Global distribution and variability of formic acid as observed by MIPAS-ENVISAT



[1] Formic acid (HCOOH) vertical profiles have been retrieved from MIPAS-ENVISAT satellite data in the upper troposphere for the first time. Based on new spectroscopic line-strength measurements by Vander Auwera et al. (2007) of HCOOH, a retrieval setup was developed and optimized to study its global distribution between 2002 and 2008. A strong seasonality, directly associated to plant growth and corresponding biogenic emissions, is observed and dominates in the middle latitudes of the Northern Hemisphere. Here, the mean monthly volume mixing ratios (VMR) at 8 km altitude typically reach 100–110 parts per trillion in volume (pptv) during the summer and decrease to about 45 pptv in the early winter. At 16 km and higher altitudes, the VMRs remain under the 20 pptv level and have a much smaller amplitude (<10 pptv). In the Southern Hemisphere, strong signals (up to 1 ppbv at 10 km in a single measurement) are detected from biomass burning during the August–October time period and can enhance the monthly mean background levels above specific tropical and midlatitudinal regions by a factor of 2 or more. In-plume production of HCOOH through photochemical processes has been identified during an extreme event in September 2006, although it is not likely to contribute significantly to the overall upper tropospheric abundances of formic acid.

1. Introduction

[2] Organic acids are important in atmospheric chemistry since they contribute to the acidity in precipitation and cloud water, particularly in remote areas [Keene and Galloway, 1988; Khare et al., 1999]. They are also known to play important roles in gas- and aqueous-phase chemistry by their oxidative capacity and, together with other oxygenated organic species, they are present in concentrations that are comparable to those of the nonmethane hydrocarbons [Khare et al., 1999; Singh et al., 2000]. Even though sulfuric and nitric acids are known to dominate the pH dependence of wet deposition in the polluted regions, the carboxylic acids (mainly formic and acetic) can contribute to about 30% of this acidification in polluted regions and over 60% in the remote areas [Keene and Galloway, 1988; Andreae et al., 1988].

[3] The main mechanisms recognized as sources of formic acid in the atmosphere are by direct emission (both biogenic and anthropogenic) and by the chemical transformation of precursors. Although their relative contributions are not known with certainty, there have been several investigations showing the importance of emissions from vegetation and soils [Keene and Galloway, 1988; Talbot et al., 1988; Andreae et al., 1988; Seco et al., 2007], as well as from biomass burning [Helas et al., 1992; Lefer et al., 1994] which significantly influence the atmospheric abundance of HCOOH. It has also been shown that the chemical oxidation of hydrocarbons could be the main source of formic acid in the atmosphere [Jacob and Wofsy, 1988; Talbot et al., 1990; Madronich et al., 1990; Sanhueza et al., 1996], via ozonolysis of isoprene, for example, but also from other precursors. Estimates for the lifetime of HCOOH range from several hours in the boundary layer to a few weeks in the free troposphere with wet and dry deposition as its primary sinks and reactions with OH of lesser importance [Jacob and Wofsy, 1988; Keene and Galloway, 1988; Hartmann et al., 1991].

[4] Formic and acetic acids have been widely measured within the boundary layer and their mixing ratios show large variations in the different environments ranging between 0.05 and 16 ppbv, as reviewed by Chebbi and Carlier [1996], Khare et al. [1997], and Seco et al. [2007]. Middle- and upper tropospheric measurements are scarce. However, during some aircraft studies such as one over the North American Arctic and sub-Arctic regions, volume mixing ratios (VMRs) of 166 ± 81 pptv during the summer were measured [Talbot et al., 1992]. These authors report a remote background concentration of 70 pptv for “clean” remote air between 2 and 7 km. Singh et al. [2000] have reported mean VMRs of 54 and 120 pptv for the upper and lower troposphere, respectively, in midlatitudes over the Atlantic. Reiner et al. [1999] measured HCOOH over Germany with a triple quadrupole mass spectrometer (TQMS). Their mean mixing ratios at 7 to 12 km altitudes from the five flights carried out during September of 1991 lie between 60 and 560 pptv. Other aircraft missions have measured boundary layer VMRs in the hundreds of ppt to low ppb levels [Talbot et al., 1990; Klemm et al., 1994; Chapman et al., 1995].

