AMTAtmospheric Measurement TechniquesAMTAtmos. Meas. Tech.1867-8548Copernicus PublicationsGöttingen, Germany10.5194/amt-10-1831-2017High-resolution airborne imaging DOAS measurements of NO2 above Bucharest during AROMATMeierAndreas Carlosameier@iup.physik.uni-bremen.dehttps://orcid.org/0000-0001-5918-3233SchönhardtAnjaBöschTimhttps://orcid.org/0000-0003-4230-8129RichterAndreashttps://orcid.org/0000-0003-3339-212XSeylerAndréRuhtzThomashttps://orcid.org/0000-0003-4646-3791ConstantinDaniel-EduardShaiganfarRezaWagnerThomasMerlaudAlexisVan RoozendaelMichelBeleganteLivioNicolaeDoinaGeorgescuLucianBurrowsJohn Philiphttps://orcid.org/0000-0003-1547-8130Institute of Environmental Physics, University of Bremen, Otto-Hahn-Allee 1, 28359 Bremen, GermanyInstitute for Space Sciences, Free University of Berlin, Carl-Heinrich-Becker-Weg 6–10, 12165 Berlin, Germany“Dunarea de Jos” University of Galati, Str. Domneasca 111, Galati 800008, RomaniaMax Planck Institute for Chemistry, Hahn-Meitner-Weg 1, 55128 Mainz, GermanyRoyal Belgian Institute for Space Aeronomy (BIRA-IASB), Avenue Circulaire 3, 1180 Brussels, BelgiumNational Institute of R&D for Optoelectronics (INOE), Magurele, Street Atomistilor 409, Magurele 77125, RomaniaAndreas Carlos Meier (ameier@iup.physik.uni-bremen.de)22May2017105183118574October201617November201622March201716April2017This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by/3.0/This article is available from https://amt.copernicus.org/articles/10/1831/2017/amt-10-1831-2017.htmlThe full text article is available as a PDF file from https://amt.copernicus.org/articles/10/1831/2017/amt-10-1831-2017.pdf
In this study we report on airborne imaging DOAS measurements of
NO2 from two flights performed in Bucharest during the AROMAT
campaign (Airborne ROmanian Measurements of Aerosols and Trace gases) in
September 2014. These measurements were performed with the Airborne imaging
Differential Optical Absorption Spectroscopy (DOAS) instrument for
Measurements of Atmospheric Pollution (AirMAP) and provide nearly gapless
maps of column densities of NO2 below the aircraft with a high
spatial resolution of better than 100 m. The air mass factors, which
are needed to convert the measured differential slant column densities
(dSCDs) to vertical column densities (VCDs), have a strong dependence on the
surface reflectance, which has to be accounted for in the retrieval. This is
especially important for measurements above urban areas, where the surface
properties vary strongly. As the instrument is not radiometrically
calibrated, we have developed a method to derive the surface reflectance from
intensities measured by AirMAP. This method is based on radiative transfer
calculation with SCIATRAN and a reference area for which the surface
reflectance is known. While surface properties are clearly apparent in the
NO2 dSCD results, this effect is successfully corrected for in the
VCD results. Furthermore, we investigate the influence of aerosols on the
retrieval for a variety of aerosol profiles that were measured in the context
of the AROMAT campaigns. The results of two research flights are presented,
which reveal distinct horizontal distribution patterns and strong spatial
gradients of NO2 across the city. Pollution levels range from
background values in the outskirts located upwind of the city to about
4 ×1016moleccm-2 in the polluted city center.
Validation against two co-located mobile car-DOAS measurements yields good
agreement between the datasets, with correlation coefficients of R= 0.94 and
R= 0.85, respectively. Estimations on the NOx emission rate of
Bucharest for the two flights yield emission rates of
15.1 ± 9.4 and 13.6 ± 8.4 mols-1,
respectively.
Introduction
NOx, the sum of NO and NO2, plays a
key role in the chemistry of the atmosphere. In the troposphere, it is
produced by natural and anthropogenic processes, such as fossil fuel
combustion, biomass burning, lightning and bacterial degradation of
fertilizers. Because its main sources are combustion processes, NOx
can serve as an indicator of anthropogenic pollution. The release of
NOx by anthropogenic activity leads to adverse effects on the
environment, such as the formation of tropospheric ozone, eutrophication, and
acid rain, as well as negative impacts on human health. Consequently, there is
a large societal interest in knowing the amounts and spatial distributions,
sources and sinks of NOx.
NO2 exhibits characteristic absorption structures in the UV–visible
spectral range, enabling the application of the DOAS (Differential Optical
Absorption Spectroscopy) method to measure column densities of this trace gas
.
Using this DOAS technique, tropospheric NO2 can be measured from
space-borne satellite instruments, such as GOME ,
SCIAMACHY , OMI
or GOME-2 . While the
satellite measurements demonstrate their value in providing a global picture
of the NO2 distribution, their spatial resolution of several tens of
kilometers is too coarse to investigate the horizontal NO2
distribution on smaller scales such as individual cities.
Measurements of NO2 on smaller scales are usually achieved by
ground-based instruments of different types. Stationary measurements provide
long time series of data and facilitate the investigation of diurnal and
seasonal variability as well as long-term trends. They can thus also be used
for validation of satellite data and chemical models
. Mobile DOAS measurements are also performed,
mostly from cars, offering the advantage of covering a large area at
comparably low costs
.
However, practical issues such as having to follow the pattern of roads or
traffic jams limit the spatial extent of these measurements.
Application of DOAS instruments on airborne platforms may bridge the gap
between ground-based and satellite measurements, because they can cover a
large area in a relatively short time with less limitations on the covered
area and measurement pattern. Airborne DOAS measurements were, for example,
performed using the AMAX-DOAS (airborne multi-axis DOAS) technique
facilitating the validation of satellite instruments and the retrieval of
trace-gas profiles
or the measurement of shipping emissions . Small
instruments have also been applied on ultralight aircraft
.
In more recent years, imaging DOAS (iDOAS) instruments were developed.
demonstrated the applicability for trace-gas
retrievals in a ground-based setup. Installed on aircraft, these systems
enable the creation of maps of the horizontal trace-gas distribution. The iDOAS
measurements of anthropogenic point source emissions of NO2 were
performed by and by above
the South African Highveld plateau and above a German power plant,
respectively. performed measurements of
volcanic emissions of BrO, OClO and SO2 at Mt. Etna,
Italy. Urban NO2 distributions were measured by
above the city of Zürich, Switzerland, and
by above the city of Leicester, England.
Measurements of NO2 in urban areas are essential to understand the
distribution of pollutants and to develop strategies for the mitigation of
air pollution events. Thus, many cities have installed in situ systems to
monitor air quality at ground level. However, these in situ stations have a
sparse spatial sampling. They are often on-road sites for good reasons and
thus directly impacted and dominated by automobile exhaust plumes. Mapping of
NO2 distributions above cities with airborne iDOAS provides a
holistic view on pollution levels across the city and may be used to identify
the contribution of different NOx sources, such as industry and
traffic, to pollution levels in a city.
In this study we present airborne imaging DOAS measurements of NO2
performed in the framework of the AROMAT campaign (Airborne ROmanian
Measurements of Aerosol and Trace gases), which took place in September 2014
in Romania. One purpose of this campaign was to test and compare
state-of-the-art instruments in preparation for the validation of the
upcoming Sentinel-5 precursor satellite mission (S5p;
). The campaign comprised a variety of remote
sensing and in situ instruments used to measure atmospheric composition and
aerosol load, facilitating a detailed characterization and comparison of the
different measurements. The campaign had two target sites, the city of
Bucharest and a power plant in Turceni, with the latter representing an isolated
point source in a rural area. However, this work concentrates on measurements
in the urban area of Bucharest.
The study focuses on the retrieval of accurate VCDs (vertical column
densities) by applying scene-specific surface reflectances determined from
intensities measured by the same instrument, which is not radiometrically
calibrated. The retrieval and application of surface reflectances, i.e., the
method to retrieve the VCD, is explained. The sensitivity of the VCD retrieval
is investigated for several aerosol load scenarios which are based on
measurements close to Bucharest. Results from two research flights above
Bucharest are presented and the dataset is compared to ground-based car-DOAS
measurements for validation. In the last section, the emission flux of
NOx is estimated by applying Gauss's divergence theorem to our data.
The campaignThe target area
The AROMAT campaign took place in Bucharest in September 2014. Bucharest,
located at 44.4∘ N, 26.1∘ E, is the largest city and capital of Romania. It
has around 1.9 million inhabitants
and covers an area of
228 km2.
According to , traffic is the dominant source of NOx in the area, but industrial sources
also contribute to NOx pollution. Nine industrial pollution sources in the area are listed in the European Pollution
Release and Transfer Register (E-PRTR; ), which may have a strong influence
on urban pollution levels.
Map of Bucharest with an overview of the measurements. Blue lines show the flight tracks performed by AirMAP.
Circles and triangles mark the measurement locations of the supporting ground-based mobile DOAS systems. No car-DOAS
measurements are available on 9 September 2014 because on this day these systems were transferred to the other campaign site
at Turceni. Additionally, the locations of the INOE research institute, where ground-based aerosol
measurements were performed, and Baneasa Airport are shown.
The research flights
Two flights above Bucharest with almost the same flight pattern were performed during the AROMAT campaign, cf. Fig. .
The flights aimed at producing maps of the NO2 field above Bucharest. To achieve that, parallel tracks were flown with a
distance of 1900 m. This pattern provides sufficient overlap of the swaths between adjacent flight tracks to produce a
gapless map (see also Sect. ). Both flights were performed on a weekday, Monday and Tuesday,
respectively. During the research flights, an area of about 560 km2 was covered in approximately 1.5 h. The flight on
8 September 2014 started the pattern in the north, whereas the flight on the next day began the pattern in the south. Both flights were
performed around local noon under cloud-free and sunny conditions with low wind speeds (< 1.5 ms-1). Further details
about the flights are shown in Table .
Overview of the analyzed flights.
TotalAt measurement altitude (3.4 km) DayFlight time [h]Flight time [h]Duration [h]SZAmin [∘]SZAmax [∘]2014-09-088.80–11.119.09–10.741.6538.7341.412014-09-097.42–9.207.89–9.201.3141.349.83
Times are provided in UTC; local time (EEST) is UTC+3.
Instrument and data acquisitionThe AirMAP instrument
The Airborne imaging Differential Optical Absorption Spectroscopy instrument for Measurements of Atmospheric Pollution (AirMAP)
has been developed for the purpose of trace-gas measurements and pollution mapping. A detailed description of the instrumental
setup, its performance, the viewing geometry and the georeferencing is given in . Thus, the instrument
is only briefly described here.
AirMAP is a push-broom UV–visible imager with a wide field of view of around 51.7∘ across track in its current setup, leading to
a swath width of about the same size as the flight altitude.
Scattered sunlight from below the aircraft is collected with a wide field-of-view objective. The light is coupled into an
imaging grating spectrograph via a sorted fiber bundle, retaining the spatial information. The fiber bundle consists of 35
individual fibers that are stacked vertically at the spectrometer entrance slit and are oriented orthogonal to the flight direction in
the focal plane of the objective.
