Acetylene (C2H2) and hydrogen cyanide (HCN) from IASI satellite observations: global distribution, validation, and comparison with model.

We present global distributions of C 2 H 2 and HCN total columns derived from the Infrared Atmospheric Sounding Interferometer (IASI). These distributions are obtained with a fast method allowing to retrieve C 2 H 2 abundance globally with a 5 % precision and HCN abundance in the tropical (subtropical) belt with a 10 % (30 %) precision. 5 IASI data are compared for validation purposes with ground-based Fourier Transform Infrared (FTIR) spectrometer measurements at four selected stations. We show that there is an overall agreement between the ground-based and space measurements. Global C 2 H 2 and subtropical HCN abundances retrieved from IASI spectra show the expected seasonality linked to variations in the anthropogenic emissions and seasonal 10 biomass burning activity, as well as exceptional events, and are in good agreement with previous spaceborne studies. IASI measurements are also compared to the distributions from the Model for Ozone and Related Chemical Tracers, version 4 (MOZART-4). Seasonal cycles observed from satellite data are reasonably well reproduced by the model. However, the model seems to overestimate (underestimate) anthropogenic 15 (biomass burning) emissions and a negative global mean bias of 1 % (16 %) of the model relative to the satellite observations was found for C 2 H 2 (HCN).

shown that HCN and C 2 H 2 columns can be routinely retrieved from IASI spectra, even in absence of exceptional columns or uplift mechanisms, when CO 2 line mixing is accounted for in the inversion scheme. These previous works were based on an optimal estimation method (OEM) developed and formalized by Rodgers (2000).
In this paper, we first present a fast scheme for the global detection and quantification 5 of HCN and C 2 H 2 total columns from IASI spectra. We describe 2008-2010 time series and analyze the seasonality of the columns of these two species above four NDACC sites in comparison with ground-based FTIR measurements. We finally present the global distributions for the years 2008 to 2010 that we compare with model outputs for these two species. twice a day with a footprint of 12 km at nadir. IASI is a Fourier transform spectrometer that measures the thermal infrared radiation emitted by the Earth's surface and atmosphere in the 645-2760 cm −1 spectral range with a spectral resolution of 0.5 cm −1 apodized and a radiometric noise below 0.2 K between 645 and 950 cm −1 at 280 K . The IASI spectra used in this study are calibrated radiance 20 spectra provided by EUMETCast near-real-time service.

Retrieval strategy
Up to now, 24 trace gases have been detected from IASI radiance spectra, including HCN and C 2 H 2 (see Clarisse et al., 2011a, for the list of detected species), with an OEM (Rodgers, 2000) implemented in a line by line radiative transfer model called Introduction Atmosphit (Coheur et al., 2005). In the cases of HCN and C 2 H 2 , the accuracy of the retrievals has been recently improved by taking into consideration the CO 2 line mixing in the radiative transfer model . This retrieval method, relying on spectral fitting, needs a high computational power and is time consuming, especially when a large number of spectra has to be analyzed and fitted. This is therefore not 5 suitable for providing global scale concentrations distributions of these trace gases in a reasonable time.
One of the commonly used methods for the fast detection of trace gases is the brightness temperature difference (BTD) between a small number of channels, some being sensitive to the target species, some being not. Such a method has been used 10 from IASI spectra for sulfur dioxide (SO 2 ) (Clarisse et al., 2008) and ammonia (NH 3 ) . It is of particular interest in operational applications (quick alerts) or when large amounts of data need to be processed. However, relying on a cautious selection of channels to avoid the contamination with other trace gases, the BTD method does not fully exploit all the information contained in hyperspectral measure- 15 ments. Especially, low concentrations of the target species may not be detected with such a method. Walker et al. (2011) presented a fast and reliable method for the detection of atmospheric trace gases that fully exploits the spectral range and spectral resolution of hyperspectral instruments in a single retrieval step. They used it to retrieve SO 2 total 20 column from a volcanic plume and NH 3 total column above India. More recently, Van Damme et al. (2014) presented a retrieval scheme to retrieve NH 3 from IASI spectra based on the work of Walker et al. (2011), and introduced a metric called Hyperspectral Range Index (HRI). We use in the present study a similar approach. 25 The method used in this study is a non-iterative pseudo retrieval method of a single physical variable or target species x expressed as, following the formalism developed Introduction  Rodgers (2000):

Hyperspectral Range Index (HRI)
where y is the spectral measurements, x 0 is the linearization point, F is the forward model (FM), S tot is the covariance of the total error (random + systematic), and the Jacobian K is the derivative of the FM to the target species in a fixed atmosphere. 5 S tot can be estimated considering an appropriate ensemble of N measured spectra which can be used to build up the total measurement error covariance S obs y : where y is the calculated mean spectrum for the ensemble.
To generate S obs y , we randomly chose 1 million spectra observed by IASI all over 10 the world, above both land and sea, during the year 2009. Then, we applied a BTD test to remove the spectra contaminated by the target species. For HCN (C 2 H 2 ), the wavenumbers 716.5 and 732 cm −1 (712.25 and 737.75 cm −1 ) were used as reference channels and 712.5 cm −1 (730 cm −1 ) was used as test channel (Fig. 1, middle panel). Given the medium lifetimes of the target species (few weeks for C 2 H 2 to few months for 15 HCN), and the limited accuracy of the BTD test due to the weak spectral signatures of the target species, it is likely that such randomly chosen and filtered spectra still contain a small amount of the target species whose signal may come out from the noise. This limitation decreases the sensitivity of the method, which is discussed in Sect. 2.2.3. The spectral ranges considered to compute the S obs y matrices are 645-800 cm −1 for 20 HCN and 645-845 cm −1 for C 2 H 2 ( Fig. 1, top panel). These ranges were chosen as they include parts of the spectrum which have a relatively strong signal from the target species but also from the main interfering species (CO 2 , H 2 O and O 3 , Fig. 1 and y, the HRI of a measured spectrum y can be defined as: with G the measurement contribution function The HRI is a dimensionless scalar similar, other than units, to the apparent column 5 retrieved in Walker et al. (2011). Unlike the optimal estimation method, no information about the vertical sensitivity can be extracted. Note also that the use of a fixed Jacobian to calculate HRI does not allow generating meaningful averaging kernels.

