We present a new MAX-DOAS profiling algorithm for aerosols and trace gases,
BOREAS, which utilizes an iterative solution method including Tikhonov
regularization and the optimal estimation technique. The aerosol profile
retrieval is based on a novel approach in which the absorption depth of
Performance tests are separated into two parts. First, we address the general sensitivity of the retrieval to the example of synthetic data calculated with the radiative transfer model SCIATRAN. In the second part of the study, we demonstrate BOREAS profiling accuracy by validating the results with the help of ancillary measurements carried out during the CINDI-2 campaign in Cabauw, the Netherlands, in 2016.
The synthetic sensitivity tests indicate that the regularization between
measurement and a priori constraints is insufficient when knowledge of the
true state of the atmosphere is poor. We demonstrate a priori pre-scaling and
extensive regularization tests as a tool for the optimization of retrieved
profiles. The comparison of retrieval results with in situ, ceilometer,
Aerosols and trace gases play an important role for life on Earth as high
concentrations have adverse impacts on human health and climate. Furthermore,
aerosols impact the Earth's energy budget by radiative forcing. They interact
with solar radiation by scattering and absorption processes. Additionally,
they have an impact on the formation of clouds when acting as cloud
condensation nuclei (CCN). Despite considerable efforts by the scientific
community, uncertainties in aerosol radiative forcing are still high
In addition to aerosols, atmospheric trace gases are of importance in an
increasingly urbanized world as they impact human health, agriculture,
acid deposition and climate (e.g.
For more than 15 years, Multi-Axis Differential Optical Absorption
Spectroscopy (MAX-DOAS) measurements have been used to investigate the
chemical composition of the troposphere
The strength of the absorption signal detected in scattered sunlight depends
on the absorber amounts and their vertical distribution but also on the
length of the light path. In general, this length in a certain altitude layer
is a function of the measurement geometry. Therefore, a set of MAX-DOAS
radiance measurements taken at different elevation angles (lines of sight,
LOSs) contains information about the vertical distribution of trace gases,
which can be retrieved. However, the retrieval of absorber profiles from
MAX-DOAS measurements is an ill-posed problem that needs additional
constraints (see Sect.
Profile retrieval algorithms used in the scientific community are either
inversion algorithms or parameterized approaches. Inversion algorithms
directly link the measurement quantity to the vertical profile of the target
absorber with the help of a forward model
Parameterization, on the other hand, means that a forward model, based on a
limited set of parameters, is used to describe the measurement quantity.
Frequently used forward model parameters are integrated values and profile
information such as shape and height of certain absorber layers
Inversion algorithms have the advantage that they are not limited to the scenarios used when creating the LUT but unrealistic profiles are possible when the measurement, inversion or regularization (weighting between the information from measurements and a priori information) is poor. On the other hand, parameterized approaches evaluate profiles much faster, as the slow forward model computation was already done when creating the LUT.
Although efforts to derive concentration profiles from MAX-DOAS
measurements have been made for more than one decade, profiling is still a
difficult task and the results of different algorithms can differ strongly when
the absorber of interest is highly variable and inhomogeneous on spatial and
temporal scales
The purpose of this study is the introduction of IUP Bremen's new MAX-DOAS
profile retrieval algorithm for aerosols and trace gases BOREAS (Bremen
Optimal estimation REtrieval for Aerosols and trace gaseS), which has been
developed to improve the earlier profile retrieval algorithm
The development of BOREAS is aimed at several key properties:
Flexibility: the algorithm should retrieve aerosol and trace gas profiles
from any MAX-DOAS measurement with pre-filtering options for the data. Accuracy and stability: the algorithm should be stable in terms of varying
atmospheric conditions when retrieving profiles for several years of data. The profiling
results (modelled observation) should fit the measured observations with a high accuracy. Automation: the retrieval should respond to problematic data/settings
(e.g. low information content, wrong regularization) automatically with an included problem solution scheme. Fast: the algorithm should be fast enough to allow near-real-time profile
retrievals.
