We compare the nitric oxide measurements in the mesosphere and lower thermosphere (60 to 150 km) from four instruments: the Atmospheric Chemistry Experiment–Fourier Transform Spectrometer (ACE-FTS), the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS), the SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY), and the Sub-Millimetre Radiometer (SMR). We use the daily zonal mean data in that altitude range for the years 2004–2010 (ACE-FTS), 2005–2012 (MIPAS), 2008–2012 (SCIAMACHY), and 2003–2012 (SMR).

We first compare the data qualitatively with respect to the
morphology, focussing on the major features, and then
compare the time series directly and quantitatively.
In three geographical regions, we compare the vertical density profiles on
coincident measurement days.
Since
none of the instruments delivers continuous daily measurements in this
altitude region, we carried out a multi-linear regression analysis.
This regression analysis considers annual and semi-annual
variability in the form of
harmonic terms and inter-annual variability by responding
linearly to the solar Lyman-

We find that the data sets are consistent and that they only disagree on minor aspects. SMR and ACE-FTS deliver the longest time series in the mesosphere, and they agree with each other remarkably well. The shorter time series from MIPAS and SCIAMACHY also agree with them where they overlap. The data agree within 30 % when the number densities are large, but they can differ by 50 to 100 % in some cases.

Climate models aim to predict the trend of Earth's climate, considering
the composition of the atmosphere.
This composition is influenced by a number of factors, including
anthropogenic emissions and solar variability.
To disentangle these effects, the evaluation of the solar influence
is important.
Solar particles and soft solar X-rays produce nitric oxide (NO) in the
mesosphere and lower thermosphere (MLT, 50–150 km)

To relate atmospheric composition changes to solar activity, global NO measurements over long time periods deliver important information. These data are provided by satellite instruments using different measurement methods. The consistency of these measurements is crucial for using the results for further work, for example to validate climate models and to find climate-relevant forcing parameters. We compare the daily zonal mean NO number densities from four space-borne instruments: the Michelson Interferometer for Passive Atmospheric Sounding (MIPAS, infrared limb emission) and the SCanning Imaging Absorption spectroMeter for Atmospheric CHartographY (SCIAMACHY, UV–vis–NIR limb and nadir emission) on Envisat, the Sub-Millimetre Radiometer (SMR, sub-millimetre limb emission) on the Odin satellite, and the Atmospheric Chemistry Experiment–Fourier Transform Spectrometer (ACE-FTS, infrared solar occultation) on SCISAT. The aim of this comparison is to answer the question of whether zonal mean data sets from different instruments consistently constrain the nitric oxide in the MLT.

Continuous global NO measurements in the MLT region are still rare, and, with the loss of the Envisat satellite in April 2012, two important instruments are missing. With the end of MIPAS and SCIAMACHY measurements, only ACE-FTS and SMR continuously measure NO between 80 and 120 km. From these two, only SMR delivers global data. ACE-FTS, however, offers only limited latitudinal coverage since it observes solar occultations, scanning the atmosphere at sunrise and sunset. There are still more satellite instruments measuring NO in the middle atmosphere: OSIRIS, SABER, and SOFIE. However, OSIRIS measures only between 85 and 100 km, and SABER only above 100 km. SOFIE is another solar occultation instrument and therefore also covers only a limited latitude range, similar to ACE-FTS.

The instruments are introduced in Sect.

Here we briefly introduce the instruments and MLT NO data
sets used in this comparison study.
In particular, we focus on the typical features of the
measurements, such as satellite orbits, spectral ranges,
MLT measurement schedules, retrieval algorithm and references,
as well as uncertainty estimates.
At the end of this section, we summarise the instruments and
data characteristics in Table

Two of the instruments considered here, MIPAS and SCIAMACHY, are on board the now-defunct European research satellite Envisat. This satellite had been orbiting on a sun-synchronous orbit at 800 km and at Equator-crossing times of 10:00 and 22:00 since 2002. Communication to the satellite was lost in April 2012, which is therefore the latest date for which MIPAS and SCIAMACHY data are available.