The dispersed light is imaged onto an FT-CCD (frame transfer charge coupled device), Princeton Instruments PhotonMAX 512B. The
frame transfer technique of the CCD provides a fast frame rate, because the charges are quickly shifted into a masked storage
area for readout. This procedure allows gapless measurements, because the
next image can be recorded within milliseconds. For data safety reasons, the
CCD readout is interrupted and restarted every few minutes, resulting in small
measurement gaps.
The spectrometer is an Acton SP-300i imaging spectrograph with a focal length of 300 mm, with an f-number of f/3.9,
and temperature stabilized at 35 ∘C. The wavelength region can be chosen according to the chemical species of interest, with
a spectral coverage of either 41 or 86 nm, using a 600 g mm-1 grating blazed at 500 nm
or a 300 g mm-1 grating blazed at 300 nm, respectively.
For the flights above Bucharest described in this paper, the 600 g mm-1 grating was used for measurements in the
visible spectral range (420–461 nm).
Table lists important properties of the AirMAP instrument as an overview.
Properties of the AirMAP instrument during AROMAT.
The across-track spatial resolution depends on the flight altitude, whereas the along-track resolution depends of the ground speed
of the aircraft and the exposure time. During AROMAT, AirMAP was installed on a Cessna 207 Turbo aircraft, operated by the Free
University of Berlin. The AirMAP instrument as well as the aircraft are equipped with an attitude and heading reference system
(AHRS) and a GPS sensor, allowing for accurate georeferencing.
For typical conditions during the AROMAT campaign, (flight altitude 3.4 km, ground speed 60 ms-1, exposure
time 0.5 s), the footprint of one ground pixel is 94 × 30 m2 at nadir.
With the AirMAP setup it is thus possible to examine the sub-pixel variability within one OMI pixel (13 × 24 km2
at nadir) or as S5p-satellite pixel (3.5 × 7 km2 at nadir).
Data preparation and spectral analysis
During the flights, spectra of scattered sunlight from below the aircraft are recorded. The datasets are series of images from
the square CCD chip, with the spectral information on the horizontal axis and spatial information on the vertical axis. In the
post-processing, adjacent rows of the CCD are averaged according to the illumination by the individual light fibers. This results
in time series of individual spectra for each viewing direction. The spectra are georeferenced according to the Cessna's AHRS data,
interpolated to 8 Hz, using a nearest-neighbor synchronization of the GPS timestamps. The GPS altitude is provided as the
altitude above the WGS84 reference ellipsoid and is corrected for altitude above ground level with a digital elevation model
(DEM;
). Calibration and dark measurements are performed on ground. Spectral calibration is
performed using the emission lines of an HgCd spectrum and a high-resolution solar atlas . Subsequently, the
DOAS method is applied to the calibrated spectra. Using an extraterrestrial solar spectrum as
background spectrum in the DOAS analysis, as is done in some satellite retrievals, yields slant column densities (SCDs),
which are the number densities of an absorber, integrated along the light path.
The background spectrum I0 used in the DOAS analysis of the AirMAP spectra is an average over 60 s (120 individual spectra)
taken from a scene of the same flight having low absorber abundances. This back-scattered radiance spectrum may contain small
amounts of the absorber. Thus, the results of the DOAS analysis are the differential slant column densities (dSCDs), representing
the difference of the absorber number densities between the scene studied and the background spectrum.
Because the background spectrum is taken at a clean location, the dSCD is only slightly smaller than the SCD.
Because of aberrations in the imaging system, the spectral resolution of the measurements varies across track. Therefore,
each individual viewing direction has its own spectral calibration and background measurement.
The most important settings used in the DOAS NO2 retrieval are displayed in Table .
Fitted absorption cross sections and other important settings used in the retrieval of NO2 dSCDs.
Molecule/parameterTemperatureReference/propertyO3223 KNO2294 KO4293 KH2O293 KRing effect–Intensity offsetConstantFit window425–450 nmPolynomialQuadraticI060 s average, per viewing directionGridding of data
To generate composite maps from overlapping measurements of NO2, as
well as for the comparison of multiple overpasses, it is necessary to grid
the data. To produce these gridded datasets, a simple gridding algorithm was
used. If not stated otherwise, a regular grid with a spatial resolution of
0.0008∘× 0.0008∘, corresponding to 89 × 64 m2, was
defined, which is the approximate size of two subsequent measurement
footprints during a level flight, i.e., when the aircraft's pitch and roll
angle is small. The measurements are spatially binned by the pixel centers.
All measurements with pixel-center coordinates falling into a grid cell are
assigned to that grid cell. Multiple measurements in one grid cell are
averaged using the unweighted arithmetic mean. This approach was chosen to
optimize computation time but may introduce small biases in the geolocation.
Furthermore, when the size of the footprint becomes larger, gaps are
introduced between the neighboring grid cells. This effect can be observed in
aircraft turns when the projected footprint becomes larger.
Differential slant column densities
Figures and show the retrieved
differential slant column densities measured during two flights performed on
subsequent days, 8 and 9 September 2014, along with major roads. The data
is gridded to having a spatial resolution of 0.0008∘× 0.0008∘. Different
light path lengths, caused by different viewing zenith angles (VZA), were
geometrically corrected by multiplication with the factor cos(VZA).
The dashed white box shows the region where the background spectrum, I0, was taken.
Some lightly negative values occur in the background region as a result of instrumental noise, because the dSCD values are scattered around zero.
In the most polluted areas, dSCD values of up to
6.1×1016moleccm-2 are observed.
The dSCDs show a plume of NO2 spreading south–westwards from the city center. The NO2 field inside the
plume shows small-scale structures with high values. Some of these structures, e.g., the pronounced values at the ring
road in the south-west, are not associated with NO2 emissions from traffic but are related to bright surfaces.
These bright surfaces can enhance the NO2 dSCDs by about 50 % compared to neighboring pixels, having a darker
surface, see also Fig. in Sect. .
Section explains the origin of these spatial patterns above bright surfaces and describes the
approach used to account for these radiative transfer effects. The results of this correction will be shown in Sect. .
Geometrically corrected differential slant column densities of NO2 above Bucharest measured
on Monday, 8 September 2014. The white dashed box shows the area where the background spectrum was taken.
Major roads (black lines) are overlaid for orientation. The pink ellipse highlights some of the small-scale
structures described in the text.
Geometrically corrected differential slant column densities of NO2 above Bucharest measured
on Tuesday, 9 September 2014. The white dashed box shows the area where the background spectrum was taken. Major
roads (black lines) are overlaid for orientation. The pink ellipse highlights some of the small-scale structures
described in the text.
Derivation of vertical column densities
Using the dSCDs resulting from the DOAS fits, vertical column densities, defined as the absorber concentration integrated along the vertical
direction, can be computed. The conversion of the retrieved dSCD to a VCD
enables comparisons of measured trace-gas column densities irrespective of
the instrument viewing geometry, solar position and surface properties. The
change in light path length, as compared to a vertical path, is usually
expressed in the form of an air mass factor (AMF), which is defined as the
ratio of slant and vertical column densities:
AMF=SCDVCD.
Conversion of retrieved dSCD to tropospheric VCD
Satellite platforms have the advantage of being able to measure a solar
spectrum without atmospheric absorption as a background spectrum for the DOAS
analysis. This is not the case for platforms operating within the Earth's
atmosphere. In addition, AirMAP has no option to point into the zenith
direction. Thus, the background spectrum for the DOAS retrieval is taken
above a rural scene with small NO2 concentrations on the same flight.
The following equations describe the conversion of the retrieved dSCDs to
tropospheric VCDs (VCDtrop). In these equations the terms dSCD and
AMF refer to the trace-gas amount fitted from a single spectrum and its
corresponding air mass factor.
The superscripts “trop” and “strat” refer to tropospheric and stratospheric parameters, respectively. The subscript
“0”
refers to conditions of the DOAS background spectrum measurement.
First, the dSCDs are converted to tropospheric slant column densities by
correction of the absorber amount in the rural background scene,
SCD0trop, and by changes in the stratospheric slant column, relative to
the background spectrum, ΔSCDstrat:
The individual terms introduced in Eqs. () and () are discussed in the following subsections.
Accounting for the tropospheric amount of NO2 in the background spectrum
In order to correct for the amount of tropospheric NO2 in the
background spectrum, SCD0trop, we take an approach similar to the one
used by . A tropospheric vertical column of
VCD0trop=1×1015moleccm-2 is assumed over the
background region, which is a representative value for Europe during the
summer period as shown in . The background
spectrum is an averaged spectrum from 120 individual spectra, which may have
different AMFs caused by changing conditions (geometry, surface reflectance)
between these measurements. For each individual measurement during the
integration time of the background spectrum, the AMF is computed for the
respective condition during a single exposure, as will be shown in
Sect. . The AMF of each single spectrum is multiplied
with the VCD0trop, see Eq. (). The average of the
product is used as the tropospheric part of the reference background
(SCD0trop).
Stratospheric correction
Changes in the stratospheric slant column, as compared to the background
scene, propagate to the tropospheric columns measured by AirMAP. Thus, we
correct for changes in the stratospheric column of NO2 in the term
(ΔSCDstrat). The stratospheric correction is applied using the
Bremen 3D Chemical Transport Model (B3dCTM) model. The B3dCTM is a combination of
the Bremen transport model with the
“chemistry code” of the Bremen two-dimensional model of the stratosphere and
mesosphere , which
evolved from SLIMCAT . The model is
driven by ECMWF ERA-Interim meteorological reanalysis fields
. The description of the model setup can be found
in . The B3dCTM model provides stratospheric
VCDs of NO2 on a global grid in a resolution of 2.5∘× 3.75∘.
The model can reproduce relative changes in the stratosphere quite well, but
biases exist in the absolute amounts of NO2. Thus, we do not
use the model data directly but scale the values to match GOME-2 satellite
measurements over the clean Pacific at the latitude of Bucharest. The scaling
factor fmodsat is derived from the following formula:
fmodsat=VCDsat(Pacific)VCDB3d(Pacific),
where VCDsat is the average stratospheric VCD of NO2 in the
longitude range 180–220∘ and at the latitude of Bucharest on the day of
the measurement. VCDB3d is the modeled VCD in the same region at the
time of the satellite overpass.
To correct the change in the stratospheric NO2 over time, the
following formula is applied for each measurement, in which the stratospheric
AMF is approximated geometrically. The geometric approximation of the
stratospheric AMF is valid for solar zenith angles (SZA) < 70∘p. 91 and can be applied, because the solar
zenith angle was much smaller during the measurements, cf. Table . A model value is obtained for each location
(lat, long) and time (t) of the measurements.
As the measurements shown in this study were performed around noon, the
diurnal variations in SCDstrat are very small and stratospheric
correction is of minor importance.
However, the correction for variations of SCDstrat becomes more important for flights performed at large SZA and was
therefore implemented in the data processing chain.
For the two flights on 8 September 2014 and 9 September 2014, the
stratospheric VCD was estimated to be around
3 ×1015moleccm-2. The maximum change in the
stratospheric SCD with respect to the reference spectrum, SCDstrat, was
1.5 ×1014 and
3 ×1014moleccm-2, respectively.
Computation of air mass factors
The tropospheric AMF is simulated by a radiative transfer model (RTM). Here
the SCIATRAN RTM is used .
Table lists the parameters that are used to
calculate AMFs for the individual measurements.
