NO2 remote sensing from APEX

38 Pages·2012·3.13 MB·English

Atmos.Meas.Tech.Discuss.,5,2449–2486,2012 Atmospheric Dis
www.atmos-meas-tech-discuss.net/5/2449/2012/ Measurement c AMTD
u
doi:10.5194/amtd-5-2449-2012 Techniques s
s
©Author(s)2012.CCAttribution3.0License. Discussions io 5,2449–2486,2012
n
P
a
p
Thisdiscussionpaperis/hasbeenunderreviewforthejournalAtmosphericMeasurement e NO remote sensing
r 2
Techniques(AMT).PleaserefertothecorrespondingfinalpaperinAMTifavailable. from APEX
|
D C.Poppetal.
is
c
High resolution NO remote sensing from u
s
2 s
io
the Airborne Prism EXperiment (APEX) n TitlePage
P
a
p Abstract Introduction
imaging spectrometer e
r
Conclusions References
|
C. Popp1, D. Brunner1, A. Damm2, M. Van Roozendael3, C. Fayt3, and D Tables Figures
is
B. Buchmann1 c
u
ss J I
1Empa,SwissFederalLaboratoriesforMaterialsScienceandTechnology, io
n
8600Du¨bendorf,Switzerland P J I
2RemoteSensingLaboratories,UniversityofZurich,Winterthurerstrasse190, ap
e
8057Zurich,Switzerland r Back Close
3BelgianInsituteforSpaceAeronomy(BIRA-IASB),AvenueCirculaire3,
| FullScreen/Esc
1180Brussels,Belgium
D
Received: 1March2012–Accepted: 19March2012–Published: 28March2012 isc Printer-friendlyVersion
u
s
Correspondenceto: C.Popp([email protected]) s
io InteractiveDiscussion
n
PublishedbyCopernicusPublicationsonbehalfoftheEuropeanGeosciencesUnion.
P
a
p
e
r
2449 |
Abstract D
is
c AMTD
u
We present and evaluate the retrieval of high spatial resolution maps of NO vertical s
2 s
column densities (VCD) from the Airborne Prism EXperiment (APEX) imaging spec- ion 5,2449–2486,2012
trometer. APEX is a novel instrument providing airborne measurements of unique P
a
p
5 spectral and spatial resolution and coverage as well as high signal stability. In this er NO2 remote sensing
study,weusespectrometerdataacquiredoverZurich,Switzerland,inthemorningand from APEX
|
lateafternoonduringaflightcampaignonacloud-freesummerdayinJune2010. NO
2
VCD are derived with a two-step approach usually applied to satellite NO retrievals, D C.Poppetal.
2 is
i.e.aDOASanalysisfollowedbyairmassfactorcalculationsbasedonradiativetransfer c
u
s
computations. s
10 io
Our analysis demonstrates that APEX is clearly sensitive to NO VCD above typi- n TitlePage
2 P
cal European tropospheric background abundances (>1×1015moleccm−2). The two- a
p Abstract Introduction
e
dimensional maps of NO2 VCD reveal a very plausible spatial distribution with strong r
gradientsaroundmajorNO sources(e.g.Zurichairport,wasteincinerator,motorways) Conclusions References
x |
andlowNO inremoteareas. ThemorningoverflightsresultedingenerallyhigherNO
15 2 2 D Tables Figures
VCDandamoredistinctpatternthantheafternoonoverflightswhichcanbeattributed is
c
to the meteorological conditions prevailing during that day (development of the bound- u
ss J I
ary layer and increased wind speed in the afternoon) as well as to photochemical loss io
n
of NO2. The remotely sensed NO2 VCD are also highly correlated with ground-based P J I
20 in-situ measurements from local and national air quality networks (R=0.73). Airborne ap
e
NO remote sensing using APEX will be valuable to detect NO emission sources, to r Back Close
2 2
provide input for NO2 emission modeling, and to establish links between in-situ mea- | FullScreen/Esc
surements, air quality models, and satellite NO products.