[5] As far as the spectroscopic detection of HCOOH in the atmosphere is concerned, the reported VMRs at the different locations and altitudes is limited by the accuracy of the absorption cross sections used for the analysis which were available at the time. The absolute line intensities of the ν6 and ν8 vibrational bands of HCOOH centered at 1105.4 and 1133.5 cm−1 have been recently updated by Vander Auwera et al. [2007] and Perrin and Vander Auwera [2007], taking into account the dissociation of the formic acid dimer. The integrated line intensity reported in that work compares well with other studies [Sharpe et al., 2004; Notholt et al., 1991] but differs significantly with the 2004 version of HITRAN [Rothman et al., 2005], thus the 2007 update of HITRAN that contains the line parameters by Vander Auwera et al. [2007] was used in this investigation.

[6] HCOOH was first identified in the atmosphere from infrared (IR) solar spectra by Goldman et al. [1984] during a balloon flight in 1981. The ν6 and ν8 vibrational bands of HCOOH were then measured and identified from ground-based FTIR spectra in Kitt Peak [Bumgarner et al., 1988] and further investigated by Rinsland et al. [2004]. Formic acid over Europe was also identified from the balloon-borne MIPAS-B2 instrument at 7.5 and 10.4 km altitudes launched in 1998 [Remedios et al., 2007]. HCOOH and other organic species have also been measured from aircraft in the free troposphere by in situ instruments based on FTIR [Goode et al., 2000] and TDLAS [Herndon et al., 2007] flying over biomass burning plumes. The first space-borne detection of HCOOH is reported by Rinsland et al. [2006] from the Atmospheric Chemistry Experiment-Fourier transform spectrometer (ACE-FTS), which records solar occultation spectra in the infrared. Further studies have followed [Coheur et al., 2007; González Abad et al., 2009], improving both the analysis and the coverage. Finally, analysis of IR radiances from the nadir-viewing instrument IASI, on board the MetOp satellite, has also been successful in detecting enhancements of HCOOH and many other compounds [Clerbaux et al., 2009; Coheur et al., 2009].

[7] In this investigation, HCOOH was retrieved from the limb-infrared spectra measured by the Michelson Interferometer for Passive Atmospheric Sounding-European environmental satellite (MIPAS-ENVISAT) instrument. Data were processed between 2002 and 2008 from which the global distribution and the seasonal variability of upper tropospheric formic acid in specific regions is reported.

2. Experimental Procedure

[8] MIPAS is a limb emission spectrometer measuring the infrared radiation from the Earth's atmosphere [European Space Agency, 2000; Fischer et al., 2008]. MIPAS was launched on board the ENVISAT on 1 March 2002 and has been on a sun-synchronous polar orbit at about 800 km altitude since. The instrument has been operating in different measurement modes with an interruption of approximately 9 months between March and December of 2004. The two modes used for the analysis of HCOOH in this study are summarized in Table 1. The main difference is a reduction in spectral resolution from 0.025 to 0.0625 cm−1 (unapodized) and an improvement in the spatial resolution going from FR (full resolution) to RR (reduced resolution) measurement modes. The latter results in a vertical oversampling since the instantaneous field of view of MIPAS at the tangent point is about 3 km [von Clarmann et al., 2009]. Also, the number of limb sequences measured per orbit has been increased, as can be seen from Table 1. The spectral coverage of the instrument is 685 to 2410 cm−1 measured in five channels.

Table 1. Measurement Modes of MIPAS-ENVISAT Used in This Studya
Measurement ModesFRRR
  • a

    FR = full resolution; RR = reduced resolution nominal modes.

Time period07/2002–03/200401/2005–present
Spectral resolution (unapod.)0.025 cm−10.0625 cm−1
No. of tangent heights (range)17 (6–68 km)27 (6–70 km)
Scan steps heights (range)3 (6–42) km1.5 (6–21), 2.0 (21–31) km
No. of limb sequences/orbit7296