Input parameters for the calculation of AMFs and the atmospheric correction to derive surface reflectances.
ParameterSourceAMFSurface reflectanceFlight altitude (H)GPS + correction by DEMxxGround surface reflectance (A)Intensity + atm. corr.xxViewing zenith angle (VZA)CalculatedxxRelative azimuth angle (RAA)CalculatedxxSolar zenith angle (SZA)CalculatedxxWavelength (λ)Center of fit windowxxNO2 profileAssumption: box profile in lowest 500 mx–AerosolsINOE Raman lidarxxFUBISS-ASA-2 (not coincident)
The angles VAA, RAA and SZA are calculated from GPS position, AHRS and AirMAP's viewing geometry. The grid points used
in the RTM calculation are listed in the appendix.
The flight altitude, H, is the altitude of the aircraft above ground level.
The NO2 profile and the flight altitude affect the sensitivity of the
measurements for NO2, because only part of the photons received at
the instrument may have passed through atmospheric layers close to the
ground.
Due to scattering, the measurement sensitivity for the presence of an
absorber generally decreases towards the ground, and this effect is more
pronounced with increasing flight altitude and small surface reflectances.
This is illustrated by Fig. , showing the box-AMF
(BAMF) for a typical flight scenario in a Rayleigh atmosphere for two surface
reflectances, along with the assumed NO2 box profile. For a detailed
explanation of the BAMF concept see and
references therein.
Figure showing the box air mass factor (BAMF) for a flight altitude of 3.4 km, an SZA of
40∘ and
a VZA of 0∘ at a wavelength of 437.5 nm for two different surface reflectances of 0.04 and 0.1 in an
atmosphere without aerosols. The shaded area shows the assumed box profile of NO2 with a constant mixing
ratio in the lowest 500 m.
The BAMF describes the sensitivity of the measurements for an absorber in a certain altitude layer. Almost all photons received at
the aircraft have passed the layer just below the flight altitude, thus exhibiting the highest sensitivity to that layer.
The ground spectral surface reflectance determines the wavelength-dependent fraction of light reflected at the surface. Bright
surfaces increase the relative contribution of light reflected by the surface to the signal received at the aircraft, thereby
increasing the sensitivity to absorbers located close to the ground. Areas with a high surface reflectance in the fitting window
will therefore generally yield larger dSCDs for the same amount of the trace gas present below the aircraft. This is the reason
for the observed small-scale structures mentioned in Sect. .
Figure shows the dependence of the AMF on the surface reflectance for the same scenario as described in
Fig. . The AMF has a strong nonlinear dependence on the surface reflectance, especially for dark surfaces.
Consequently, good knowledge of the surface reflectance is required in order to appropriately correct for its influence on the retrieved trace-gas amounts.
Dependence of the tropospheric AMF on the surface reflectance for a flight altitude of 3.4 km,
an SZA of 40∘ and a VZA of 0∘ at a wavelength of 437.5 nm in an atmosphere without aerosols.
The observation geometry relevant for the RTM calculations is described by
three angles: the solar zenith angle (SZA), viewing zenith angle (VZA) and
relative azimuth angle (RAA). Figure illustrates the meaning
of these angles.
Illustration of the angles describing the observation geometry relevant for RTM calculations. The RAA (not shown)
is the difference between VAA and SAA.
The VZA is the deviation from the direct nadir observation geometry. As the
VZA increases, the light paths get longer. The VZA changes with the viewing
direction but is also altered with the aircraft's attitude. The RAA is the
difference between the solar azimuth angle (SAA) and the viewing azimuth
angle (VAA) of the measurement.
Following the SCIATRAN convention, the RAA is defined as 0∘ if the instrument is pointed towards the Sun
(forward scattering) and 180∘ for the direction away from the Sun (backward scattering).
The SZA is the angle between the zenith and the center of the Sun's disc and
impacts on the length of the light path through the Earth's atmosphere.
The input parameters to SCIATRAN are either measured directly or are
calculated from other known parameters, see
Table . Estimations or assumptions have to be
made for the NO2 profile, the aerosol load and the ground surface
reflectance.
NO2 profile
No information about the NO2 vertical distribution is available for
the conditions of the flights. Thus, assumptions have to be made. In order to
keep the definition of the unknown profile simple, we have chosen to use a
box profile with a homogeneous mixing ratio. The altitude in which the
NO2 resides depends on many parameters, such as emission altitude,
boundary layer height, orography and temperature.
derived NO2 profiles from MAX-DOAS
measurements in China, showing that the NO2 profile height is between
500 and 1000 m. studied the
urban NO2 profile of Toronto and found the average characteristic
profile height to be around 500 m during summer. These studies suggest that
the assumption of a 500 mNO2 layer is a reasonable guess.
Aerosol profile and properties
For the specific flights, presented here, no direct information on the
aerosol profile exist. However, aerosol extinction profiles were measured
during the flight at the EARLINET station INOE with a Raman lidar at
532 nm. This
EARLINET station (www.earlinet.org) is located in Magurele at the
outskirts of Bucharest, cf. Fig. . Furthermore, aerosol
profiles were measured with an airborne Sun photometer, the FUBISS-ASA2
instrument , in the vicinity of Bucharest
during another campaign (AROMAT-2) 1 year after the flights discussed here.
The FUBISS profiles were measured on 30 and 31 August 2015. The
extinction profiles derived from the ground-based Raman lidar (INOE) and the
airborne Sun photometer (FUBISS) are displayed in
Fig. . The initially higher vertical resolution of
the measured profiles was reduced by binning to vertical layers of
200 (INOE) and 240 m (FUBISS). This step was necessary for
the RTM calculations. The names of the profiles indicate the day of
measurement. So for example the profile named “INOE 08 8h–10h” corresponds to
measurements taken on 8 September 2014 in the time interval 08:00 to 10:00 UTC.
The legend also shows the corresponding aerosol optical depth (AOD), which is the integral of the
aerosol extinction profile. For better comparability, the INOE profiles were
only integrated from the ground to the flight altitude. Because no
measurements are available at altitudes close to the ground, the extinction
coefficient of the lowest available altitude was applied to all layers below.
Aerosol profiles used in this study. The FUBISS profiles were derived from measurements of the FUBISS-ASA2
instrument in the vicinity of Bucharest at a wavelength of 450 nm. These measurements were performed during
the AROMAT-2 campaign 1 year later. The profiles with the label INOE correspond to profiles derived from ground-based
Raman-lidar measurements at INOE. The legend also shows the respective value of the aerosol optical depth (AOD), which
is the integral of the extinction coefficient. In order to allow a better comparison of the AOD values, the INOE extinction
profiles were only integrated from the ground to the flight altitude.
For the analysis of both flights in Sect. , we have used the FUBISS profile 31a. The AOD of this profile is
the closest match to the monthly average AOD, available from AERONET measurements, having a value of AOD450=0.26±0.1
(mean ±SD; https://aeronet.gsfc.nasa.gov/new_web/V2/climo_new/Bucharest_Inoe_500.html#2014).
The effects of the aerosol assumptions on the AMFs will be discussed in Sect. .
Due to the lack of further information, we assume that the aerosol profiles
shown in Fig. represent the prevailing aerosol
conditions during the flights investigated here. For all aerosol profiles and
all atmospheric layers, an asymmetry factor of 0.7 and a single-scattering
albedo (SSA) of 0.9 was assumed in the RTM calculations, which is a
reasonable choice for urban aerosols . A
constant Ångström exponent of 1.5 was used in order to convert the extinction
coefficients to the desired wavelength.
To our knowledge, no data product provides information on the ground spectral
surface reflectance in sufficient spatial resolution to be used for our
measurements. Such data is only acquired on campaign basis from instruments
such as APEX or HySpex
. Since the surface reflectance has
a large impact on the AMF, cf. Figure , we estimate
the surface reflectance from our own measurements. The approach used is
described in the following.
Derivation of surface reflectance
The measured spectra contain information on the recorded light intensity
during the exposure. The scalar quantity “intensity” is computed for each
spectrum, which is the average light intensity in the fitting window of the
DOAS analysis, normalized to an illumination time of 1 s. These
measured intensities are largely influenced by the surface reflectance. The
variability of the measured intensities, however, does not only depend on the
surface reflectance, A, but also on other parameters such as the
observation geometry, the solar position and the aerosol load. As the AirMAP
instrument is not radiometrically calibrated, the measured intensities can be
used only in a relative sense. The influence of the observation geometry and
the surface reflectance on the intensities measured by AirMAP were modeled
for a given aerosol scenario (green line in Fig. )
with the SCIATRAN RTM and compiled into a look-up table (LUT). This LUT of
modeled intensities is used to correct for atmospheric effects affecting the
intensities measured by AirMAP. The parameters accounted for in the RTM
calculations are listed in Table . In order to
isolate the contribution of the surface reflectance to the measured
intensity, a six step procedure (a–f) is applied. The individual steps are
explained in the following.
A reference area with a known surface reflectance is identified, which is
large enough to contain footprints of several measurements of each viewing
direction in that area. The reference area and its surface reflectance value
are taken from the ADAM database (A surface reflectance DAtabase for ESA's
earth observation Missions; ). The ADAM database
includes a climatology of surface reflectances derived from MODIS (Moderate
Resolution Imaging Spectroradiometer) satellite data (MOD09A1 MODIS/Terra
Surface Reflectance 8-Day L3 Global 500 m SIN Grid V005). It contains
normalized surface reflectances (SZA = 45∘, VZA = nadir) on a global grid of
0.1∘× 0.1∘ for each month. The shortest wavelength range for which
surface reflectances are available, without extrapolation, is the
459–479nm band. The fit window used in our DOAS retrieval is
425–450 nm. However, large differences of the surface reflectance
between the two spectral ranges are not expected.
Thus, the respective surface reflectance value for the 459–479 nm band is applied here as reference surface
reflectance ARef. Two grid cells of the ADAM database were used as the reference region, both having a surface
reflectance value ARef of 0.0394 ± 0.0043.
The assumption of a small variation between the two spectral bands can be verified by comparison to the OMLER database
, which is a gridded climatology of surface reflectances determined from measurements of the OMI instrument,
covering both spectral ranges at a spatial resolution of 0.5∘× 0.5∘. The grid cell closest to the measurement location
and time (latitude is 44.5∘, longitude is 26∘; month is September; dataset is MonthlyMinimumSurfaceReflectance) has a mean surface
reflectance of 0.041 in the 425–450 nm band and a mean surface reflectance of 0.042 in the 459–479 nm band,
which corresponds to a relative difference of less than 3 % between the two spectral bands and agrees well with the value from the ADAM database.
As for the NO2 retrieval, the individual viewing directions are
treated separately. For each measurement of one viewing direction, with a
footprint in the reference area i, the intensity is averaged to the value
ImeasRef. For simplification of the notation, the parameters of the
observation geometry are summarized in the parameter set P. The flight
altitude throughout the measurements was constant.ImeasRef=1n∑i=1nImeasi(Hi,VZAi,RAAi,SZAi,λ)=1n∑i=1nImeasi(Pi,λ)
The measured intensities in the fit window along with the reference region can be seen in Fig .