2 D
is
c Printer-friendlyVersion
u
s
1 Introduction s
io InteractiveDiscussion
n
P
25 Nitrogen dioxide (NO2) is an important reactive trace gas in the troposphere. NO2 ap
acts as an ozone and aerosol precursor and can directly or indirectly affect human e
r
2450 |
health (e.g. pulmonary or cardiovascular diseases) (Brunekreef and Holgate, 2002) D
is
and ecosystem functions and services (e.g. damage of leaves, reduction of crop pro- c AMTD
u
s
duction, acidification) (Bell and Treshow, 2002). Besides natural sources such as s
io 5,2449–2486,2012
lightning and soil emissions the major fraction of tropospheric NO is related to an- n
2
P
thropogenic activities, notably fossil fuel combustion by traffic and industry. Despite a
5 p
significant improvements of air quality in European countries during the past two e NO remote sensing
r 2
decades, air quality thresholds are still frequently exceeded and further efforts are from APEX
|
neededparticularlyregardingreductionsofparticulatematter,ozone,andnitrogenox-
ides (NOx=NO+NO2). Measurements of NO2 in the troposphere are performed with Dis C.Poppetal.
c
various in-situ, airborne, and spaceborne instruments. Tropospheric vertical column u
10 s
densities (TVCD) retrieved from satellites (e.g. from the Scanning Imaging Absorp- sio
n TitlePage
tion Spectrometer for Atmospheric Chartography (SCIAMACHY), the Global Ozone
P
Monitoring Experiment (GOME(-2)), or the Ozone Monitoring Instrument (OMI)) have a
p Abstract Introduction
e
largely contributed to a better understanding of the global distribution of NO as well r
2
as its sources and trends (e.g. Boersma et al., 2004; Richter et al., 2005; van der A Conclusions References
15 |
et al., 2008; Zhou et al., 2012). The spatial resolution of satellite products in the order
D Tables Figures
of multiple tens of kilometers is only sufficient to detect aggregate sources like entire is
c
cities(Beirleetal.,2011)andindividualsourceslikeemissionsfrompowerplants(Kim uss J I
etal.,2006)orships(Beirleetal.,2004)iftheyaresufficientlyseparatedinspacefrom io
n
20 other sources. Ground-based in-situ instruments, on the other hand, provide accu- P J I
a
rate and continuous trace gas measurements but lack of homogeneous geographical p
e
coverage. Airborne remote sensing observations can in this regard provide a valuable r Back Close
link between ground-based and spaceborne NO2 information. For example, airborne | FullScreen/Esc
multi-axis differential optical absorption spectroscopy (AMAXDOAS) was used to re- D
25 trieve NO2 TVCD (Wang et al., 2005), NO2 profile information (Bruns et al., 2006), or iscu Printer-friendlyVersion
to validate SCIAMACHY NO TVCD (Heue et al., 2005). Heue et al. (2008) demon- s
2 s
strated the capability of an imaging DOAS instrument to retrieve two dimensional NO io InteractiveDiscussion
2 n
distributions over the highly polluted Highveld plateau in South Africa. P
a
p
e
r
2451 |
The Airborne Prism EXperiment (APEX) imaging spectrometer is a state-of-the art D
is
instrument with an unprecedented combination of high spectral and spatial resolution, c AMTD
u
s
good two dimensional geographical coverage, and high signal stability. Two NO dis- s
2 io 5,2449–2486,2012
tribution maps were retrieved from imaging spectrometer data acquired over Zurich, n
P
Switzerland, in the morning and the late afternoon of 26 June 2010. Our results are a
5 p
considered as one of the first spatio-temporal investigations of the NO distribution on e NO remote sensing
2 r 2
aregionaltolocalscale. Inparticular,wepresentthefirsthighresolutionmapsofNO from APEX
2 |
VCD in a city measured by an airborne imaging spectrometer. The specific objectives
D C.Poppetal.
ofthisstudyare(i)thepresentationofaretrievalschemetoobtainNO2 quantitiesfrom is
c
APEX imaging spectrometry data and (ii) the qualitative and quantitative assessment u
10 s
of the NO2 products, considering amongst others in-situ measurements of NO2. sio
n TitlePage
P
a
p Abstract Introduction
2 Instrument and data acquisition e
r
Conclusions References
2.1 The APEX instrument |
D Tables Figures
APEX is a dispersive pushbroom imaging spectrometer for environmental monitoring is
c
u
15 developedbyaSwiss-BelgiumconsortiumintheframeworkoftheESA-PRODEXpro- ss J I
gramme (Itten et al., 2008). APEX consists of the imaging spectrometer itself, a Cali- io
n
brationHomeBase(CHB)forinstrumentcalibration,andadataProcessingandArchiv- P J I
a
ing Facility (PAF) for operational product generation (Jehle et al., 2010). Table 1 gives pe
r Back Close
a brief overview of the sensor characteristics.