The paper is organized as follows: in the first section, a typical
MAX-DOAS measurement and its information content are shortly described. The
next section focuses on the theoretical background of the retrieval
algorithm. This section is followed by a detailed sensitivity study. The
fourth section is divided into an analysis of synthetic data, a discussion of error
sources and an example of application to real measurements, which shows that
BOREAS is able to retrieve profiles with high reliability and accuracy. The
final section (Sect. 5) concludes and summarizes the results presented in
this study.
Modern ground-based MAX-DOAS instruments are capable of measuring scattered
sunlight in the ultraviolet (UV) and visible (VIS) spectral range with a full
azimuthal and elevation angle coverage of the hemisphere (see
Fig.
Schematic representation of typical MAX-DOAS measurement geometries.
In general, the spectral least-squares fit of absorption cross sections and a
polynomial to the slant optical thickness (logarithm of a Sun-normalized
radiance at a certain LOS, also called slant optical thickness, SOT)
When using a single zenith spectrum measured around noon as
The ratio of slant column density
Because of the altitude-dependent concentration of absorbers, there is a need
for a similar quantity describing the ratio of
The profile retrieval algorithm BOREAS was developed in order to retrieve
aerosol and trace gas vertical profiles from MAX-DOAS measurements. The
aerosol retrieval is fully implemented within the RTM SCIATRAN
BOREAS analysis flow chart. The algorithm is separated into two steps. Step 1: aerosol retrieval within a Tikhonov regularization. Step 2: optimal-estimation-based trace gas retrieval.
The standard DOAS fit does not provide direct information about aerosols
present in the atmosphere. However, the scattering and absorption properties
of aerosols have an impact on the measured differential slant column density
If
Generally, the vertical profiles of species in the troposphere are unknown
because of the temporal and spatial variability of emission sources,
transport and conversion processes. However, the oxygen monomer
In the BOREAS aerosol retrieval algorithm, the difference between modelled
and measured
The inverse problem with respect to the aerosol optical depth is then
formulated as
Since
It follows that the weighting function provides a linear relationship between
the variation of DSOT and the variation of an aerosol number density
The solution of the minimization problem given by Eq. (
The minimization problem can be reformulated in the following vector–matrix
form:
The minimization problem given by Eq. (
There are three criteria to stop the iteration process:
Convergence in parameter space, i.e. the maximum difference between the
components of the state vector at two subsequent iterative steps does not exceed the
selected criterion (e.g. 0.0001 km The root mean square difference between measured and simulated The maximum number of iterations is reached.
In addition to the equations above, here we introduce other quantities which
are useful for describing the retrieval. The gain matrix
SCIATRAN settings for the calculation of differential slant column densities.
Compared to aerosols, the inverse problem for trace gases is easier to solve
because, under the assumption of an optically thin atmosphere, the
relationship between trace gas concentration and measured differential slant
column density is linear. Then, the forward model
From the perspective of MAX-DOAS, this relationship is strongly ill posed as
the number of retrieval parameters is usually much higher than the number of
measurements. Furthermore, as the light paths of two geometrically close
elevation angles traverse similar vertical layers near the surface,
measurements cannot be considered independent. Consequently, the retrieval
parameter vector
In this section, we first present and discuss the results of synthetic
sensitivity studies. For this purpose, noise-free differential slant columns
densities of different aerosol and
Parameters for aerosol and
Exponential aerosol profiles for the synthetic sensitivity study. Also shown is the a priori profile for the aerosol retrieval (dashed line).
A synthetic data set of differential slant column densities of
Retrieval results with a fixed
The trace gas and aerosol profiles in Table
Figure
The retrieval uses vertical grid steps of 50 m for synthetic data because
smaller step widths introduce retrieval noise, whereas larger steps result in
vertical structure being overly smoothed. Since the retrieval tends to
overestimate the upper layers due to missing sensitivity (see
Fig.
On the left side of Fig.
True and retrieved AOT for SNR
Variation of the relative difference in AOT and bottom extinction (BOT) compared to the true value as a function of SNR and Tikhonov parameter, colour coded for scenario E3 with a priori pre-scaling.