For each measurement with a footprint in the reference area i, the LUT
is queried for a modeled intensity at the respective observation geometry and
the surface reflectance value of the reference region, using linear
interpolation. The wavelength, λ, was set to the center of the fit
window (437.5 nm). These modeled intensities are then averaged to the
mean value ImodRef.ImodRef=1n∑i=1nImodi(Hi,VZAi,RAAi,SZAi,ARef,λ)=1n∑i=1nImodi(Pi,ARef,λ)
In the next step, each measured intensity of the flight is normalized to
match the modeled intensities by scaling with the ratio of modeled and
measured intensities above the reference region. This procedure assumes that
the uncalibrated intensities of AirMAP can be calibrated using a single
factor, per viewing direction, derived over the reference region.
Iscaled=Imeas×ImodRefImeasRef
For each measurement, a vector of corresponding modeled intensities for
the viewing geometry, P, and for all surface reflectances is retrieved from
the LUT.
The surface reflectance for the measurement is then determined by
selecting the surface reflectance value for which the modeled intensity from
the LUT best fits the scaled measured intensity. In order to improve
accuracy, linear interpolation is applied to determine the surface
reflectance.
Measured intensities on 8 September 2014 along with the reference region from the
ADAM database (green boxes).
Surface reflectances derived from measured intensities for the flight
on 8 September 2014 inside the ADAM reference region.
The above procedure yields scene-specific surface reflectances for each
measurement which are later used in the AMF calculations.
Figure shows a histogram of the derived
surface reflectances inside the reference region using the method described.
The mean of the surface reflectances agrees with the value of ARef. The
values range from 0.005 to about 0.1. Only very few measurements show larger
values.
A map of the intensity derived surface reflectances for the flight on
8 September 2014 is displayed in Fig. .
Surface reflectances for the Bucharest region derived from measured intensities for the flight on 8 September 2014
The smallest surface reflectances are found in the forest and water areas,
see map in Fig. . The largest values are found above
bright rooftops, which is in qualitative agreement with surface reflectances
for urban surfaces in the literature
.
Surface reflectance data quality
Although the surface reflectance is derived individually for each of the 35
viewing directions, the surface reflectances show a consistent behavior. This
can be seen over areas with a homogeneous surface type, such as the large
forest area in the north-east of the flight pattern. Although the surface
reflectances were derived for all viewing directions independently, the
homogeneous surface type also yields a homogeneous surface reflectance.
To investigate further on the precision of the method, the derived surface
reflectances of two flights on different days were spatially binned on the
same grid with a resolution of 0.0008∘× 0.0008∘. A pixel-wise
comparison is shown in Fig. a. The derived values of
the two flights agree well, having a correlation coefficient of R= 0.91 and a
slope close to 1. Figure b shows the corresponding
histogram of absolute differences. The mean of the differences is close to
0 and the FWHM (Full Width Half Maximum) of the differences is less than
0.01. These results indicate that the applied corrections for viewing
geometry and atmospheric effects are reasonably consistent.
Pixel-wise comparison of co-located surface reflectances,
derived from flights on different days, 8 and 9 September 2014. (a) shows the correlation between
the two flights. The text box shows the correlation coefficient R, the number of data points N and the result
of a linear orthogonal fit (red line). (b) shows the corresponding histogram of the absolute differences.
Another and similar approach to test the precision of the method is to
investigate the derived surface reflectances of multiple overpasses over the
same location during one flight. A large fraction of the flight area was
covered twice, due to an intended overlap between adjacent tracks.
The individual tracks were flown around noon in from east to west and west to east, respectively.
This implies that approximately half of the sensor swath is pointing towards
the Sun (0∘< RAA < 90∘), while the other half is pointing away from the
Sun (90∘< RAA < 180∘). Figure a shows a
correlation plot of the derived surface reflectances from the flight on
8 September 2014 in dependence of the RAA.
Figure b shows the corresponding histogram
of the absolute differences.
Surface reflectance in
dependence of the relative azimuth angle (RAA), as determined from one flight
on 8 September 2014. (a) shows the correlation of surface reflectances
acquired under forward and backward scattering regimes. The text box shows
the correlation coefficient R, the number of data points N and the result of
a linear orthogonal fit (red line). (b) shows the corresponding
histogram of absolute differences.
It can be seen that the surface reflectances acquired under an RAA < 90∘ (pointing towards Sun) generally yield lower values.
The spatial distribution of these differences is displayed in
Fig. . The largest differences occur in the urban, most
densely populated and built-up areas and over water bodies, which can be
clearly identified, cf. Fig. . Regions covered by
vegetation, such as the forest in the north-east or the regions without
buildings, show much lower differences.
Absolute differences of surface reflectance in dependence of the relative azimuth angle (RAA), as determined
from one flight on 8 August 2014. Additionally shown is a zoom in on the lake, highlighting the differences observed over water bodies.
Refer to Fig. to relate the spatial patterns to the surface type.
The reason for this behavior is attributed to the angular dependency of the
ground surface reflection, as represented by the BRDF (Bidirectional
Reflectance Distribution Function). This can be illustrated by a simple
example. In the urban area, the reflecting surface is not usually flat but
rather structured. Houses will appear brighter on the side facing the Sun,
while they appear darker on the opposite side. In general, flat surfaces such
as lakes appear brighter when observed at RAA < 90∘ and darker at RAA > 90∘,
while the opposite is true for non-flat surfaces such as woodlands. Full
consideration of these effects is out of scope of this study. Missing
treatment of BRDF effects introduces a relative uncertainty on the surface
reflectance of around 10 % as can be seen from
Fig. . However, when applying the intensity
derived surface reflectance to the LUT of AMFs, discarding these directional
effects has only a minor impact on the retrieval. The important information
for a correct AMF, matching the scene, is the amount of light that was
reflected on the ground. Since each spectrum in the DOAS analysis has its own
surface reflectance value, recorded under the same observation geometry, the
AMF is corrected implicitly. The derived surface reflectances can thus be
regarded as effective reflectance values.
Effects of the applied surface reflectance on the results
Figure shows a zoom in on the map with the highest
pollution levels during the flight on 8 September 2014. When comparing the measured
intensity (a) and the dSCDs (c), the spatial correlation of bright surfaces
with high dSCDs, as mentioned in Sect. , is clearly
visible. The AMF, (b), is dominated by the amount of light reflected by the
surface. This was accounted for in the AMF by the application of the
intensity derived surface reflectances. The VCDs in (d) are much smoother
than the dSCDs. Small-scale structures, originating mainly from the surface
reflectance, are successfully eliminated.
Overview of the intensity, AMF, dSCD and VCD for the flight on 8 September 2014. Application of the
intensity derived surface reflectances generates a smooth NO2 VCD distribution.
NO2 VCD above Bucharest
Figures and show the VCD of
NO2 retrieved from the flights on 8 and 9 September 2014,
respectively. Both figures show bin-averaged VCD values on a regular grid
with a spatial resolution of 0.0008∘× 0.0008∘.
Individual measurements with a flight altitude lower than 3000 m, a
large fitting error (RMS larger than 0.02) or a VZA lager than 40∘ were filtered
out prior to the gridding procedure.
Vertical column densities measured on 8 September 2014. The numbered labels show NOx emitters listed in
the E-PRTR, cf. Table .
Vertical column densities measured on 9 September 2014. The numbered labels show NOx emitters listed
in the E-PRTR, cf. Table .
The numbered labels show NOx emitters that are listed in the European
Pollutant Release and Transfer Register
. Further details about
these pollution sources are given in Table .
NOx Emitters listed in the European Pollutant Release and Transfer Register (E-PRTR).
Compare entries with locations in Figs. and .
The E-PRTR lists NOx emitters exceeding a threshold of 1×105kgyr-1.
Besides the large plume above the city, several emission hot spots are
identified, which correlate well with the locations of the facilities listed
in the E-PRTR (e.g., 5, 9 in Fig. and 2, 3 in
Fig. ). It should be noted that not all of the listed
emitters have necessarily been active during the time of the research flight.
Despite the fact that both flights were performed under similar conditions
(Monday and Tuesday, around local noon, similar wind speed), the horizontal
NO2 distribution is quite different. The flight on 8 September 2014 has
a larger NO2 VCD of up to 4.2 ×1016moleccm-2,
while the maximum NO2 VCD for the flight on 9 September 2014 is
3.4 ×1016moleccm-2.
It should be noted that despite the different appearances and peak values, the mean NO2 amount of both flights are
similar: the average NO2 VCD is 9.3 ×1015moleccm-2 on 8 September 2014 and
8.9 × 1015moleccm-2 on 9 September 2014.
The reason for the different NO2 distribution is not completely
understood but is probably attributed to the wind conditions.
Figure shows the wind properties for the 2 investigated
days and 1 day before, measured at Baneasa Airport in the north of
Bucharest, cf. Fig. . The shaded areas indicate the
times of the measurements. Comparing the apparent wind direction, as seen in
the NO2 distribution, to the data record of the meteorological
station at Baneasa airport, it is worth noting that there is good agreement
on 9 September 2014, whereas a mismatch of the wind direction is observed on
8 September 2014.
Wind data for the 2 investigated days and the day before measured at the Baneasa Airport. The gray shaded area
indicates the times of the flights. Data provided by Meteo Romania.
On both days there was a similar low wind speed from northern or
east–north-east directions, respectively, during the time of the
measurements. Before the first flight, however, there was actually no
significant wind at all during the morning rush hour. This results in a
stronger accumulation of NO2 close to the sources. If this is the
case, it is possible that the NO2 could reach higher altitudes
before being transported away. This results in a higher sensitivity towards
the NO2 because of a different AMF, see
Sect. .
Assuming a wind speed of 1.4 ms-1, an air parcel would be transported 5 km per hour. The measured plumes
extend to about 15 km from the city center, where the densest traffic is expected. This means that the observed NO2
could already be up to 3 h old.
However, from the available data, no unambiguous and firm conclusion can be
drawn. To investigate on this issue further, reliable high-resolution data
describing the meteorological conditions and emissions of NOx would
be necessary. Because of the very low wind speed, uncertainties on the
transport of NO2 are relatively large.
The area covered during a research flight corresponds to almost 2 OMI
pixels or 23 S5p pixels. Averaging the AirMAP measurements could thus be used
for satellite data validation. We do not show a comparison to OMI satellite
data in this study, because, on the investigated days, the OMI footprints did
not match well with the covered area.
Considering the large spatial gradients of the NO2 field, a comparison of only partially overlapping measurement areas is not
meaningful. However, a comparison to mobile car-DOAS measurements will be shown in Sect. .
Discussion of uncertainties
The total uncertainty on the vertical column originates from (1) uncertainties on the retrieved dSCDs, (2) uncertainties in the applied AMFs
and (3) uncertainties in the background column. The contribution of the
different uncertainties on the final VCD result are discussed in the
following.
Uncertainty on the differential slant column densities
Several effects contribute to the dSCD uncertainty: shot noise from the
radiance, electronic noise from the instrument, uncertainties from the
cross sections (typically around 2–3 %) and errors from spectral
interference in the DOAS retrieval. The resulting individual relative
fitting errors of NO2 above the polluted city center of Bucharest
range from 5 to 10 %. For smaller NO2 abundances, the relative
error is much larger as some of the error sources are absolute errors that do
not scale with the NO2 signal. Therefore, a relative as well as an
absolute error needs to be stated. The combined uncertainty of the dSCDs is
then the sum of the relative and the absolute errors. The random error of the
dSCDs can be estimated from the noise of the retrieved dSCDs in the time and
region where the background spectrum was taken
Chap. 8. Provided that
the tropospheric NO2 column in that area is small and constant, the
observations are scattered around zero, and the RMSE (root-mean-squared
error) provides an estimate on the magnitude of the random errors. Due to the
variation in spectral resolution, the RMSE of the dSCDs varies with the
viewing direction, and ranges from 2.1 ×1015moleccm-2
in the central viewing directions to
2.7 ×1015moleccm-2 in the outer viewing directions.