TheAPEXimagingspectrometerconsistsofaCCDdetectorforthevisibleandnear | FullScreen/Esc
20
infrared (VNIR) and a CMOS detector for the shortwave infrared (SWIR) wavelength D
is
region. TheNO retrievalisbasedonabsorptionbandsintheUV/VISspectraldomain c Printer-friendlyVersion
2 u
s
and, hence, only the VNIR specification is discussed hereinafter. Since atmospheric s
io InteractiveDiscussion
trace gases exhibit spectrally narrow absorption features, the (spectrally) unbinned n
P
25 configurationwasappliedtoprovidehighestspectralresolution. Thespectralsampling a
p
interval (SSI) and the full width half maximum (FWHM) are non-linear functions of e
r
2452 |
the wavelength and increase with longer wavelengths. According to pre-flight sensor D
is
calibration, the SSI increases from 0.63 to 1.44nm and the FWHM from 0.83 to 1.66 c AMTD
u
s
between 420nm and 520nm where NO slant column densities (SCD) are usually s
2 io 5,2449–2486,2012
derived. APEXisapushbroomscannerandmeasuresradiancesin1000spatialpixels n
P
across-track. The extent of the flight line along-track depends on the pre-defined flight a
5 p
pattern. The spatial resolution in across-track direction is determined by the sensor’s e NO remote sensing
r 2
instantaneousfieldofview(IFOV)of0.028◦. Thespatialresolutionalong-trackdepends from APEX
|
on the integration time. This unparalleled combination of high resolution, geographical
D C.Poppetal.
coverage, and high signal stability makes APEX very attractive for a range of remote is
c
sensing applications, e.g. in the fields of vegetation, atmosphere, limnology, geology, u
10 s
or natural hazard studies. sio
n TitlePage
P
2.2 Test site and data a
p Abstract Introduction
e
r
Within the APEX acceptance flight campaign in 2010, extensive data acquisition ac- Conclusions References
|
tivities took place over Belgium and Switzerland. Six of these image data sets were
acquiredinunbinnedmodeoverZurich,Switzerland,onSaturday26June(c.f.Fig.1). D Tables Figures
15 is
Three of them were flown around 10:00 local time (08:00UTC) and three in the late cu
afternoon around 17:30 local time (15:30UTC). The data was acquired approximately ss J I
io
5000mabovegroundwithanintegrationtimeof57ms,leadingtoapixelsizeofabout n
P J I
2.5macross-trackandapproximately6malong-track,respectively. Theentiredaywas a
p
20 cloud-free. er Back Close
Zurich is Switzerland’s largest city with nearly 400000 inhabitants surrounded by
| FullScreen/Esc
an agglomeration of more than one million inhabitants. The test site includes a wide
D
range of surface types like buildings, roads, parks, forests, and part of Lake Zurich is
c Printer-friendlyVersion
(Fig. 1, c.f. also APEX true color composite in Fig. 7a). It also includes more rural u
s
s
25 areas with decreased atmospheric pollution levels like the Uetliberg mountain range. io InteractiveDiscussion
n
SeveralmajorNO sourceswerecovered. Threenationalmotorwayssurroundthecity
x P
a
to the north, west and south (A1, A3, A4) and major transit roads lead through the city p
e
with a usually high traffic volume. In addition, two waste incinerators as well as part r
2453 |
of the approach corridor of Zurich airport are located in the test site area. Zurich was D
is
also selected for the flight experiments because of the dense ground-based air quality c AMTD
u
s
networkwitheightstationswithintherangecoveredbyAPEX(Fig.1). TheNationalAir s
io 5,2449–2486,2012
Pollution Monitoring Network (NABEL) provides half-hourly averaged measurements n
P
of classical air pollutants like NO , O , SO , PM at Zurich Kaserne (urban back- a
5 2 3 2 10 p
ground, situated in a park near the city center) and Duebendorf (suburban). Additional e NO remote sensing
r 2
air quality measurements and meteorological parameters are available from the inter- from APEX
|
cantonal network Ostluft which maintains eight sites in the area of interest from which
D C.Poppetal.