MIPAS is an
infrared Fourier transform spectrometer. It has a spectral range from 4.15 to 14.6

MIPAS measured atmospheric emissions in a limb observation geometry

The NO data used here were produced using the MIPAS data processor developed at
the Institute of Meteorology and Climate Research (IMK) in cooperation with
the Instituto de Astrofísica de Andalucía (IAA)

NO in the altitude region of interest (70–120 km) is derived
from the fundamental NO band emission at 5.3

The single-profile vertical resolution of NO in the
70–100 km region is 15–20 km, and the single-profile precision ranges from

The retrieval of NO in the upper atmosphere (100–170 km) is described in detail
in

SCIAMACHY

Nitric oxide is retrieved by observing the NO gamma bands with
SCIAMACHY's UV channel 1 (230–314 nm)

The SCIAMACHY retrieval derives NO number densities. Here we use the NO data version 2.0; overall, SCIAMACHY contributes the daytime data from 78 MLT measurement days from 26 July 2008 until 30 March 2012 with about 450 scans per day.

ACE-FTS is a Fourier transform spectrometer on board the Canadian
Atmospheric Chemistry Experiment (ACE) satellite

ACE-FTS retrievals of NO use 36 microwindows between 5.18 and
5.43

Odin is a Swedish-led satellite, in cooperation with the Canadian, French and
Finnish space agencies

SMR is one of the instruments on board this
satellite. It is a limb emission sounder measuring globally a variety of trace
gases and the temperature in the whole middle atmosphere.
SMR uses four sub-millimetre channels (486.1–503.9, 541.0–558.0,
547.0–564.0, 563.0–581.4 GHz) and one millimetre-wave
channel (118.25–119.25 GHz)

We use the data version 2.1 in this study, and SMR contributes NO observations on 301 days from 7 October 2003 to 25 December 2012, with about 600 scans per day.

Instrument overview

Uncertainty estimates; see text for details.

The individual measurements of nitric oxide (NO) from each instrument were
averaged to daily zonal mean values binned into 5

To prepare these data,
volume mixing ratios had to be converted to number densities.
The ACE-FTS VMRs were converted to number density using the density from
the simultaneously retrieved temperatures

The averaged data were additionally filtered by the sensitivity
of the instruments.
The ACE-FTS measurements have been filtered based on the Data Issues List
(

Figures

These figures give an overview of the available data set. ACE-FTS, as a solar occultation instrument, has only limited geographical coverage. MIPAS and SMR have limited sensitivity at altitudes below 85 km, in particular at middle and low latitudes. Additionally, MIPAS data from 75 to 100 km are at present only available from July 2008. The SCIAMACHY data are restricted to daytime measurements, which adversely affects the number of measurements, in particular at high latitudes in the polar winter. SMR data are relatively sparse before 2007, when the Odin astronomy mission ended, and more measurement days have been dedicated to NO observations since then.

The zonal mean data of all four instruments are consistent with respect to the annual variation of the NO density in the MLT region. Throughout the latitude range, the number densities are low at times of low solar activity, 2008 and 2009, and increase with growing solar activity, 2010 and 2011. The NO density increases most in the polar regions and at higher altitudes. Between 95 and 115 km, the density increases also at lower latitudes, in particular after 2011.

NO zonal mean time series at 85 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO zonal mean time series at 105 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

To put the comparisons from Sect.

We first compare the values in the Northern Hemisphere
at 67.5

The results are consistent throughout the altitude range; the largest values are observed between 95 and 105 km, and smaller values below and above these altitudes. Only SMR and ACE-FTS provide data below 100 km for the years 2004 to 2008. Above 100 km, MIPAS contributes some data points from 2005 onwards, and SCIAMACHY data are available only from the mid-2008.