Several authors have already highlighted retrieval problems from MAX-DOAS
measurements under highly variable (in time and space) atmospheric conditions
A comparison of all three profiles and their optimal
For scenario E3, high
In contrast to this pragmatic approach, the so-called
The optimal-estimation-based trace gas profile retrieval faces similar
problems to the regularization difficulties discussed for the aerosol
retrieval in the previous section. In many studies, an a priori variance of a
fixed percentage of the a priori profile is used in combination with the differential
slant column density errors from the DOAS fit to constrain the
solution
Again, a priori pre-scaling is used to achieve a better estimate of the true
atmospheric conditions. In contrast to the aerosol retrieval, which needed a
pre-run for a better first guess, the trace gas a priori pre-scaling is
achieved using the 30
The different
Exponential
Figure
The results again show the importance of pre-scaling. Whereas the E2 scenario
is well retrieved for all
Retrieval results with a fixed
Retrieval errors for E1–E3 of the synthetic aerosol
The specific optimal
The total error of both aerosol and trace gas retrieval can be separated into
three different error sources
Figure
The total errors for both aerosol and trace gas retrieval are dominated by
the smoothing error with negligibly small retrieval noise. This shows that
the measurement itself can be considered a small error source in contrast
to the generally limited information content of MAX-DOAS measurements. Note
that the error components of the
Averaging kernels of the aerosol retrievals for scenarios E1–E3 (first three figures on the left side). Also shown are the area and FWHM of the AK on the right side. FWHM values are depicted on their nominal heights (coloured) and on the height of their individual peaks (black).
For the quantification of the vertical sensitivity of the retrieval,
Fig.
The trace of the averaging kernel matrix (degrees of freedom,
DOFs) is often discussed as if it
is the number of independent pieces of information for a retrieval.
In Fig.
Rms between measured and simulated
In Fig.
Averaging kernels of the
In addition to the above mentioned errors and resolution properties, further
error sources and problems should be noted. For real measurements, the true
meteorological conditions are mostly unknown.
The 2nd Cabauw Intercomparison of Nitrogen Dioxide Measuring Instruments
campaign (CINDI-2) took place in Cabauw (the Netherlands) from
25 August to 7 October 2016 and was funded by the European Space Agency (ESA)
and the participating research groups. The campaign goals were the
characterization of differences between
The measurement site Cabauw is located in a rural region dominated by
agriculture but is surrounded by four of the largest cities in the
Netherlands (Rotterdam, Amsterdam, Den Haag and Utrecht). Thus, depending on
the wind direction, long-range transport from highly industrialized areas is
likely, which results in high-pollution events. Here, 3 of the 5
investigated days show a more or less steady wind from south-easterly
directions, with a change in wind direction in the evening of 15 September
(see Fig.
Wind speed and direction for the investigated days
In this study, the instrumental set-up for the validation of aerosol and
trace gas profiles consists of in situ and remote-sensing instruments.
Near-surface concentration and extinction values are provided by a ceilometer,
The profile validation is realized on 2 cloud-free days
(13–14 September 2016) and 3 days with broken clouds (15 September and
23–24 September 2016). The operators of MAX-DOAS instruments were asked to
perform elevation scans at the beginning of each hour and at 11:15 and
11:45 UTC every day. The ancillary measurements introduced in the following
subsections were resampled or averaged in this time interval to prevent
discrepancies due to time lags between two observations. Each scan took
around 8 min and consisted of the following elevation angles: 1,
2, 3, 4, 5, 6, 8, 15 and 30
In this section, BOREAS retrievals were performed on data from the IUPB
MAX-DOAS instrument on a 100 m step width vertical grid from the surface up
to 4 km. The aerosol retrieval uses the a priori pre-scaling option with an
exponential a priori profile described by a 1 km scale height and an aerosol
optical thickness of 0.18. The asymmetry factor and single-scattering albedo were
chosen to be 0.92 and 0.68 respectively. The SNR was set to 2500 with a
Tikhonov parameter of 2. The a priori variance decreased with altitude
from 1.5 at the surface to 0.01 at 4 km. This definition was chosen to
improve the profiling results for lower altitudes, where the main aerosol
load/concentration can be expected, and in order to suppress instabilities at
the upper grid boundary, where the sensitivity is lowest. The trace gas
retrieval again uses a priori pre-scaling with an exponentially decreasing
profile with a scale height of 1 km and a vertical column density of
Aerosol profiles were validated using three different instruments. The
retrieved AOT was compared with values from an AERONET station (AErosol
RObotic NETwork,
The bottom extinction coefficient of the retrieved aerosol profiles was
compared with in situ PM
Furthermore, we used AERONET-scaled ceilometer near-surface extinction as a further validation data set. The ceilometer (CHM15k Nimbus) was operated by the Royal Netherlands Meteorological Institute (KNMI) and sampled backscattering signals every hour at a wavelength of 1064 nm. The integrated backscattering signal was divided by the 1020 nm AERONET AOD to get a conversion factor which was applied to the backscattering signal for the conversion into extinction coefficients. This new ceilometer profile was again scaled with AERONET AOT at 477 nm. The error is expressed as the standard deviation calculated from the temporal and vertical averaging.