six sites provided data for 26 June 2010: Heubeeribuehl (periphery), Wettswil Filderen is
c
(rural,closetomotorway),WettswilWeieracher(rural),Stampfenbach(urban,kerbside, u
10 s
moderate traffic), Schimmelstrasse (urban kerbside, high traffic volume), and Opfikon sio
n TitlePage
(kerbside, motorway).
P
a
p Abstract Introduction
2.3 Data preparation e
r
Conclusions References
|
APEX data were acquired in unbinned mode to provide highest spectral resolution.
Thisinstrumentalsetting,however,isattheexpenseofthesignalstability. Toincrease D Tables Figures
15 is
the signal-to-noise (SNR) ratio, the imaging spectrometer data were spatially aggre- cu
gated. Aboxsizeof20×20pixelswasappliedwhichisexpectedtoincreasetheSNR ss J I
√ io
twenty times (or 400) assuming uncorrelated noise. An image based SNR estima- n
P J I
tionappliedontheoriginalunbinnedAPEXdatarevealedaSNRof158.5at490nmfor a
p
20 darksurfaces(water). Spatialaggregation,hence,increasestheSNRtoapproximately er Back Close
3170 for dark surfaces. It is relevant to note that image based SNR estimates usually
| FullScreen/Esc
overestimate noise or underestimate the SNR respectively, as surface variability is in-
D
herent in the image statistics. The spatial averaging resulted in a decreased pixel size is
of around 50×120m2. cu Printer-friendlyVersion
s
s
25 The APEX spectrometer is spectrally and radiometrically calibrated pre-flight. How- io InteractiveDiscussion
ever,itisworthmentioningthatthiscalibrationdoesnotcompensateforcertaineffects n
P
occurring in-flight. For example, pushbroom sensors are typically affected by slight ap
e
spectral miss-registrations across-track, e.g spectral smile effects (D’Odorico et al., r
2454 |
2010) which may depend on the specific flight conditions. In order to minimize these D
is
effects, the NO retrieval was performed on geometrically uncorrected data to allow a c AMTD
2 u
s
scan-raw wise processing of the data. In addition, NO is determined from raw data s
2 io 5,2449–2486,2012
(digital numbers or DN) and a spectral calibration is performed directly as part of the n
P
retrieval algorithm to account for spectral effects under flight conditions (c.f. Sect. 3.1). a
5 p
In order to obtain surface reflectance as an important input parameter of the re- e NO remote sensing
r 2
trieval algorithm a software binning was applied to transform the unbinned data from APEX
|
to the standard binning pattern and the data were subsequently calibrated to at-
D C.Poppetal.