ACE-FTS, SMR, and MIPAS show that the NO number density is correlated with solar activity. From 2004 to 2007, a period of moderate solar activity, the number densities were generally larger than in 2008/2009, when solar activity was low. The NO density increases then again after 2009 with the onset of the next solar cycle. This correlation is visible at all altitudes, and it is particularly strong in the main production region from 95 to 105 km. Unfortunately, the SCIAMACHY data set is too short to show the same correlation. The SCIAMACHY number densities are always on the low side compared to the other instruments. This is less pronounced at 75 km but is clearly visible at altitudes of 85 km and above.

In addition to the overall correlation of the NO densities with the long-term solar activity, the seasonal cycle is clearly visible in the data from all instruments. This annual variation is more pronounced at 85 km, but it is also visible at 105 km.

NO time series comparison of all four instruments at 67.5

Figure

The annual cycle is also visible in all data sets, as it is in the Northern Hemisphere. Again, this cycle is more pronounced at 85 km but is also visible at 105 km. The SCIAMACHY data above 95 km are low compared to the other three instruments but are still within the error range.

Figure

In general, the magnitude of the NO number density in these
regions is smaller than at higher latitudes by a factor of 5 to 10,
in particular at polar winter, as also discussed in

At altitudes above 95 km, the number densities follow the solar cycle activity.
They decline at the beginning and increase again at the end of
the investigated period.
A distinct annual cycle of the NO density is not clearly identifiable
at these latitudes.
This is in contrast to the time series at higher latitudes
and the result of different production mechanisms depending on latitude.
It is already known that, at high latitudes and under auroral
conditions, the production of NO is larger than that at equatorial
latitudes.
The differences between the polar summer and polar
winter regions come from larger photochemical losses
in the summer, leading to smaller NO
densities

NO time series comparison
as in Fig.

The time series from all instruments are consistent in all regions we compared, in particular considering the sometimes large error bars (equal to the 95 % confidence interval of the daily zonal mean). Larger differences indicate short-term variations that are measured by one instrument on a particular day when none of the other instruments observed NO in the upper atmosphere.

The error bars in the figures indicate the statistical error only. Random retrieval errors are most likely much smaller than the natural variability (the latter hence dominating the standard errors of the mean), and biases are very difficult to estimate consistently for all instruments in a bottom-up manner. Here we determine biases between different instruments in a top-down approach.

We obtain the most direct comparison of the individual results by comparing the vertical density profiles. Discrepancies in the profiles provide insight into the characteristic strengths and weaknesses of the individual instruments and their NO retrieval. However, we focus here on daily zonal mean data, and, apart from SCIAMACHY and MIPAS on the same satellite, the local solar times of the measurements differ substantially. In addition, the different distributions of the individual measurement or tangent points make comparing vertical profiles difficult.

To obtain reliable statistics about the profiles,
we first calculate the difference profile for each coincident
5

First we compare the SCIAMACHY data to the other instruments
because it provides the most regular data throughout the
altitude and latitude range (see Figs.

NO vertical profile comparison of the SCIAMACHY NO number
density (

MIPAS and SCIAMACHY share the same satellite, and therefore they performed the most congruent measurements. In addition, their limb scans were scheduled to measure the mesosphere and lower thermosphere during the same orbits once a month. Therefore, we get the best statistics from this pair of instruments.

SCIAMACHY has fewer coincident days with SMR than with MIPAS. We also have to consider the different local times of the measurements, which are 10:00 for SCIAMACHY and between 06:00 and 07:00 for SMR (at the equator). This timing of the SMR measurements makes them susceptible to the NO diurnal cycle and may lead to systematic differences in the measured number densities, in particular in the lower mesosphere. The coincidences with ACE-FTS amount to only about 10 to 20 usable profiles. Since ACE-FTS measures at sunrise and sunset primarily at higher latitudes, the NO diurnal cycle also affects the retrieved number densities.

The patterns of the MIPAS to SCIAMACHY differences in Fig.