Figure
In the scatter plot, BOREAS error bars were calculated by integrating the
total error (Eq.
In the evening, the results of both instruments vary more. Different reasons
for this finding are possible. First, a developing planetary boundary layer
(PBL) might exceed the vertical extension where BOREAS has sufficient
sensitivity. Second, when pointing towards the Sun, saturation of the CCD
becomes a problem, leading to low integration times, which decrease SNR and
fitting quality. In addition, RTM calculations might introduce uncertainties
at high SZA, when the aerosol load is high and the light path is strongly
increased. Furthermore, the aerosol phase function used leads to
uncertainties, as the forward scattering peak might be underestimated by the
Henyey–Greenstein parameterization. The other days show a higher variability
due to an increase in cloudy scenes. Note that, especially in the morning and
around noon, clouds may influence the instruments in various ways because of
their different azimuthal viewing directions. The BOREAS profile retrieval
cannot be considered reliable when one or several elevation angles point at
clouds during the measurements, as intensity and light path vary in
unpredictable ways. This might also lead to large iteration numbers or to no
convergence at all. On 15 September, the temporal patterns of the two
instruments differ strongly. Although the extinction values around noon agree
well, AERONET's AOT decrease before and the
increase afterwards cannot be found this strongly by BOREAS. Especially in
the evening, both curves deviate, indicating highly variable atmospheric
conditions, which might also be introduced by a change in wind direction (see
Fig.
The correlation coefficient of 0.83 is high and shows the general good agreement on 3 of 5 days. The regression line indicates the above-mentioned small underestimation of the AOT by BOREAS.
Figure
The correlation of BOREAS bottom extinctions with data from both validation instruments is high, indicating good agreement on cloud-free days. The correlation with ceilometer near-surface values is high as well but the regression line shows a general underestimation of ceilometer near-surface values in comparison to BOREAS.
Figure
Ceilometer aerosol layer height within the planetary boundary layer.
The underlying extinction coefficient profiles for the above-discussed bottom
extinction and AOT values can be seen in Fig.
On 15 September, the averaged ceilometer data show high and thick clouds in the morning and a rising PBL in the afternoon. In the AK-smoothed data, these clouds cannot be identified any more. BOREAS introduces elevated aerosol layers which can be understood as a retrieval artefact due to these cloudy scenes. In the afternoon, BOREAS finds an upward expansion in the PBL similar to the ceilometer, with the exception of the last profile, which was influenced by the telescope pointing towards the Sun. BOREAS extinction values are smaller in the PBL, which is a consequence of the already explained underestimation of BOREAS AOT to AERONETs AOT, which was used for the backscatter signal scaling.
Comparison of aerosol extinction coefficient profiles from BOREAS
and ceilometer for 13 September 2016
Several instruments were used for the validation of BOREAS nitrogen dioxide
profiling results. The VCD results were validated with the
help of a Pandora instrument (no. 128) operated by LuftBlick
The Pandora instrument was used with its direct Sun capability to retrieve
total nitrogen dioxide columns on hourly time steps (see
Fig.