sensor radiance using the APEX-PAF. The re-binned radiance data were then at- is
c
mospherically corrected using the ACTOR-4 software tool (Richter and Schla¨pfer, u
10 s
2002) to obtain Hemispherical-Conical-Reflectance (HCRF) data. In a last step, the sio
n TitlePage
unbinned and binned APEX data were geometrically corrected using the PARGE
P
orthorectification software (Schla¨pfer and Richter, 2002). This processing step is a
p Abstract Introduction
e
needed to re-project auxiliary data (e.g. a digital elevation model (DHM25, http://www. r
swisstopo.admin.ch/internet/swisstopo/en/home/products/height/dhm25.html, last ac- Conclusions References
15 |
cess:12March2012))totherawgeometryoftheAPEXdatatoallowascan-rawwise
D Tables Figures
processing as mentioned before. Further, the geocorrected data is essential to re- is
c
late the APEX NO2 vertical column density (VCD) to the in-situ measurements of NO2 uss J I
concentrations and to identify objects of interest like large NO point sources. io
x n
P J I
a
3 NO retrieval p
20 2 er Back Close
The derivation of NO maps from APEX follows the two step approach usually applied
2 | FullScreen/Esc
to satellite NO retrievals. In a first step, differential slant column densities (dSCD) are
2 D
derived by the well known differential optical absorption spectroscopy (DOAS) tech- is
c Printer-friendlyVersion
nique(PlattandStutz,2008). Subsequently,thedSCDareconvertedtoVCDbymeans u
s
s
25 of air mass factors (AMF) calculated with a radiative transfer model. Detailed informa- io InteractiveDiscussion
n
tion about the physical principles, applications, and accuracies of DOAS and AMF
P
a
computationscanbefoundelsewhere(Palmeretal.,2001;Boersmaetal.,2004;Platt p
e
and Stutz, 2008). r
2455 |
3.1 DOAS analysis D
is
c AMTD
u
Differential slant column densities (dSCD) were derived with the QDOAS software s
s
io 5,2449–2486,2012
(http://uv-vis.aeronomie.be/software/QDOAS/,lastaccess:12March2012,Faytetal., n
2011). DOAS analysis requires reference spectra which we obtained from the imag- P
a
p
5 ing spectrometer data themselves. Reference spectra were selected for each individ- er NO2 remote sensing
ual (aggregated) flight column (along-track) to account for and minimize errors in the from APEX
|
DOAS analysis due to spectral miss-calibration (e.g. spectral smile). For this purpose,
ten (across-track) rows were chosen by visual inspection in a forested and elevated D C.Poppetal.
is
area to the south of the city (c.f. Fig. 7a) which is assumed to only contain a back- c
u
s
ground abundance of NO (pollution free). Subsequently, the ten rows were averaged s
10 2 io
in the columnar direction to increase the SNR of the reference spectra. This finally n TitlePage
P
resulted in 50 across-track reference spectra per flight. a
p Abstract Introduction
NO absorptioncrosssections(at293K,Voigtetal.,2002)weresubsequentlyfitted e
2 r
tothedifferentialopticaldepthderivedfromtheAPEXmeasurementsandthereference Conclusions References
|
spectra. Spectrallyslowlyvaryingsignatures(e.g.fromaerosolsorsurfacereflectance)
15
D Tables Figures
were accounted for by including a fifth-order polynomial in the fit and instrumental ef- is
c
fects such as dark current and/or straylight are dealt with in QDOAS by introducing an u
offset spectrum. Inelastic Raman scattering was considered by a Ring cross section ssio J I
n
computed by QDOAS (c.f. Fayt et al., 2011) and the interference with O2-O2 was ac- P J I
20 counted for by fitting an appropriate absorption cross section (Hermans et al., 2002). ap
e
Smallest fitting errors were found by using the 470–510nm wavelength region (c.f. red r Back Close
rectangle in Fig. 2). The usage of a window at shorter wavelengths (e.g. 420–470nm) | FullScreen/Esc
leadtoincreasedfittingerrors,probablyduetothelowersignallevelsandhighernoise
D
(and therewith lower SNR). is
c Printer-friendlyVersion
u
25 Accurate wavelength calibration is an important prerequisite for the DOAS analysis. s
s
As indicated above, raw DN (digital number) data was used to keep the highest sensi- io InteractiveDiscussion
n
tivityofthemeasurementsfortheNO retrieval. Thespectralcalibrationwastherefore P
2 a
performed with the QDOAS algorithm itself. A high resolution solar spectrum (Chance p
e
r
2456 |
and Kurucz, 2010) was applied to obtain spectral calibration information which was D
is
subsequently used to convolve and shift the high resolution absorption cross sections c AMTD
u
s
to the APEX specifications. Two exemplary (pre-processed) APEX spectra from the s
io 5,2449–2486,2012
VNIR detector recorded over a residential and a remote vegetated area are illustrated n
P
inFig.2. Thecorrespondingslantcolumnfitofthepixelovertheresidentialareaispre- a
5 p
sentedinFig.3whichalsogivesanimpressionofthesensor’sspectralresolution. The e NO remote sensing
r 2
quality of the DOAS fit depends on the spectral and radiometric characteristics of the from APEX
|
instrument. The spectral calibration step in QDOAS disclosed some differences worth
D C.Poppetal.