[9] The retrieval strategy consists of a multiparameter nonlinear least-squares fitting of measured and modeled spectra [von Clarmann et al., 2003]. The calibrated level-1B radiance spectra from the European Space Agency (ESA) are modeled with the Karlsruhe optimized and precise radiative algorithm (KOPRA) described by Stiller [2000] and the retrieval, performed on a 1 km grid up to 44 km and 2 km above, uses a Tikhonov-type regularization scheme with a first-order finite differences constraint as in similar retrievals from MIPAS data [see, e.g., Steck, 2002; Glatthor et al., 2007; von Clarmann et al., 2007]. Spectra contaminated by cloud or strong aerosol signal are discarded from the analysis. Cloud identification is based on the radiance contrast between the spectral ranges 788.2–796.25 cm−1 and 832.3–834.4 cm−1, following the approach of Spang et al. [2004]. For our analysis a threshold of 4.0 was chosen (i.e., spectra where this ratio was below 4.0 were discarded). Optically thin aerosols that passed this cloud filter were considered in the retrieval by means of an empirical background continuum that was jointly fitted to the spectra along with the target variables (for details of this approach see von Clarmann et al. [2003]).

[10] The strongest emission of HCOOH centered near 1005 cm−1 due to the Q branch of its ν6 vibrational mode has been chosen for the retrievals. The line intensities of this absorption feature, which has been widely used before for atmospheric sounding, were recently updated [Perrin and Vander Auwera, 2007; Vander Auwera et al., 2007] and are considerably lower (up to a factor of 2) than those reported in previous studies. The retrievals were done within the two microwindows shown in Figure 1 from 1103.5 to 1106.1 and 1112.5 to 1116.9 cm−1; ozone (O3), dichlorodifluoromethane (CFC-12), and chlorodifluoromethane (HCFC-22) were jointly fitted. Other parameters (spectral shift, temperature, and tangent heights) as well as a series of trace gases (H2O, N2O, CH4, HNO3, CFC-11, and ClONO2) were fitted in advance and the respective results were used for the subsequent HCOOH retrieval. A possible influence of HDO on the retrieved profiles was investigated as suggested by Rinsland et al. [2004]. The assumed and fixed abundance of the HDO isotopolog from its standard mean ocean value was doubled and halved to establish its impact on the retrieved profiles of HCOOH. No significant influences were obtained on the results (<2% change in the upper troposphere) and hence the HDO was considered in the radiative transfer calculations with a standard ratio relative to the retrieved main isotopologe H2O. A zero a priori profile and an altitude-independent smoothing constraint was applied to avoid artificial structures and instabilities in the results, respectively.

Figure 1.

Spectral identification of formic acid at 11 km from one MIPAS measurement (21 October 2003, 7:22Z) over Africa. The bottom traces correspond to residual spectra (measurement − calculation) when HCOOH is included (black) and omitted (red) in the radiative transfer calculation. The difference between these two is given in blue and resembles the simulated contribution of formic acid alone (gray) in the atmosphere at this altitude.

[11] The individual measurement presented in Figure 1 is from 21 October 2003 (FR measurement mode) viewing at 11 km altitude over Africa. The radiative transfer calculations were performed with and without the HCOOH line parameters; the residuals of the best fits (measurement-simulation) are shown as black and red traces, respectively. The blue trace represents the difference of these and corresponds to the net emission spectrum of formic acid, which is shown separately in the gray trace and serves as proof for its spectral identification from a single measurement.

[12] In Figure 2, a profile of HCOOH (black trace) retrieved from a single limb sequence also over Africa but now on 25 September 2006 (RR measurement mode) is presented. This profile corresponds to an extreme biomass burning event when the VMR reached ∼1.1 ppbv at 10 km altitude. The averaging kernels are plotted as inset within Figure 2 and shows the sensitivity to the true HCOOH profile from the upper troposphere up to the stratosphere. The degrees of freedom, given by the trace of the averaging kernels, were in average 6.4 ± 1.0 for FR- and 6.8 ± 1.3 for the RR-measurement mode. Also in Figure 2 monthly mean profiles for July and December of 2006 from middle latitudes (20°–60°N) are plotted to show the significant seasonal dependence, which will be discussed in the next section.

Figure 2.

Vertical profile (black trace) retrieved from a single MIPAS scan during an extreme event over Africa on the 25 September 2006. The corresponding rows of averaging kernels for that measurement are displayed in the inset. The green and blue traces are the mean profiles calculated from all processed measurements in midlatitudes (20°–60°N) during the months of July and December 2006, respectively.