SCIAMACHY consistently measures smaller number densities than SMR by 10 to 20 %. The data agree from 80 to 100 km in the northern polar region. At middle and low latitudes, the SCIAMACHY and SMR densities agree at 95 km and at 110 km.

The SCIAMACHY data agree well with the ACE-FTS data in the Northern Hemisphere. In the Southern Hemisphere, the SCIAMACHY measurements are smaller than the ACE-FTS number densities between 90 and 105 km. Both number densities are consistent below 90 km considering the statistical error. At middle and low latitudes the instruments agree within the large error range.

Figure

NO vertical profile comparison of MIPAS NO data

In all three regions, MIPAS measures the largest NO number
densities between 100 and 120 km, between
80 and 120 %
larger than measured by the other instruments.
As seen in the previous section,
MIPAS and SCIAMACHY data agree in all three latitude regions
at altitudes between 120 and 140 km.
Above about 140 km, MIPAS densities are consistently smaller than
SCIAMACHY densities, up to 50 % at around 155 km.
However, they are still consistent considering the uncertainty
of the MIPAS NO densities because of the uncertain amount of
atomic oxygen used in the MIPAS retrieval

Between 80 and 100 km in the southern polar region, the MIPAS and the ACE-FTS data agree well. Below 80 km and above 100 km, in this region, the MIPAS number densities are larger by 50 to 100 %. At middle and low latitudes MIPAS and ACE-FTS have only a few coincident measurement days and even fewer comparable data points when considering the instruments' sensitivity. In the upper usable altitude region, between 95 and 105 km, MIPAS and ACE-FTS are consistent. In the lower altitude region from 65 to 70 km, MIPAS number densities are larger by 50 to 80 %. In the northern polar region, the MIPAS number densities are also larger than the ACE-FTS measurements in the same altitude region. They are smaller than the ACE-FTS number densities between 85 and 100 km.

Compared to SMR, the MIPAS NO number densities are significantly smaller in the southern polar region from 80 to 100 km by about 40 to 50 %. In the northern polar region from 80 to 100 km, the MIPAS number densities are about 30 to 50 % smaller than the SMR data. Above and below, MIPAS and SMR agree within the statistical error. The number densities of each agree well at middle and low latitudes between 90 and 100 km.

Figure

NO vertical profile comparison of the SMR data

In all three regions, the SMR number densities are consistently larger than the SCIAMACHY data above 100 km. We observe the largest differences in the southern polar region, up to 80 %. However, the number densities of each agree in the northern polar region from 80 to 100 km. At middle and low latitudes, the difference between SMR and SCIAMACHY vary between 10 and 40 %, reaching larger but insignificant values between 85 and 90 km.

Compared to MIPAS, the SMR number densities are significantly smaller in all three regions from 100 to 120 km by 20 to 50 %. From 85 to 100 km, the SMR number densities are consistently larger than the MIPAS data, in the southern polar region between 50 and 80 %, and in the northern polar region up to 120 % but with a large uncertainty. At middle and low latitudes, both measurements are consistent between 90 and 105 km.

Compared to ACE-FTS, the SMR number densities are substantially larger at high southern latitudes above 100 km, differing by 100 % at 105 km. ACE-FTS and SMR data agree well between 80 and 100 km. Below 80 km in that region, the SMR number densities differ from the ACE-FTS results between 50 and 80 % but with a large statistical uncertainty. The NO number densities are comparable at middle and low latitudes over the whole altitude range, considering the statistics. The differences in the northern polar region behave similarly to the results at high southern latitudes; the maximum deviation is 150 % at 105 km. Here, the two data sets agree well between 90 and 100 km. The SMR number densities are up to 60 % larger than the ACE-FTS data below 90 km in that region.

Figure

NO vertical profile comparison of the ACE-FTS data

The vertical resolution differs between the instruments; see
Sect.

Here we briefly assess the impact of these different resolutions on the daily zonal mean data. In the first case, we apply the SCIAMACHY averaging kernels to the ACE-FTS densities. In the second case, we apply the MIPAS upper-atmosphere NO (100–170 km) averaging kernels to the SCIAMACHY densities.