The mean dSCD error for the flight on 8 September 2014, as output of the DOAS fit,
is 2.2 ×1015moleccm-2, which is in agreement with the
RMSE in the background I0 region.
Uncertainties on the air mass factors
The AMF converts the dSCDs to VCDs. Thus, uncertainties on the AMF will
affect the uncertainties on the VCDs directly, mainly in the form of relative
errors. The largest uncertainties contributing to the error on the AMF are
the aerosol effects, followed by the surface reflectance and the unknown
NO2 profile.
Uncertainties on the surface reflectance
Figure shows the strong nonlinear dependence of the
AMF on the surface reflectance for a typical observation scenario. While the
varying surface reflectances are captured well in our retrieved surface
reflectances, they all depend on the surface reflectance value of the
reference area. Missing treatment of BRDF effects might also cause an
uncertainty of around 10 % on the retrieved surface reflectance itself. For
the application in the AMF-LUT, however, the AMFs are implicitly corrected, as
discussed in Sect. .
From the comparison of the surface reflectance derived on 2 different days
in Fig. b, the precision of the surface reflectance
retrieval was assessed to be 0.006. Assuming a surface reflectance of 0.04,
this results in a statistical uncertainty of 6 % on the AMF. The value of
the precision on the surface reflectance describes an upper limit, because
the value determined is influenced by gridding artifacts and includes
variations caused by directional reflectance properties of the surface. The
effect of aerosols on the surface reflectance retrieval is discussed in
Sect. .
Uncertainties introduced by the NO2-profile assumptions
Figure shows altitude-dependent box air mass factors
for a flight altitude of 3.4 km, a constant surface reflectance of
0.04 and a solar zenith angle of 40∘. The BAMF describes the sensitivity of
the retrieved slant column to the presence of a given amount of an absorber
like NO2 in a given altitude.
From the figure the influence of the absorber profile on the AMF can be estimated. For the displayed scenario,
a surface reflectance of 0.04 and a well-mixed NO2 box profile of 0.5 km height, the resulting AMF would
be 1.7. If the maximum profile height were increased to 1 km, the resulting AMF would be 2. Under the assumption
that the profile altitude is in the range of 500 to 1000 m the uncertainty of the profile on the AMF
is on the order of 10 %. The magnitude of the uncertainty introduced in the AMF by the profile uncertainty increases
with decreasing albedo and with increasing SZA and can be on the order of 20 % or larger for more deviating profiles at lower Sun.
Uncertainties related to aerosols
Aerosols can have several impacts on the retrieved vertical columns. If a
layer of aerosols is present above a trace gas it obscures the view on the
trace-gas layer by shielding it through the increased scattering probability.
This effect would bias the VCD low if not taken into account. On the other
hand, aerosols can lead to multiple scattering effects which extend the light
path within the aerosol layer. If the aerosols and the trace gas are present
in the same layer, this will lead to a larger absorption of the trace gas of
interest, biasing the VCD high. These considerations assume aerosols with
a large SSA. For absorbing aerosols, the light path
enhancement effect is reduced and VCDs are low biased also in case of a
well-mixed trace-gas and aerosol layer. Studies for satellite observations
report on sensitivities varying between a few
percent and up to 20 % for aerosol layers located above the trace gas of
interest.
Figure shows aerosol profiles used in the SCIATRAN
RTM calculations, which were derived from ground-based Raman-lidar and
airborne Sun-photometer measurements. The profile used in the analysis
(FUBISS 31a) represents a scenario with a rather low aerosol load.
Furthermore, the used aerosol profile assumes a small extinction in the
lowest 500 m, whereas the NO2 probably mainly resides in that
layer. Thus, the examined aerosol profiles mainly reflect the shielding
effect of aerosols. Light path enhancements in the trace-gas layer by
aerosols are thus not represented well. Therefore, the AMFs are probably
underestimated and the VCDs are more likely biased high. To study the
influence of the aerosol profile on the AMF, all available profiles were used
for AMF calculations and the results compared to the profile used in this
study. Figure compares AMFs of the other
aerosol scenarios shown in Fig. and AMFs in a pure
Rayleigh atmosphere to the profile applied in this study. The shaded area
describes a ±30 % deviation of the AMFs depending on the used aerosol
profile. In the case of a Rayleigh atmosphere, the AMF is larger, because no
shielding aerosol effect is present. With increasing extinction above the
trace-gas layer, the AMF becomes smaller.
Influence of the assumed aerosol profile on the AMFs for an NO2 box profile in the lowest 500 m.
The shaded area represents a ±30 % uncertainty on AMFs of the used aerosol profile. The legend also shows the mean
value of the individual ratios between the AMF of the respective scenario and the scenario used.
Certainly, a single profile cannot describe the aerosol distribution across
the whole extent of the investigated area. The INOE profiles were measured at
the outskirts of Bucharest. Due to restrictions on the air space, the
profiles derived from the FUBISS-ASA2 instrument were also measured outside
the city and are not coincident in time. The uncertainties thereby are
difficult to quantify. Nevertheless, the range between the investigated
aerosol scenarios provides an estimate on the impact of the aerosol profile
on our measurements.The range of AMFs obtained when using the different
aerosol profile assumptions supports the uncertainty values reported in
. It should be noted that the aerosol load also
influences the derived surface reflectances. If the true aerosol load is
larger than assumed in the RTM calculations of the modeled intensities, the
scaling factor in Eq. () becomes smaller, resulting in a
low biased surface reflectance. Underestimating the surface reflectance
results in a low biased AMF and consequently in high biased VCDs. This effect of
an underestimation of the aerosol load in the RTM calculations for the
surface reflectance is opposing to the effect on the AMF, because an
underestimation of the aerosol in the modeled AMF underestimates the
shielding effect of aerosols. As discussed above, an underestimation of the
shielding effect causes high biased AMFs, resulting in low biased VCD. It can
thus be argued that these opposing effects partly compensate each other. The
surface reflectance was derived by using a single simplified aerosol scenario
with a low AOD. Applying AMFs corresponding to a larger aerosol load than
accounted for in the surface reflectance retrieval leads to low biased AMFs.
Uncertainty resulting from the NO2 amount in the background spectrumUncertainties in the tropospheric background
As no direct measurements of the NO2 column in the background scene
(VCD0trop) exists, this value is quite uncertain in relative terms.
Assuming a 100 % uncertainty on the 1 ×1015moleccm-2
value, a tropospheric AMF of 1.2 and a tropospheric AMF over the background
scene of 0.8, this adds an uncertainty of
7 ×1014moleccm-2 to the tropospheric vertical column.
Uncertainties related to changes in stratospheric NO2
The stratospheric NO2 signal changes with the SZA and by
photochemical reactions. Our measurements were performed around local noon.
Thus, the relative changes in the SZA and the light intensity are rather
small. Furthermore, we apply a correction for changes in the stratospheric
NO2 amount with respect to the background spectrum. The maximum of
the change in the stratospheric SCD is
3 ×1014moleccm-2. Assuming a 100 % uncertainty on the
applied correction, and a tropospheric AMF of 1.2, the uncertainty on the
stratospheric contribution is around 3 ×1014moleccm-2.
Summary of uncertainties
Table summarizes the major uncertainties deduced
from the considerations made above. Assuming that the sources of the
uncertainties are unrelated, the combined uncertainty on the AMF can be
estimated by the square root of the quadratic sum of individual
uncertainties. The largest uncertainty on the AMF arises from the assumptions
on the aerosol load and properties, followed by the NO2 profile and
the surface reflectance. The resulting combined uncertainty on the AMF is
less than 26 %. The assumption of independent uncertainties is not
completely valid, because of the link between the retrieved surface
reflectance and the shielding effect, which depends on the aerosol load.
Because the uncertainty on the dSCD is an absolute value, it does not scale with the NO2 signal. A relative uncertainty
is stated for a typical dSCD value. Taking the mean dSCD value and the mean dSCD error as a reference for a typical value, and
taking into account the main influences listed in Table , the overall uncertainty on the NO2
VCD is about 35 %.
Major contributors to uncertainties.
ParameterUncertainty assumptionReference caseRelative uncertaintyRemarkAMF/dSCDSurface reflectance0.0060.046 %cf. Fig. bNO2-box-profile height+500 m500 m10 %cf. Fig. Aerosolsall aerosol scenariosFUBISS 31a3–23 %cf. Fig. dSCD error2.2 ×1015moleccm-21 ×1016moleccm-222 %From mean dSCD errorand mean dSCDTropospheric background1 ×1015moleccm-21 ×1016moleccm-27 %cf. Sect. Comparison to mobile car-DOAS measurements
On 8 September 2014 mobile car-DOAS measurements were performed by the Max Planck
Institute for Chemistry in Mainz (MPIC) and the University of Galati (UGAL).
For the flight on the next day, no supporting measurements are available
because the car-DOAS instruments were transferred to the Turceni power plant,
which is the other campaign measurement site.
The most important properties of the mobile car-DOAS measurement are shown in Table .
More detailed information about the MPIC and UGAL VCD retrieval algorithms used can be found in and
as well as , respectively.
The data provided by the two groups can be used to validate the VCD retrieved
from AirMAP to independent ground measurements. For the comparison, the
car-DOAS VCD data was filtered to contain only measurements during the
research flight and gridded to a resolution of 0.03∘× 0.03∘. The same
grid was applied to the AirMAP VCDs and a pixel-wise comparison of the
co-located pixels was performed. For an overview of the locations of the car
measurements refer to Fig. . Figures a and a show correlation plots, with the VCD
retrieved by AirMAP on the x axis and the VCD retrieved by the MPIC and UGAL
car-DOAS instrument on the y axis. Figures b and b show a time series for the car measurements, along
with AirMAP's retrieved VCDs at the respective car positions. In this sense,
the time axis is only valid for the car-DOAS measurements. The lower panel
shows the temporal difference between the airborne and the ground-based
measurements.
(a) Correlation between VCD retrieved by AirMAP and the VCD retrieved by the UGAL car-DOAS instrument
measured on 8 September 2014. The text box shows the correlation coefficient R, the number of data points N and the result
of a linear orthogonal fit (red line). (b) Time series of the car-DOAS measurements along with AirMAP measurement
at the respective car positions. The lower panel shows the temporal difference between the measurements compared.
(a) Correlation between VCD retrieved by AirMAP and the VCD retrieved by the MPIC car-DOAS instrument
measured on 8 September 2014. The text box shows the correlation coefficient R, the number of data points N and the result
of a linear orthogonal fit (red line). (b) Time series of the car-DOAS measurements along with AirMAP measurement
at the respective car positions. The lower panel shows the temporal difference between the measurements compared.
Both comparisons reveal a good correlation between the datasets, with
correlation coefficients of R= 0.85 (MPIC) and R= 0.94 (UGAL), respectively.