[13] Global processing of more than 344,000 limb sequences was performed that corresponds to over 4,300 orbits or 337 complete days between September of 2002 and July of 2008. Zonal means were calculated for every month and their typical standard deviations ranged between 50% and 90% in the altitude range of interest. The total estimated precision of the HCOOH retrieval in the relevant 8–12 km altitude range of one geolocation, measured over Africa on 21 October 2003 during a biomass burning event, is about 12% as seen in Figure 3 (solid line). About 4%–5% of the error in this altitude range is due to noise (dotted line) and the rest are the contributions of the major random parameter errors such as the instrumental line-of-sight (LOS) with 10%–11%, temperature (temp) with 2%–3%, gain calibration of the measured spectra (gain) with 2%–3%, and the horizontal temperature gradients (grad T) with about 1%. In the case of single measurements where no enhancement of HCOOH was detected, the total retrieval error is dominated by the noise, which can be as high as 30% in this altitude range. The noise error, which also becomes important above 14 km in cases like what is shown in Figure 3, is reduced considerably in the results presented here since substantial averaging was performed. The major systematic error, not shown in Figure 3 arises from the spectroscopic information available. For example, the accuracy of the absolute line intensities of HCOOH reported by Vander Auwera et al. [2007] is about 7%.

Figure 3.

Total retrieval estimated error (solid line) of a single profile measured over Africa on 21 October 2003, when the VMR at 11 km altitude reached 364 pptv. The noise error and the contribution of the major random parameter errors are given as dotted and dashed lines, respectively. LOS = instrumental line-of-sight; temp = temperature; gain = gain calibration of the measured spectra; grad T = temperature gradient; shift = spectral shift; ILS = instrumental line shape.

3. Results and Discussion

3.1. Time Series

[14] Global MIPAS data processed between September 2002 and July 2008 for deriving formic acid mixing ratios were plotted in Figure 4 as monthly mean VMR values for different regions. When only midlatitudes in the Northern Hemisphere are considered, as can be seen in Figure 4 (bottom), a strong seasonal pattern can be observed in the 8–18 km range. The vertical dependence shows a steady decrease of the VMR with height, as can also be seen in the blue and green profiles shown in Figure 2. These values match extraordinarily well the mean values reported for the periods between February 2004–December 2005 and December 2004–July 2007 from solar occultation measurements by the ACE-FTS instrument [González Abad et al., 2009]. MIPAS's wide spatial and temporal coverage allows to study the variability in detail. The largest values (typically between 100–110 pptv at 8 km) are observed between April and August, followed by a rapid decline in September and minima detected during the winter months (November through January). The amplitude of this oscillation goes from about 55 pptv at 8 km down to below 10 pptv above 16 km.

Figure 4.

Monthly mean volume mixing ratios of HCOOH measured by MIPAS between September 2002 and July 2008. (bottom) Northern middle latitude VMRs at different altitudes. (top) Mean VMRs at 10 km in different regions of the world are presented (N pole = 70°–90°N, SAF = South Africa, SA = South America, IND+AUS = Indonesia and Australia, S pole = 70°–90°S).

[15] Also in Figure 4 (top), the evolution of monthly mean VMR's at 10 km height are investigated in specific regions of the world. The light blue trace corresponds to the north-polar region (70°–90°N) and its evolution resembles that observed in the midlatitudes, with the difference that the increase takes place later in the spring. The VMRs detected by MIPAS in the Northern Hemisphere, as gathered from the regular seasonal dependence, are thus mainly shaped by biogenic emissions from vegetation and soils as suggested by Talbot et al. [1988] and Andreae et al. [1988].

[16] Three regions from the Southern Hemisphere were chosen and investigated separately. These regions were used by Li et al. [2009] to describe the emissions of HCN produced from biomass burning, and southern Africa (SAF), South America (SA), and Indonesia and Australia (IND+AUS) were found to have the largest contributions. The corresponding geographical areas are defined by 48°–1°S and 5°–60°E, 57°S–16°N and 95°–32°W, and 50°S–7°N and 88°–165°E, respectively. The evolution of the monthly mean values over these regions are plotted as red, pink, and green traces in Figure 4. A strong HCOOH signal at 10 km can be observed peaking during the months of September and October but with very distinct strengths in the different years. Among the biomass burning seasons measured by MIPAS during the years 2002, 2003, 2006, and 2007, the strongest HCOOH signals over southern Africa was observed during September of 2006. The mean VMR in this region reached 200 pptv at 10 km altitude, which is more than twice what is regularly measured during the summer at midlatitudes in the Northern Hemisphere.