mentioningtothepre-flightreportedAPEXspecifications(Sect.2.1andTable1)inthe is
c
470–510nmwavelengthrange. TheDOASanalysisindicatedthattheindividualchan- u
10 s
nels are positioned roughly 0.6nm higher than reported pre-flight and the (over wave- sio
length) averaged across-track differences of their corrected position (spectral smile) is n TitlePage
P
∼0.25nm. This is in line with recent findings from an in-flight and scene based APEX a
p Abstract Introduction
e
performance assessment (D’Odorico et al., 2011). Furthermore, the QDOAS wave- r
length calibration and slit function characterisation points to a FWHM about double Conclusions References
15 |
the spectral resolution measured in the laboratories at the CHB. The reasons for this
D Tables Figures
discrepancy are not fully understood yet and further analysis is currently carried out. is
c
Finally,notethatnegativedSCDcanoccurwhentheSCDfromthereferencespectrum uss J I
is larger than that from the fitted spectrum or due to noise. io
n
P J I
3.2 Air mass factor calculation a
20 p
e
r Back Close
The AMF expresses the ratio between slant and vertical column of a trace gas:
| FullScreen/Esc
SCD=VCD×AMF (1)
D
is
and is a measure of the average backscatter path through the atmosphere of the pho- c Printer-friendlyVersion
u
tons observed by the sensor. The AMF can be calculated as follows (Palmer et al., ss
io InteractiveDiscussion
25 2001; Boersma et al., 2004): n
P
Pm (bˆ)x a
AMF= L a,L (2) pe
Px r
a,L
2457 |
where the subscript “L” denotes a specific atmospheric layer and m is the (box) air D
L is
mass factor per layer. Besides the layer subcolumns of the a priori NO profile (x ), c AMTD
2 a,L u
s
the AMF depends on forward model parameters (bˆ) such as the solar and viewing s
io 5,2449–2486,2012
zenithandazimuthangles,surfacealbedo,aerosolextinctionprofile,andsurfacepres- n
P
sure. The box air mass factors were calculated using the Linearized Discrete Ordinate a
5 p
Radiative Transfer model (LIDORT, Spurr, 2008): er NO2 remote sensing
from APEX
m =−1 ∂I (3) |
L I ∂τ
L D C.Poppetal.
is
where I is the intensity of the backscattered radiance and τ the optical depth of layer c
L u
“L”. ss
io
10 With regard to radiative transfer computations, the surface albedo for every pixel n TitlePage
P
was derived from re-binned and atmospherically corrected APEX data themselves for
a
p Abstract Introduction
the central wavelength of the fitting window (490nm). Surface height was taken from e
r
the digital elevation model DHM25 previously projected to the raw geometry of the Conclusions References
|
individual flight lines. Surface pressure for every pixel was subsequently obtained ap-
plying the US Standard Atmosphere 1976 (http://www.pdas.com/coesa.html, last ac- D Tables Figures
15 is
cess: 12 March 2012). The a priori NO2 profile was taken from the EURAD chemi- cu
caltransportsimulations(http://www.eurad.uni-koeln.de/,lastaccess:12March2012) ss J I
io
over Switzerland at 5×5km2. The coarse resolution profile was subsequently scaled n
P J I
to the corresponding surface height of the aggregated APEX grid cell according to a
p
e
20 Zhou et al. (2009). Aerosol optical depth (AOD) at 500nm was taken from the Aerosol r Back Close
Robotic Network (AERONET, Holben et al., 1998) site Laegeren which is within ap-
| FullScreen/Esc
proximately 20km from the city of Zurich. The AOD was converted to an extinction
D
profileforeverypixelassuminganexponentialdecreasewithheightandascaleheight is
c Printer-friendlyVersion
of two kilometers. The box air mass factors from three exemplary (aggregated) APEX u
s
s
25 pixels are depicted in Fig. 4. APEX’s sensitivity toward a NO2 signal is highest in the io InteractiveDiscussion
n
atmospheric layer below the aircraft (red horizontal line) and is decreasing toward the P
a
surfaceandtowardhigheratmosphericlayers. Amongalltheabovementionedparam- p
e
eters, the surface albedo has the largest impact on the AMF, e.g. the bright surface r
2458 |