First we compare ACE-FTS with and without applied SCIAMACHY averaging kernels
to each other and to SCIAMACHY.
Since the SCIAMACHY retrieval yields number densities,
we applied the respective averaging kernels to the ACE-FTS
number densities after conversion from VMRs.
The medians of the relative differences of the convolved ACE-FTS densities
to the original ACE-FTS and SCIAMACHY densities are shown in
Fig.

We find that at middle and low latitudes the original and convolved ACE-FTS number densities agree within the statistical error of the median. At high northern and southern latitudes, the convolved densities are systematically smaller than the original densities. This improves the agreement between ACE-FTS and SCIAMACHY at high southern latitudes above 90 km, but it adversely affects their consistency at high northern latitudes. The large relative difference at low latitudes between 80 and 88 km is a result of small number densities measured by ACE-FTS in this region. They are about 1 to 2 orders of magnitude smaller than above and below.

NO vertical difference profile of the convolved ACE-FTS
number densities

To conclude, we observe an additional high bias of SCIAMACHY compared to the convolved ACE-FTS data at high southern latitudes below 90 km and at high northern latitudes above 85 km. In almost all regions, the SCIAMACHY number densities are consistently larger than the degraded ACE-FTS number densities.

Next we compare the convolved SCIAMACHY daily zonal mean densities to the original data and to the MIPAS densities whose averaging kernels were applied. Note that we only used 5 days at the end of 2008 for this comparison. The MIPAS V4O upper-atmosphere NO averaging kernels are only defined above 100 km. Thus, they only marginally overlap with ACE-FTS (up to about 107 km) and SMR (up to about 115 km). Therefore, we only compare SCIAMACHY and MIPAS this way. Furthermore, since the SMR vertical resolution is similar to the SCIAMACHY resolution, the results would be comparable.

The MIPAS upper-atmosphere NO processor retrieves

The median of the relative differences are shown in Fig.

NO vertical difference profile of the convolved SCIAMACHY densities

The original and degraded SCIAMACHY densities agree within the
statistical error at high latitudes above about 135 km.
Between 135 and 120 km at high northern latitudes, the original and convolved SCIAMACHY
profiles agree. In this region, they differ by less than

The low vertical resolution of MIPAS
between 100 and 110 km
is also
indicated by a maximum in the width of the averaging kernel
functions shown in

Diurnal variations play a role when comparing measurements
at different local times. At 106 km and 65

Diurnal variations were also investigated by

One would expect that the diurnal variations distinguish in particular
the SMR measurements at early morning local time (06:00) and the
ACE-FTS measurements at sunrise from the MIPAS and SCIAMACHY
measurements at early noon local time (10:00).
According to

Furthermore, at low latitudes (30

The direct quantitative comparison of the NO data of the four instruments is difficult for several reasons. Coincident measurements are sparse and the local times of the individual measurements differ substantially between some of the instruments. The solar UV radiation influences the NO density annually because of the different solar inclination, and inter-annually due to its varying intensity during the 11-year solar cycle. Thus, the NO density can vary substantially from day to day depending on particle precipitation rates, for example at times of high geomagnetic activity.

All instruments scanned the MLT region only on particular single days, for example MIPAS every 10 days and SCIAMACHY every 14 days. In addition, the MIPAS and SCIAMACHY data are only available for the later part of the time period. This makes capturing all variations of NO in the upper atmosphere difficult. To overcome these shortcomings, we carry out a multi-linear regression analysis of the zonal mean data.

We construct a simple transfer function for the NO number
density

The NO number density

The time series of the measurements, the regression result
fitting the data from all instruments simultaneously, and the residuals
are shown in
Figs.