The larger spread in the comparison to the MPIC instrument is probably caused
by the viewing geometry, the driven route and the spatial pattern and
temporal variability of the NO2 field. The spatial information on the
location of the car measurements is taken from the car's position during the
measurement. The location of the AirMAP measurement is the center of the
projected footprint of the ground pixel. Since the MPIC instrument is
pointing at an elevation angle α of 22∘, it integrates a different
horizontal air mass.
When comparing to the UGAL instrument the scatter is smaller, presumably
because the spatial inhomogeneities in the NO2 field affect the
comparison to a much smaller extent because the instrument is pointed to the
zenith.
The maximal horizontal mismatch dmax between the airborne and the
ground-based measurement can be approximated by a simple geometrical
approach.
dmax=HNO2tan(α)+HNO2×tan(VZA)
For the assumed 500 mNO2 box profile, HNO2, and
a maximum VZA of AirMAP during a level flight, this results in a maximal
displacement of 240 m for the zenith-pointing UGAL instrument,
whereas the possible mismatch for the MPIC instrument adds up to
1480 m. This explanation is further supported by
Fig. b, where an obvious time lag between the two
datasets can be seen shortly before 10:00 UTC, which translates to a spatial
mismatch.
A linear orthogonal fit to the data reveals a slope of 0.89 for the
comparison to the UGAL data, indicating an overestimation of AirMAP's VCDs or
an underestimation of the UGAL data. However, the slope in the comparison to
the MPIC data has a value of 1. A possible reason for the lower values
obtained from the UGAL instrument could be related to the assumptions made on
the NO2 profile. To investigate on this hypothesis, the BAMF for
ground-based zenith sky observations at an SZA of 40∘ was weighted with (a) the 500 m box profile assumed in the AirMAP retrieval and (b) the
profile used in the UGAL retrieval to yield the respective AMFs. The
NO2 profile used by UGAL was extracted from the CHIMERE model for a
small town (Timisora, Romania: ∼ 300 000 inhabitants) and assumes an
exponentially decreasing mixing ratio up to an altitude of 6 km. The
ratio of these differently weighted BAMFs (AMFs) is 0.90 (AMF500m/AMFUGAL).
This ratio is close to the slope determined from the fit
shown in Fig. a, indicating that the differences between
the two instruments can be explained by the assumptions made on the profile.
The data retrieved from the different instruments and used in this study were
analyzed independently by all groups without common assumptions on the
NO2 profile, aerosols and other properties related to the radiative
transfer. Considering the large sensitivity of the AirMAP AMFs on the surface
reflectance and the aerosol profile, the datasets show good agreement.
Estimation of the urban NOx emission rate
Several studies have investigated the NOx emission rate of point
sources and urban areas
.
Here we adapt the method presented in and
, where urban emissions are estimated from
an encircled area, by integrating along the route of a circle S. This
method is based on Gauss's divergence theorem, describing the relation
between the flux of a vector field through a closed surface (measured) and the
divergence of the vector field inside the enclosed volume (emissions inside
that volume). The NO2 emission rate FNO2 may be estimated
from
FNO2=∮SVCDNO2(s)⋅w‾⋅n‾⋅ds.
Here, n‾ indicates the normal vector parallel to the Earth's
surface and orthogonal to the azimuth of the line segment ds, and
w‾ is the mean horizontal wind vector. NOx is primarily
emitted as NO and converts to NO2 by reaction with O3.
NOx has a short lifetime on the order of hours. In order to derive
the NOx emission rate from the measured VCDNO2,
Eq. () has to be modified to account for (a) the partitioning between NO and NO2 and (b) chemical loss.
Assuming steady state conditions, the partitioning between NO and
NO2 can be described by the Leighton relationship. For typical urban
conditions at noontime, the ratio of [NO]/[NO2] is about
0.32 . Therefore, the factor cL= 1.32 is
introduced, which scales the measured NO2 to NOx. The
correction factor cτ accounts for chemical loss that occurred from
the location of emission to the location of the measurement. From the NOx lifetime τ can be estimated from the wind speed w‾ and
the distance between source and measurement d.
cτ=expd/w‾τ
The lifetime of NOx is variable and depends on ozone levels and the actinic flux. A typical lifetime of NOx is
3.8 h as shown in , where lifetimes of urban NOx were estimated from satellite data. For
simplicity, the lifetime of NOx, τ, is set to 3.8 h.
Taking into account these correction factors, the NOx emission rate
FNOx is calculated by
FNOx=cL⋅cτ∮SVCDNO2(s)⋅w‾⋅n‾⋅ds.
To apply this method to our measurements, we have used the gridded data as
shown in Figs. and . The
measurement locations were converted to a Cartesian coordinate system,
setting the origin in the center of Bucharest (lat is 43.4355∘ N, long is 26.1025∘ E).
Circles around this origin were defined with angular displacements in steps
of 0.1∘, resulting in 3600 sampling points. This step was performed for many
radii in steps of 100 m, and an interpolated value at the circle
locations was obtained. The VCDNO2 value for locations outside of the
measured area was set to the background value of
1 ×1015moleccm-2.
Because of the discrete data, the integral is approximated by the sum of the
sampled values.
FNOx=cL⋅cτ⋅∑iVCDNO2(si)⋅w‾⋅cos(βi)⋅Δsi
The inflow of NO2-enriched air is accounted for in the term
cos(β), which is the angle between the normal of the wind direction
and the azimuth of the line segment. This term ensures that NO2
transported into the encircled area becomes negative and does not contribute
to the emissions determined from within the circle.
The term Δs is the Euclidean distance between the sample locations. The wind speed, w‾, measured at Baneasa airport
was used and set constant for each flight. The wind direction was determined from the apparent distribution of the plume.
Table lists the parameters used to analyze the two flights.
Parameters used to calculate the NOx emission rate.
Parameter8 September 20149 September 2014Center coordinate44.4355∘ N,44.4355∘ N,26.1025∘ E26.1025∘ EWind speed (w‾)1.1 ms-11.4 ms-1Wind direction57∘65∘cL1.321.32NOx lifetime (τ)3.8 h3.8 h
Figure shows the NOx emission rate of
Bucharest determined from the method described above in dependence of
distance to the city center. On both of the analyzed days, the determined
emission rate increases until a distance of 10.9 km. At larger
distances, the emission rate reaches a plateau. This behavior is related to
the area covered by the flights, because at larger distances no measurements
are available south of the center coordinate and the values are set to
background values. Assuming background values is reasonable for the areas
upwind (north), but it is not appropriate for downwind areas (south) outside the
measured domain, where enhanced VCDNO2 are expected, cf.
Figs. and . Thus, only emission
rates up to a distance of 10.9 km give meaningful results. When
considering the emissions within the radius of 10.9 km, the
NOx emission rate is 15.1 mols-1 on 8 September 2014 and
13.6 mols-1 on 9 September 2014. The increase of the emission rate
with distance may have two possible reasons: (a) NOx has not yet
reached its steady state ratio or (b) the area where NOx is emitted
increases, resulting in a larger emission rate. The latter is certainly true
for this urban area, because the emissions do not only occur in the city
center but across the whole extent of Bucharest. The effect mentioned first
may explain the smaller slope at small distances to the origin.
Assuming an uncertainty of 33 % on the VCDNO2, see
Sect. , and an uncertainty on the wind speed of 50 %, the
overall uncertainty on the determined NO2 emission rate is 60 %.
estimated the uncertainty of the correction
factors, cL and cτ, to be 10 % each. Applying these values here also leads to a total uncertainty on the NOx emission rate of 62 %.
Emission rate of NOx determined on the 2 investigated days. The center coordinate (source) was set to the city center.
Conclusions
In this paper, we presented airborne imaging DOAS measurements performed
during the AROMAT campaign in September 2014. Two flights above Bucharest
were performed, covering an area of about 18 × 33 km2 within
1.5 h with a spatial resolution better than 100 m. These
flights aimed at providing a high-quality and fine-resolution map of the
horizontal NO2 distribution above this large eastern European city.
To correct for the strongly varying surface reflectance within the city and
its impact on the measurements, we have developed a method to derive surface
reflectance information from an instrument which is not radiometrically
calibrated. For this, we have used a look-up table approach, in which an
effective surface reflectance value is derived for each individual
measurement from the measured relative intensity. A combination of MODIS
retrievals over a reference region and SCIATRAN model data of atmospheric
radiation was used to link relative intensities to absolute surface
reflectances. The resulting scene-specific surface reflectances have the
advantage that they directly match the measurements, avoiding artifacts from
spatial sampling and interpolation and at least partially overcoming the need
for precise knowledge of the surface BRDF. Comparison of measurements on the
2 days as well as observations taken under different relative azimuth
angles shows excellent consistency of the derived surface reflectances.
Further validation is planned by direct comparison to APEX measurements taken
during a tandem flight over Berlin in April 2016. Using the AirMAP derived
reflectance values, vertical columns were computed from the differential
slant columns. While the NO2 dSCD distribution shows spatial patterns
related to surface properties, these are no longer observed in the
NO2 VCD distribution, indicating a successful correction for light
path effects. Uncertainties in the AMF calculation were discussed in detail
and inaccuracies in surface reflectance and aerosol assumptions identified
as main error sources. The latter could in principle be improved by aerosol
soundings over the city center, which were not possible during the AROMAT
campaign. Strong spatial gradients in the NO2 distribution could be
observed in the covered area across the city, with NO2 columns
ranging from background values in rural areas upwind to about
4.2 ×1016moleccm-2 over pollution hot spots.
Measurements on subsequent days revealed quite distinct pollution patterns,
probably related to changing meteorological conditions. Validation of the
AirMAP observations with two independent co-located car-DOAS measurements
performed on one of the measurement days shows good agreement between the
datasets, indicating the good quality of the measurements. Using the AirMAP-derived NO2 distribution and wind data, the total NOx
emission rate of Bucharest could be estimated to be about
14.4 ± 8.9 mols-1. The airborne imaging DOAS measurements
reported here illustrate the inhomogeneous and rapidly varying horizontal
distribution of pollution on a city scale that cannot be accessed by any
other observation method at present. The measurements also illustrate the
large sub-pixel variability in NO2 data from present UV–visible satellite
instruments like OMI and that AirMAP observations can be used for detailed
validation of measurements from upcoming missions such as Sentinel 5
precursor having improved spatial resolution.
The data of the AirMAP observations can be obtained from the authors upon request.