The
Near-surface concentration validation was done with the help of in situ
(NAQMN, ICAD, CAPS) and remote-sensing instruments (lidar and LP DOAS). The
in situ samplers NAQMN, CAPS and ICAD were operated by RIVM, the Royal
Belgian Institute for Space Aeronomy (BIRA-IASB) and the Institute of
Environmental Physics in Heidelberg (IUPH) respectively. In addition to the
aerosol bottom concentration, NAQMN also provided
In addition to the in situ instruments, a long-path DOAS (LP DOAS) provided
by IUPH
Nitrogen dioxide profiles were provided only by the
Figure
The correlation with Pandora is slightly better than with lidar but the regression parameter shows that Pandora VCDs are higher than those from BOREAS, whereas the lidar regression is closer to 1.
In Fig.
BOREAS surface concentrations agree very well with all data sets except during the morning hours. On 13 September, in addition to the ICAD/CAPS data at the 27 m level (blue), the 200 m points are shown (green). The anti-correlated behaviour in the morning hours confirms the strong inhomogeneity and proves that nitrogen dioxide was mainly concentrated close to the surface at that time. The mean values of both CAPS instruments (yellow line) show a better agreement with BOREAS than the 27 and 200 m data.
It seems that BOREAS cannot fully resolve this thin near-surface layer and
retrieves rather smooth profiles instead of sharp concentration peaks (see
also the discussion about vertical sensitivity in Sect.
The correlations are high for all instruments with the highest value for
NAQMN and the
Figure
Comparison of
In this study, we introduced BOREAS, a new profiling algorithm for MAX-DOAS
measurements. BOREAS retrieves aerosol extinction coefficient profiles around
The retrieval performance was demonstrated on the example of synthetic data
calculated with SCIATRAN and with real measurements taken during the
CINDI-2 campaign in Cabauw 2016. The synthetic scenarios were chosen to work
as a stress test for BOREAS. One fixed a priori profile was used for the
retrieval of different large aerosol loads and
In the second part of this study, BOREAS retrieval results from measurements
of the IUPB MAX-DOAS instrument during the CINDI-2 campaign were validated
with ancillary measurements. The agreement for all parameters was good with
correlations equal or higher than 0.75. Systematic offsets or deviations
depending on geometry, atmospheric condition and investigated air mass
indicate limitations due to the specific measurement characteristics, which
were also found in earlier MAX-DOAS profiling studies. The correlation with
AERONET AOT is 0.83 with an underestimation by BOREAS, leading to a
regression slope of 1.30. The bottom value correlation is 0.75 for the
ceilometer near-surface extinction and 0.81 for NAQMN in situ measurements.
The total column of
As a conclusion, BOREAS profiling capabilities are strong, which was proven by
high correlations with all validating instruments. Discrepancies were found
due to different azimuthal viewing directions of the instruments and the
limited vertical resolution of MAX-DOAS profiling when aerosol load or
concentration is close to the surface in layers thinner than the BOREAS
vertical resolution. A comparison of BOREAS results to the performance of other
MAX-DOAS profiling algorithms will be discussed in two upcoming papers
The profiling data are available upon request (contact persons are Tim Bösch and Andreas Richter). The individual data sets of ancillary instruments are available from the individual PIs upon request.
Box aerosol profiles for the synthetic sensitivity study. Also shown is the a priori profile for the aerosol retrieval (dashed line).
The retrieval results for the box scenarios in Fig.
Retrieval results with a fixed
Variation of SNR and Tikhonov parameter with the relative differences in AOT and bottom extinction compared to the true value, colour coded for scenario B1 with a priori pre-scaling.
Same as Fig.
Same as Fig.
Figures
Figures
Same as Fig.
Same as Fig.
Figure
Box
Retrieval results with a fixed
The authors declare that they have no conflict of interest.
This study was supported by the University of Bremen and by the FP7 Project Quality Assurance for Essential Climate Variables (QA4ECV), no. 607405. Participation of the University of Bremen in the CINDI-2 campaign received funding from ESA via RFQ/3-14594/16/I-SBo. The authors would like to thank the CINDI-2 team for excellent support on site. Furthermore, we thank François Hendrick and Marc Allaart for the provision of mean pressure and temperature profiles used within the BOREAS retrieval of CINDI2 data.The article processing charges for this open-access publication were covered by the University of Bremen.Edited by: Michel Van Roozendael Reviewed by: three anonymous referees