NO time series regression results at 67.5

NO time series regression results at 67.5

NO time series regression results at 67.5

NO time series regression results at 67.5

The solid line in these in the upper panels of these figures is the regression fit using
the composite data from all instruments simultaneously.
The residuals in the lower panels in these figures indicate
that the transfer function

Figure

NO mean residuals of the individual measurements
to the composite fit.
Shown are the values where the regression fit is above the 95 %
significance level determined using the

The values shown in Fig.

We next analyse the coefficients of the individual regressions
to compare the instruments' responses to the harmonic cycles
and in particular to Lyman-

NO regression coefficient

NO regression coefficient

Figures

We observe enhanced Kp coefficients

NO regression coefficient

NO regression coefficient

Figure

Direct comparison of the regression coefficients

Taking into account the instrument sampling patterns, the cross correlations
between these three estimators vary in general within

We note that the transfer function

The residuals shown in Figs.

The different vertical resolutions of MIPAS and
SCIAMACHY, as discussed in Sect.

In this study, we compared the measurements from four instruments, three limb sounders and one solar occultation instrument, using different spectral ranges: infrared, sub-millimetre waves, and ultraviolet. Despite these different methods and accompanying different retrieval strategies, the nitric oxide daily zonal mean densities of all four instruments are consistent during the comparison time period.

We find that the NO number density time series agree well;
almost all data points lie within the statistical error bars
(equal to the 95 % confidence interval) of the daily zonal mean values.
Additionally, considering the different vertical resolutions and
sometimes large systematic uncertainties,
we get a consistent data set.
The remaining differences are most likely caused by
the different MLT measurement schedules and latitude–time
coverage of the instruments.
For example, SCIAMACHY provides only daytime measurements and
therefore less data at high latitudes at polar winter,
where the other instruments observe enhanced NO values.
This biases the SCIAMACHY daily zonal mean NO number densities
to smaller values compared to the other instruments.
We observe this effect clearly in the comparisons shown in
Sect.

The medians of vertical profile differences in three geographic
regions (90–50

We also performed the regression analysis on an individual instrument basis and obtained consistent coefficients in the important altitude region. In particular, we found consistent responses to the estimators related to solar and geomagnetic variability. These estimators can be further used as an empirical model of NO in the middle atmosphere, which is particularly useful for climate models.

NO zonal mean time series at 75 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO zonal mean time series at 85 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO zonal mean time series at 95 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO zonal mean time series at 105 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO zonal mean time series at 115 km from ACE-FTS, MIPAS, SCIAMACHY, and SMR (from top to bottom).

NO time series comparison of all four instruments at 67.5

NO time series comparison of all four instruments at 67.5

NO time series comparison of all four instruments at 32.5

NO time series comparison of all four instruments at 2.5

NO time series regression results at 67.5

NO time series regression results at 67.5

NO time series regression results at 32.5

NO time series regression results at 2.5

S. Bender and M. Sinnhuber thank the Helmholtz Society for funding this project under the grant number VH-NG-624. The IAA team (M. López-Puertas and B. Funke) were supported by the Spanish MINECO under grant AYA2011-23552 and EC FEDER funds. The SCIAMACHY project was funded by the German Aerospace Center (DLR), the Dutch Space Agency, SNO, and the Belgian Science Policy Office (BELSPO). ESA funded the Envisat project. The University of Bremen as the principal investigator has led the scientific support and development of SCIAMACHY and the scientific exploitation of its data products. The Atmospheric Chemistry Experiment (ACE), also known as SCISAT, is a Canadian-led mission mainly supported by the Canadian Space Agency and the Natural Sciences and Engineering Research Council of Canada. Odin is a Swedish-led satellite project funded jointly by Sweden (SNSB), Canada (CSA), Finland (TEKES), France (CNES) and the Third-Party Missions programme of the European Space Agency (ESA). The provision of MIPAS level-1b data by ESA is gratefully acknowledged. We acknowledge support by the Deutsche Forschungsgemeinschaft and the Open Access Publishing Fund of the Karlsruhe Institute of Technology. The article processing charges for this open-access publication were covered by a Research Centre of the Helmholtz Association. Edited by: M. Riese