Grid points of look-up tables used RTM calculations
ReferencesBaidar, S., Oetjen, H., Coburn, S., Dix, B., Ortega, I., Sinreich, R., and Volkamer, R.: The CU Airborne MAX-DOAS instrument:
vertical profiling of aerosol extinction and trace gases, Atmos. Meas. Tech., 6, 719–739, 10.5194/amt-6-719-2013, 2013.Baumgartner, A., Gege, P., Köhler, C., Lenhard, K., and Schwarzmaier, T.:
Characterisation Methods for the Hyperspectral Sensor HySpex at DLR's
Calibration Home Base, Proc. SPIE, 8533, 85331H–85331H–8,
10.1117/12.974664, 2012.Beirle, S., Boersma, K. F., Platt, U., Lawrence, M. G., and Wagner, T.:
Megacity Emissions and Lifetimes of Nitrogen Oxides Probed from
Space, Science, 333, 1737–1739, 10.1126/science.1207824, 2011.Berg, N., Mellqvist, J., Jalkanen, J.-P., and Balzani, J.: Ship emissions of SO2 and NO2:
DOAS measurements from airborne platforms, Atmos. Meas. Tech., 5, 1085–1098, 10.5194/amt-5-1085-2012, 2012.Bovensmann, H., Burrows, J. P., Buchwitz, M., Frerick, J., Noël, S.,
Rozanov, V. V., Chance, K. V., and Goede, A. P. H.: SCIAMACHY: Mission
Objectives and Measurement Modes, J. Atmos. Sci.,
56, 127–150, 10.1175/1520-0469(1999)056<0127:SMOAMM>2.0.CO;2, 1999.Burrows, J. P., Hölzle, E., Goede, A. P. H., Visser, H., and Fricke, W.:
Earth Observation SCIAMACHY-scanning Imaging Absorption
Spectrometer for Atmospheric Chartography, Acta Astronautica, 35, 445–451,
10.1016/0094-5765(94)00278-T, 1995.Burrows, J. P., Weber, M., Buchwitz, M., Rozanov, V.,
Ladstätter-Weißenmayer, A., Richter, A., DeBeek, R., Hoogen, R.,
Bramstedt, K., Eichmann, K.-U., Eisinger, M., and Perner, D.: The Global
Ozone Monitoring Experiment (GOME): Mission Concept and First
Scientific Results, J. Atmos. Sci., 56, 151–175,
10.1175/1520-0469(1999)056<0151:TGOMEG>2.0.CO;2, 1999.Burrows, J. P., Borrell, P., and Platt, U. (Eds.): The Remote Sensing of
Tropospheric Composition from Space, Physics of Earth and Space
Environments, Springer Berlin Heidelberg, Berlin, Heidelberg,
10.1007/978-3-642-14791-3, 2011.Central Intelligence Agency: The World Fact Book (Romania),
available at:
https://www.cia.gov/library/publications/the-world-factbook/geos/print/country/countrypdf_ro.pdf, last access: 2 May 2016.Chipperfield, M. P.: Multiannual Simulations with a Three-Dimensional Chemical
Transport Model, J. Geophys. Res.-Atmos., 104,
1781–1805, 10.1029/98JD02597, 1999.
Constantin, D.-E., Voiculescu, M., Georgescu, L., Trif, C., Karakolios, E.,
Mamoukaris, A., and Xipolitos, K.: Imprint of Road Vehicle Dynamics on
Atmospheric Pollution. Case Study: Bucharest City
2007–2010, J. Environ. Prot. Ecol., 13,
837–843, 2012.Constantin, D.-E., Merlaud, A., Van Roozendael, M., Voiculescu, M., Fayt, C.,
Hendrick, F., Pinardi, G., and Georgescu, L.: Measurements of Tropospheric
NO2 in Romania Using a Zenith-Sky Mobile DOAS
System and Comparisons with Satellite Observations, Sensors, 13,
3922–3940, 10.3390/s130303922, 2013.D'Amico, G., Amodeo, A., Baars, H., Binietoglou, I., Freudenthaler, V., Mattis,
I., Wandinger, U., and Pappalardo, G.: EARLINET Single Calculus Chain
– Overview on Methodology and Strategy, Atmos. Meas. Tech., 8,
4891–4916, 10.5194/amt-8-4891-2015, 2015.D'Amico, G., Amodeo, A., Mattis, I., Freudenthaler, V., and Pappalardo, G.:
EARLINET Single Calculus Chain – Technical –
Part 1: Pre-Processing of Raw Lidar Data, Atmos. Meas. Tech., 9,
491–507, 10.5194/amt-9-491-2016, 2016.Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P., Kobayashi,
S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P., Bechtold, P.,
Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N., Delsol, C.,
Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S. B.,
Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler, M.,
Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J.-J., Park,
B.-K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.-N., and
Vitart, F.: The ERA-Interim Reanalysis: Configuration and Performance
of the Data Assimilation System, Q. J. Roy. Meteorol. Soc., 137, 553–597, 10.1002/qj.828, 2011.Eskes, H., Huijnen, V., Arola, A., Benedictow, A., Blechschmidt, A.-M., Botek,
E., Boucher, O., Bouarar, I., Chabrillat, S., Cuevas, E., Engelen, R.,
Flentje, H., Gaudel, A., Griesfeller, J., Jones, L., Kapsomenakis, J.,
Katragkou, E., Kinne, S., Langerock, B., Razinger, M., Richter, A., Schultz,
M., Schulz, M., Sudarchikova, N., Thouret, V., Vrekoussis, M., Wagner, A.,
and Zerefos, C.: Validation of Reactive Gases and Aerosols in the MACC
Global Analysis and Forecast System, Geosci. Model Dev., 8, 3523–3543,
10.5194/gmd-8-3523-2015, 2015.European Environment Agency: The European Pollutant Release and
Transfer Register (E-PRTR), Member States Reporting under
Article 7 of Regulation (EC) No 166/2006 –
European Environment Agency,
http://www.eea.europa.eu/ds_resolveuid/710e9a407c0e4fecb376ee87ec9515c1, last access: 9 June
2016.European Environment Agency: Digital Elevation Model over Europe
(EU-DEM),
http://www.eea.europa.eu/data-and-maps/data/ds_resolveuid/ca503256de1b4231b029e4145d0a8b7b (last access: 21 January 2016),
2013.General, S., Pöhler, D., Sihler, H., Bobrowski, N., Frieß, U., Zielcke,
J., Horbanski, M., Shepson, P. B., Stirm, B. H., Simpson, W. R., Weber, K.,
Fischer, C., and Platt, U.: The Heidelberg Airborne Imaging DOAS
Instrument (HAIDI) – a Novel Imaging DOAS Device for
2-D and 3-D Imaging of Trace Gases and Aerosols, Atmos. Meas. Tech.,
7, 3459–3485, 10.5194/amt-7-3459-2014, 2014.Heiden, U., Segl, K., Roessner, S., and Kaufmann, H.: Determination of Robust
Spectral Features for Identification of Urban Surface Materials in
Hyperspectral Remote Sensing Data, Remote Sens. Environ., 111,
537–552, 10.1016/j.rse.2007.04.008, 2007.Herold, M., Roberts, D. A., Gardner, M. E., and Dennison, P. E.: Spectrometry
for Urban Area Remote Sensing-Development and Analysis of a
Spectral Library from 350 to 2400 nm, Remote Sens. Environ., 91,
304–319, 10.1016/j.rse.2004.02.013, 2004.Heue, K.-P., Richter, A., Bruns, M., Burrows, J. P., v. Friedeburg, C., Platt, U., Pundt, I., Wang, P., and Wagner, T.:
Validation of SCIAMACHY tropospheric NO2-columns with AMAXDOAS measurements, Atmos. Chem. Phys., 5, 1039–1051, 10.5194/acp-5-1039-2005, 2005.Heue, K.-P., Wagner, T., Broccardo, S. P., Walter, D., Piketh, S. J., Ross,
K. E., Beirle, S., and Platt, U.: Direct Observation of Two Dimensional Trace
Gas Distributions with an Airborne Imaging DOAS Instrument, Atmos. Chem.
Phys., 8, 6707–6717, 10.5194/acp-8-6707-2008, 2008.
Hilboll, A.: Tropospheric Nitrogen Dioxide from Satellite Measurements:
SCIAMACHY Limb/Nadir Matching and Multi-Instrument Trend Analysis, PhD
thesis, University of Bremen, 2014.Hilboll, A., Richter, A., and Burrows, J. P.: Long-Term Changes of Tropospheric
NO2 over Megacities Derived from Multiple Satellite
Instruments, Atmos. Chem. Phys., 13, 4145–4169,
10.5194/acp-13-4145-2013, 2013a.Hilboll, A., Richter, A., Rozanov, A., Hodnebrog, Ø., Heckel, A., Solberg,
S., Stordal, F., and Burrows, J. P.: Improvements to the Retrieval of
Tropospheric NO2 from Satellite –
Stratospheric Correction Using SCIAMACHY Limb/Nadir Matching and
Comparison to Oslo CTM2 Simulations, Atmos. Meas. Tech., 6, 565–584,
10.5194/amt-6-565-2013, 2013b.Huijnen, V., Eskes, H. J., Poupkou, A., Elbern, H., Boersma, K. F., Foret, G.,
Sofiev, M., Valdebenito, A., Flemming, J., Stein, O., Gross, A., Robertson,
L., D'Isidoro, M., Kioutsioukis, I., Friese, E., Amstrup, B., Bergstrom, R.,
Strunk, A., Vira, J., Zyryanov, D., Maurizi, A., Melas, D., Peuch, V.-H., and
Zerefos, C.: Comparison of OMI NO2 Tropospheric Columns
with an Ensemble of Global and European Regional Air Quality Models,
Atmos. Chem. Phys., 10, 3273–3296,
10.5194/acp-10-3273-2010, 2010.Ibrahim, O., Shaiganfar, R., Sinreich, R., Stein, T., Platt, U., and Wagner,
T.: Car MAX-DOAS Measurements around Entire Cities: Quantification of
NOx Emissions from the Cities of Mannheim and
Ludwigshafen (Germany), Atmos. Meas. Tech., 3, 709–721,
10.5194/amt-3-709-2010, 2010.Itten, K. I., Dell'Endice, F., Hueni, A., Kneubühler, M., Schläpfer,
D., Odermatt, D., Seidel, F., Huber, S., Schopfer, J., Kellenberger, T.,
Bühler, Y., D'Odorico, P., Nieke, J., Alberti, E., and Meuleman, K.:
APEX - the Hyperspectral ESA Airborne Prism Experiment, Sensors, 8,
6235–6259, 10.3390/s8106235, 2008.Kleipool, Q. L., Dobber, M. R., de Haan, J. F., and Levelt, P. F.: Earth
Surface Reflectance Climatology from 3 Years of OMI Data, J. Geophys. Res.-Atmos., 113, D18308, 10.1029/2008JD010290,
2008.
Kurucz, R. L., Furenlid, I., Brault, J., and Testerman, L.: Solar Flux Atlas
from 296 to 1300 nm, US, National Solar Observatory, 1984.Lawrence, J. P., Anand, J. S., Vande Hey, J. D., White, J., Leigh, R. R.,
Monks, P. S., and Leigh, R. J.: High-Resolution Measurements from the
Airborne Atmospheric Nitrogen Dioxide Imager (ANDI), Atmos. Meas.
Tech., 8, 4735–4754, 10.5194/amt-8-4735-2015, 2015.Leitão, J., Richter, A., Vrekoussis, M., Kokhanovsky, A., Zhang, Q. J.,
Beekmann, M., and Burrows, J. P.: On the Improvement of
NO2 Satellite Retrievals – Aerosol Impact
on the Airmass Factors, Atmos. Meas. Tech., 3, 475–493,
10.5194/amt-3-475-2010, 2010.Levelt, P. F., van den Oord, G. H. J., Dobber, M. R., Malkki, A., Visser, H.,
de Vries, J., Stammes, P., Lundell, J. O. V., and Saari, H.: The Ozone
Monitoring Instrument, IEEE T. Geosci. Remote,
44, 1093–1101, 10.1109/TGRS.2006.872333, 2006.Liu, F., Beirle, S., Zhang, Q., Dörner, S., He, K., and Wagner, T.:
NOx Lifetimes and Emissions of Cities and Power Plants
in Polluted Background Estimated by Satellite Observations, Atmos. Chem.