[17] The strong signal of HCCOH detected over SAF during the biomass burning season is likely to be related to the moderate El Niño event during that year ( The El Niño-Southern Oscillation (ENSO) is known to produce interannual variability in the tropical atmosphere, resulting in important changes in large-scale circulation and convection. The moderate El Niño of 2006 was observed to have perturbations on biomass burning gases detected by the Tropospheric Emission Spectrometer (TES) when compared to the 2005 season, which was considered a neutral year [Logan et al., 2008]. Although we are unable to compare with that particular year because of insufficient processed data during those months, a clear enhancement in the SAF region is observed with respect to 2003 and 2007, which are considered neutral and weak La Niña events, respectively.

[18] The brown trace in Figure 4, which corresponds to the polar region in the Southern Hemisphere, presents a background mixing ratio of 10–20 pptv at 10 km altitude. No evidence of biogenic emissions are discernible during the Antarctic spring but rather there is a distinct influence from biomass burning from mid- and tropical latitudes when compared to the signals from South America (pink), southern Africa (red), and Indonesia and Australia (green).

3.2. Global Distribution

[19] Three time periods have been selected for data presentation based on the time series shown in the previous section. The global distributions of formic acid at 10 km for the periods March–July, August–October, and November–February are presented in Figure 5. Three time sequences are presented in order to compare side by side the distinct seasonal features that are repeated in different years.

Figure 5.

Global HCOOH maps (pptv) at 10 km altitude for the periods March–July, August–October, and November–February. Three different year sequences are presented.

[20] The March–July period is characterized by a homogenous distribution of formic acid over northern midlatitudes. In the tropics, enhancements with similar magnitudes are located over continental regions and the Atlantic. The white areas in the maps correspond to data that have been filtered out due to cloud contamination.

[21] During the period between August and October, the HCOOH mixing ratios over the middle latitudes have clearly diminished in the Northern Hemisphere but strong enrichments are detected particularly over the SAF, SA, and IND+AUS regions. The geographical delimitations of the areas discussed in the previous section, which lie within the Southern Hemispheric pollution belt, are marked in Figure 5 (top). Enhancements and similar distributions of CO [Funke et al., 2009], C2H6 [von Clarmann et al., 2007], and peroxyacetyl nitrate [Glatthor et al., 2007] have also been detected by MIPAS in this region during a biomass burning episode in October 2003. The transatlantic transport phenomenon as well as the trajectories of these biomass burning plumes have been analyzed in detail [von Clarmann et al., 2007; Funke et al., 2009]. It is evident from Figure 5 that the strongest mean VMR values of HCOOH are observed over southern Africa during the August–October period in 2006, in accordance to what has been presented in Figure 4.

[22] The lowest VMR values are detected during the period from November until February (Figure 5, bottom). Here the middle latitude mixing ratios in the Northern Hemisphere have reached their background levels (30–40 pptv at 10 km) but higher VMR persist in tropical latitudes and particularly along the Southern Hemispheric pollution belt.

3.3. Day-Night Differences

[23] The day- and nighttime mean profiles and global maps of HCOOH were separated in order to investigate possible atmospheric sources complementing the direct emissions from both biogenic and biomass burning sources. No significant difference was detected in most of the cases, confirming that the upper tropospheric lifetimes are longer than 1 day. The monotonical decrease of the VMR's usually observed in the retrieved profiles supports the idea that the HCOOH emitted near the surface can be transported to the upper troposphere and is then removed by deposition as recognized in previous studies [Jacob and Wofsy, 1988; Keene and Galloway, 1988; Hartmann et al., 1991].

[24] In Figure 6, the mean profiles at northern middle latitudes in the growing season (July 2006) are plotted and separated into measurements performed during the day (red) and night (blue). No significant difference can be established, as was for most cases investigated. However, the extreme event detected during August–October of the same year in southern Africa (SAF) clearly shows a day-night difference that was 25%–30% from the mean in the 8–12 km altitude range. An average of 242 data points were used within this altitude range, from which 43% correspond to day and 57% to night measurements. It is worth mentioning that the apparent maxima at 9 km altitude in the mean HCOOH profiles are not present when only profiles for which data are available all the way down to 8 km are plotted. The significant day-night difference observed above 9 km, however, suggests that substantial photochemical production of HCOOH is taking place during this event.

