found substantial differences in the locations
and magnitudes of isoprene hotspots, consistent with a recent study. The corresponding nighttime
infrared observations reveal unexpected hotspots over tropical South America, the Congo basin,
and Southeast Asia. We used an atmospheric chemistry model to link these nighttime isoprene
measurements to low-NOxregions with high biogenic VOC emissions; at sunrise the remaining isoprene
can lead to the production of epoxydiols and subsequently to the widespread seasonal production of
organic aerosol in the tropical upper troposphere.
T
he main source of atmospheric isoprene
[CH 2 =C(CH 3 )-CH=CH 2 ] is terrestrial
plants, including some mosses, ferns,
gymnosperms, and angiosperms ( 1 ). Iso-
prene is emitted from leaves, with the
emission rate dependent mostly on temper-
ature and photosynthetically active radia-
tion (PAR). Only at chronic levels of water
stress and temperature do isoprene emissions
cease; even after this stress is alleviated, the
emissions resume, sometimes at rates higher
than before the stress began ( 2 ). The main loss
of atmospheric isoprene is oxidation by the
hydroxyl radical (OH), resulting in a typical
lifetime of 1 hour, which varies according to
the photochemical environment. The impor-
tance of isoprene emissions lies in the resulting
atmospheric chemistry. Inventories estimate
the global annual mean flux of isoprene to be
~500 Tg C ( 3 ) but with a large uncertainty that
reflects individual model assumptions ( 4 ) and
the sparsity of measurements that underpin
inventories. Tropical ecosystems represent
~80% of this global total ( 3 ), although esti-
mates inferred from satellite observations sug-
gest that this fraction might be overestimated
by 30% ( 5 ). The fate of isoprene oxidation
products depends on the relative abundances
of nitrogen oxides (NOx=NO+NO 2 ) and
hydroperoxy (and organic peroxy) radicals.
Broadly, at comparatively high NOxlevels,
the chemistry results in the rapid productionof formaldehyde (HCHO) that is now routinely
observed by Earth-observing satellites and can
be used to infer the emissions of the parent
hydrocarbons, predominantly isoprene ( 6 – 8 ).
At comparatively low levels of NOx, isoprene
peroxy radicals react with hydroperoxy radi-
cals to form an organic peroxide, and to a lesser
extent they can undergo unimolecular isom-
erization reactions. The peroxides can then
be further oxidized by OH (lifetime of ~3 to
5 hours) to form isoprene epoxydiols (IEPOX)
( 9 ). The atmospheric lifetime of IEPOX against
oxidation by OH is ~20 to 30 hours, during
which time it can partition into the particle
phase, either by condensation or reactive
uptake by preexisting particles, to form IEPOX
secondary organic aerosol (SOA) ( 10 – 13 ).
Recent studies have reported isoprene re-
trievals from highly spectrally resolved infrared
(IR) data from the Cross-track Infrared Sounder
(CrIS) aboard the NASA/NOAA Suomi-NPP
satellite ( 14 ); these data have been used to
revise isoprene emission inventories ( 15 ).
The satellite was launched in 2011 into a Sun-
synchronous orbit with equator-crossing local
times of 0130 and 1330. CrIS measures IR
radiation in three bands that span 650 to562 4 FEBRUARY 2022•VOL 375 ISSUE 6580 science.orgSCIENCE
(^1) National Centre for Earth Observation, University of
Edinburgh, Edinburgh, UK.^2 School of GeoSciences, University
of Edinburgh, Edinburgh, UK.^3 National Centre for Earth
Observation, STFC Rutherford Appleton Laboratory, Chilton,
UK.^4 Remote Sensing Group, STFC Rutherford Appleton
Laboratory, Chilton, UK.^5 National Centre for Earth
Observation, University of Leicester, Leicester, UK.^6 School of
Physics and Astronomy, University of Leicester, Leicester, UK.
*Corresponding author. Email: [email protected]
These authors contributed equally to this work.
A
C
E
G
B
D
F
H
Fig. 1. Monthly CrIS and GEOS-Chem model distributions of nighttime
(local equatorial overpass time 0130) effective isoprene columns.
(AandB) April 2018. (CandD) July 2018. (EandF) September 2018.
(GandH) December 2018. The magnitudes of these effective columns depend
on how the isoprene vertical profile is represented in the retrieval and on
scene-dependent averaging kernels ( 16 ). The GEOS-Chem model is sampled at
the time and location of each CrIS measurement and is convolved with the
scene-dependent averaging kernel. Satellite data and model output are filtered to
exclude scenes where co-retrieved aerosol optical thickness (AOT) > 0.05 and
effective cloud fraction (= cloud fraction × cloud top height) > 4 km. Black
squares over tropical South America and tropical Africa denote regions used in
Figs. 2 and 3 and the corresponding plots over tropical Africa ( 16 ).
RESEARCH | REPORTS