Phys., 16, 5283–5298, 10.5194/acp-16-5283-2016, 2016.
Lohberger, F., Hönninger, G., and Platt, U.: Ground-Based Imaging
Differential Optical Absorption Spectroscopy of Atmospheric Gases, Appl.
Opt., 43, 4711–4717, 2004.Mattis, I., D'Amico, G., Baars, H., Amodeo, A., Madonna, F., and Iarlori, M.:
EARLINET Single Calculus Chain – Technical –
Part 2: Calculation of Optical Products, Atmos. Meas. Tech., 9,
3009–3029, 10.5194/amt-9-3009-2016, 2016.Mendolia, D., D'Souza, R. J. C., Evans, G. J., and Brook, J.: Comparison of
Tropospheric NO2 Vertical Columns in an Urban
Environment Using Satellite, Multi-Axis Differential Optical Absorption
Spectroscopy, and in Situ Measurements, Atmos. Meas. Tech., 6, 2907–2924,
10.5194/amt-6-2907-2013, 2013.Merlaud, A., Van Roozendael, M., van Gent, J., Fayt, C., Maes, J.,
Toledo-Fuentes, X., Ronveaux, O., and De Mazière, M.: DOAS
Measurements of NO2 from an Ultralight Aircraft during
the Earth Challenge Expedition, Atmos. Meas. Tech., 5, 2057–2068,
10.5194/amt-5-2057-2012, 2012.Munro, R., Lang, R., Klaes, D., Poli, G., Retscher, C., Lindstrot, R., Huckle,
R., Lacan, A., Grzegorski, M., Holdak, A., Kokhanovsky, A., Livschitz, J.,
and Eisinger, M.: The GOME-2 Instrument on the Metop Series of
Satellites: Instrument Design, Calibration, and Level 1 Data Processing
– an Overview, Atmos. Meas. Tech., 9, 1279–1301,
10.5194/amt-9-1279-2016, 2016.Oetjen, H., Baidar, S., Krotkov, N. A., Lamsal, L. N., Lechner, M., and
Volkamer, R.: Airborne MAX-DOAS Measurements over California:
Testing the NASA OMI Tropospheric NO2 Product,
J. Geophys. Res.-Atmos., 118, 7400–7413,
10.1002/jgrd.50550, 2013.Platt, U. and Stutz, J.: Differential Optical Absorption Spectroscopy:
Principles and Applications, Physics of Earth and space environments,
Springer, Berlin, 10.1007/978-3-540-75776-4, 2008.Popp, C., Brunner, D., Damm, A., Van Roozendael, M., Fayt, C., and Buchmann,
B.: High-Resolution NO2 Remote Sensing from the
Airborne Prism EXperiment (APEX) Imaging Spectrometer, Atmos.
Meas. Tech., 5, 2211–2225, 10.5194/amt-5-2211-2012, 2012.
Prunet, P., Bacour, C., Price, I., Muller, J.-P., Lewis, P., Vountas, M., von
Hoyningen-Huene, W., Burrows, J. P., Schlundt, C., Bréon, F.-M.,
Gonzales, L., North, P., Fischer, J., and Domenech, C.: A Surface Reflectance
DAtabase for ESA's Earth Observation Missions (ADAM), ESA
Final Report NOV-3895-NT-12403, Noveltis, 2013.Ricchiazzi, P., Yang, S., Gautier, C., and Sowle, D.: SBDART: A
Research and Teaching Software Tool for Plane-Parallel Radiative
Transfer in the Earth's Atmosphere, B. Am.
Meteorol. Soc., 79, 2101–2114,
10.1175/1520-0477(1998)079<2101:SARATS>2.0.CO;2, 1998.Rothman, L. S., Gordon, I. E., Babikov, Y., Barbe, A., Chris Benner, D.,
Bernath, P. F., Birk, M., Bizzocchi, L., Boudon, V., Brown, L. R., Campargue,
A., Chance, K., Cohen, E. A., Coudert, L. H., Devi, V. M., Drouin, B. J.,
Fayt, A., Flaud, J. M., Gamache, R. R., Harrison, J. J., Hartmann, J. M.,
Hill, C., Hodges, J. T., Jacquemart, D., Jolly, A., Lamouroux, J., Le Roy,
R. J., Li, G., Long, D. A., Lyulin, O. M., Mackie, C. J., Massie, S. T.,
Mikhailenko, S., Müller, H. S. P., Naumenko, O. V., Nikitin, A. V.,
Orphal, J., Perevalov, V., Perrin, A., Polovtseva, E. R., Richard, C., Smith,
M. A. H., Starikova, E., Sung, K., Tashkun, S., Tennyson, J., Toon, G. C.,
Tyuterev, V. G., and Wagner, G.: The HITRAN2012 Molecular Spectroscopic
Database, J. Quant. Spectrosc. Ra., 130,
4–50, 10.1016/j.jqsrt.2013.07.002, 2013.Rozanov, V. V. and Rozanov, A. V.: Differential Optical Absorption Spectroscopy
(DOAS) and Air Mass Factor Concept for a Multiply Scattering Vertically
Inhomogeneous Medium: Theoretical Consideration, Atmos. Meas. Tech., 3,
751–780, 10.5194/amt-3-751-2010, 2010.Rozanov, V. V., Rozanov, A. V., Kokhanovsky, A. A., and Burrows, J. P.:
Radiative Transfer through Terrestrial Atmosphere and Ocean: Software
Package SCIATRAN, J. Quant. Spectrosc. Ra., 133, 13–71, 10.1016/j.jqsrt.2013.07.004, 2014.Schönhardt, A., Altube, P., Gerilowski, K., Krautwurst, S., Hartmann, J.,
Meier, A. C., Richter, A., and Burrows, J. P.: A Wide Field-of-View Imaging
DOAS Instrument for Two-Dimensional Trace Gas Mapping from Aircraft,
Atmos. Meas. Tech., 8, 5113–5131, 10.5194/amt-8-5113-2015, 2015.
Seinfeld, J. H. and Pandis, S. N.: Atmospheric Chemistry and Physics: From Air
Pollution to Climate Change, Wiley, 2006.Serdyuchenko, A., Gorshelev, V., Weber, M., Chehade, W., and Burrows, J. P.:
High Spectral Resolution Ozone Absorption Cross-Sections –
Part 2: Temperature Dependence, Atmos. Meas. Tech., 7, 625–636,
10.5194/amt-7-625-2014, 2014.Shaiganfar, R., Beirle, S., Sharma, M., Chauhan, A., Singh, R. P., and Wagner,
T.: Estimation of NOx Emissions from Delhi Using
Car MAX-DOAS Observations and Comparison with OMI Satellite Data,
Atmos. Chem. Phys., 11, 10871–10887, 10.5194/acp-11-10871-2011,
2011.Shaiganfar, R., Beirle, S., Petetin, H., Zhang, Q., Beekmann, M., and Wagner,
T.: New Concepts for the Comparison of Tropospheric
NO2
Column Densities Derived from Car-MAX-DOAS Observations, OMI
Satellite Observations and the Regional Model CHIMERE during Two
MEGAPOLI Campaigns in Paris 2009/10, Atmos. Meas. Tech., 8,
2827–2852, 10.5194/amt-8-2827-2015, 2015.Sinnhuber, B.-M., Weber, M., Amankwah, A., and Burrows, J. P.: Total Ozone
during the Unusual Antarctic Winter of 2002, Geophys. Res.
Lett., 30, 1580, 10.1029/2002GL016798, 2003a.Sinnhuber, M., Burrows, J. P., Chipperfield, M. P., Jackman, C. H., Kallenrode,
M.-B., Künzi, K. F., and Quack, M.: A Model Study of the Impact of
Magnetic Field Structure on Atmospheric Composition during Solar Proton
Events, Geophys. Res. Lett., 30, 1818, 10.1029/2003GL017265,
2003b.Thalman, R. and Volkamer, R.: Temperature Dependent Absorption Cross-Sections
of O2–O2 Collision
Pairs between 340 and 630 nm and at Atmospherically Relevant Pressure,
Phys. Chem. Chem. Phys., 15, 15371–15381,
10.1039/C3CP50968K, 2013.Vandaele, A. C., Hermans, C., Simon, P. C., Carleer, M., Colin, R., Fally, S.,
Merienne, M. F., Jenouvrier, A., and Coquart, B.: Measurements of the
NO2 Absorption Cross-Section from 42000 cm-1 to 10000 cm-1 (238–1000
nm) at 220 K and 294 K, J. Quant. Spectrosc.
Ra., 59, 171–184, 10.1016/S0022-4073(97)00168-4, 1998.Veefkind, J., Aben, I., McMullan, K., Förster, H., de Vries, J., Otter,
G., Claas, J., Eskes, H., de Haan, J., Kleipool, Q., van Weele, M.,
Hasekamp, O., Hoogeveen, R., Landgraf, J., Snel, R., Tol, P., Ingmann, P.,
Voors, R., Kruizinga, B., Vink, R., Visser, H., and Levelt, P.: TROPOMI
on the ESA Sentinel-5 Precursor: A GMES Mission for Global
Observations of the Atmospheric Composition for Climate, Air Quality and
Ozone Layer Applications, Remote Sens. Environ., 120, 70–83,
10.1016/j.rse.2011.09.027, 2012.Vlemmix, T., Hendrick, F., Pinardi, G., De Smedt, I., Fayt, C., Hermans, C.,
Piters, A., Wang, P., Levelt, P., and Van Roozendael, M.: MAX-DOAS
Observations of Aerosols, Formaldehyde and Nitrogen Dioxide in the
Beijing Area: Comparison of Two Profile Retrieval Approaches, Atmos.
Meas. Tech., 8, 941–963, 10.5194/amt-8-941-2015, 2015.Wagner, T., Ibrahim, O., Shaiganfar, R., and Platt, U.: Mobile MAX-DOAS
Observations of Tropospheric Trace Gases, Atmos. Meas. Tech., 3, 129–140,
10.5194/amt-3-129-2010, 2010.Wang, P., Richter, A., Bruns, M., Rozanov, V. V., Burrows, J. P., Heue, K.-P.,
Wagner, T., Pundt, I., and Platt, U.: Measurements of Tropospheric
NO2 with an Airborne Multi-Axis DOAS Instrument,
Atmos. Chem. Phys., 5, 337–343, 10.5194/acp-5-337-2005, 2005.Winkler, H., Sinnhuber, M., Notholt, J., Kallenrode, M.-B., Steinhilber, F.,
Vogt, J., Zieger, B., Glassmeier, K.-H., and Stadelmann, A.: Modeling Impacts
of Geomagnetic Field Variations on Middle Atmospheric Ozone Responses to
Solar Proton Events on Long Timescales, J. Geophys. Res.-Atmos., 113, D02302, 10.1029/2007JD008574, 2008.Zieger, P., Ruhtz, T., Preusker, R., and Fischer, J.: Dual-Aureole and Sun
Spectrometer System for Airborne Measurements of Aerosol Optical Properties,
Appl. Opt., 46, 8542, 10.1364/AO.46.008542, 2007.