Figure 6.

Day and night mean profiles for northern middle latitudes during July 2006 and for the southern African region (SAF) during August until October 2006.

[25] It can be assumed that the in situ formation through mechanisms such as those suggested from olefin ozonolysis, isoprene oxidation, and gas phase reaction of HCHO with HO2 [Jacob and Wofsy, 1988; Talbot et al., 1990; Madronich et al., 1990; Sanhueza et al., 1996] can take place and could be contributing under very specific conditions to the upper tropospheric levels of HCOOH. The savannah region of southern Africa undergoes a flaming-type of combustion, which is characterized to favor the emission of simple compounds during the pyrolisis process. On the other hand, during the lower-temperature smoldering process, which is known to dominate in the equatorial America, large amounts of incompletely oxydized products are emitted as described by Andreae and Merlet [2001]. von Clarmann et al. [2007], for example, have observed that plumes originated from the southern African fires contain considerably larger O3/C2H6 ratios compared to those from South America. This and other peculiarities in the chemical composition of the plumes, which in turn could also be influenced by their distinct resident times, could explain why no significant day-night difference could be observed in the SA and IND+AUS regions. For these night-day differences to become noticeable from the SAF savannah fires of 2006, moreover, not only the in situ production but also the removal processes need to be efficiently taking place. This example provides enough evidence that formic acid can be produced photochemically and contributing to the upper tropospheric abundances under specific plume conditions. However, there is no indication as to what the relative contribution of this production mechanism might be when compared to its direct emission from the surface.

4. Summary and Conclusion

[26] Formic acid in the upper troposphere has been retrieved from MIPAS-ENVISAT infrared radiances measured between September 2002 and July 2008 with unprecedented spatial and temporal coverage. Global distributions and time series of monthly mean HCOOH show distinct seasonal and regional features. The monthly mean profiles show a fast decline of the mixing ratios from 8 to 16 km altitudes and a strong seasonal cycle in the Northern Hemisphere. This pattern, which is strongly associated to plant growth and corresponding biogenic emissions, reaches 100–110 pptv in middle latitudes at 8 km altitude around July and drops considerably during the winter months. This seasonal variation is in good agreement with that reported from a fixed ground-based measurement site using solar absorption spectroscopy [Rinsland et al., 2004], although a different set of spectroscopic line parameters were employed in that study. Also, the VMR values observed from MIPAS coincide with in situ measurements from balloons in the free troposphere [Goldman et al., 1984; Remedios et al., 2007] and with aircraft measurements [Talbot et al., 1992; Reiner et al., 1999].

[27] In the Southern Hemisphere, there is a large contribution from biomass burning affecting the observed abundances in the upper troposphere. Monthly mean VMRs over specific regions, including southern Africa, Indonesia, Australia, and South America, can be increased by a factor of 2 or more from their mean background levels in the upper troposphere due to the impact of biomass burning emissions. During September 2006, for example, a mean monthly value of 200 pptv over southern Africa was registered and a single measurement on 25 September resulted in VMRs larger than 1 ppbv at 10 km and lower altitudes. It is shown here that atmospheric or secondary production of HCOOH can significantly contribute to the observed VMRs under specific conditions also at these altitudes. In-plume production of HCOOH has been observed from aircraft measurements and described in detail by Lefer et al. [1994]. The day-night differences observed during the 2006 savannah fires imply that both photochemical and removal processes are efficiently occurring. No such differentiation was detected in the northern middle latitudes during growing season, nor from other regions with large biomass burning emissions, suggesting that the type of combustion, the plume's residence times, and/or the uplifting processes might be decisive in favoring the secondary production of HCOOH.

[28] Further investigations are required in order to explore the nature and the abundances of organic acids in the atmosphere. For example, acetic acid has been found to be present in the boundary layer quantities comparable to HCOOH. Their relationship could give more information about their origin since from biomass burning and other combustion processes, acetic acid is emitted in larger proportions [Talbot et al., 1988; Hartmann et al., 1991], whereas biogenic emissions contain in proportion larger amounts of HCOOH.


[29] The Alexander von Humoldt Foundation is thanked by the first author for supporting a research stay of 1 year at the Institute of Meteorology and Climate Research in Karlsruhe. ESA is acknowledged for providing MIPAS level-1B data.