Conversion of HRI into total columns
Having calculated the matrices G for HCN and C 2 H 2 , each observed spectrum can be 10 associated through Eq.
(3) with a value of HRI for HCN (HRI HCN ) and C 2 H 2 (HRI C 2 H 2 ). These HRIs are only metrics for determining whether levels of the gas are enhanced with respect to the climatological background over the vertical levels where the instrument is sensitive. For a given atmosphere atm, the main challenge is then to link the HRI to a column amount of the target molecule, i.e. to find B HCN atm and B C 2 H 2atm such 15 as: [X ] being the species abundance in molec cm −2 .
To determine these coefficients linking the HRIs to total column amounts, HCN and C 2 H 2 profiles have been constructed, with enhanced concentrations of the species 20 located in a 1 km thick layer, whose altitude is varied from the ground up to 30 km for HCN and up to 20 km for C 2 H 2 (the choice of these maximum altitudes are made with respect to the Jacobians of the FM that are shown in Fig. 3  . Each of the constructed profile has been associated with a spectrum through the FM of Atmosphit considering standard absorption profiles. The associated values of HRI HCN and HRI C 2 H 2 have then been computed for each of the simulated spectra. Figure 2 shows the look up tables (LUTs) of HRI HCN (top) and HRI C 2 H 2 (bottom) as a function of the abundance of the target molecule and of the altitude of the pol-5 luted layer in a standard tropical modeled atmosphere (Anderson et al., 1986). Similar LUTs have been computed for standard temperate (US standard atmosphere) and polar (Anderson et al., 1986) atmospheres (data not shown). The satellite viewing angles were taken into account in the HRI calculation similarly to Van Damme et al. (2014).
One can see that, for a given atmosphere and for a given altitude of the polluted layer, 10 the abundances of both species linearly depend on the HRI value, which validates Eq. (5). For a given atmosphere atm and a given species X , the different values of B with respect to the altitude z of the polluted layer will be noted b X atm (z) and b X atm (z) in the following. Figure 3 shows the normalised Jacobians of the FM for HCN and C 2 H 2 aver-15 aged over the spectral ranges given in Sect. 2.2.1 (645-800 cm −1 for HCN and 645-845 cm −1 for C 2 H 2 ) and for each of the three standard modeled atmospheres. These Jacobians express the sensitivity of the FM, i.e. both the radiative transfer model and IASI (through its instrumental function), to the target species abundance X in a fixed atmosphere atm: We then obtain the coefficients B X atm by multiplying the b X atm (z) by the value of the Jacobian at the altitude z:  Figure 4 gives the resulting values of B HCN (blue) and B C 2 H 2 (green) in function of the latitude.

Sensitivity and stability of the method
The sensitivity of the method can be assessed from the Jacobians presented in Fig. 3. For HCN, one can see that there is no sensitivity at the surface and above ∼ 30 km, and the altitude of the sensitivity peak is located close to the tropopause at ∼ 9, ∼ 11 and ∼ 14 km for the polar, temperate and tropical atmospheres, respectively. For C 2 H 2 , 10 there is no sensitivity above ∼ 20 km, and the maximum sensitivity is reached at ∼ 8, ∼ 10 and ∼ 11 km for the polar, temperate and tropical atmospheres, respectively. The HRIs presented here above are sensitive to the abundance of the target species -this is what they are made for -and to their vertical distribution. However, the measured column amount may also depend on: (1) the proper suppression of the 15 spectral background, (2) the conditions of thermal contrast with the surface (TC), and (3) the accuracy of the FM to simulate the spectra used to build up the LUTs. The latter was discussed already by Duflot et al. (2013). In order to test the impact of the two first factors (spectral background suppression and TC) on the retrieved column amount, HCN and C 2 H 2 profiles have been constructed with varying TC and concentrations of 20 the interfering and target species. The TC is defined here as the difference between the skin (surface) temperature and that of the air at an altitude of 1.5 km. These variations in interfering species abundances and TC were considered to be independent and were taken within the range ±2 % for CO 2 and ±20 % for H 2 O and O 3 , and in the range ±10 K for the TC. For a fixed column amount of the target species, the HRIs 25 were compared one by one to a HRI corresponding to a standard spectrum (i.e. with background concentrations of the interfering species and a TC equal to zero) and if the Introduction difference between the two HRIs was lower than 10 %, then this fixed abundance of the target species was tagged as detectable independently from the listed parameters. The TC was found to be the major source of HRI variation for both target species, and a serious cause of limitation only for HCN. Figure 5 shows the variation of HRI HCN caused by a TC equal to ±10 K. One can see that HCN column amount can be 5 detected with a variation due to the TC below 10 % when its abundance is higher than 0.28, 1.2 and 1.6 × 10 16 molec cm −2 for the tropical, temperate and polar atmospheres, respectively. This gives the stability thresholds above which HCN column amount can be measured with a 10 % confidence in the independence of the retrieval method to the atmospheric parameters. Consequently, as the stability thresh-10 olds of the method for HCN in temperate and polar atmospheres are too high (1.2 and 1.6 × 10 16 molec cm −2 , respectively) to allow the detection of HCN background abundances as compared to usual background column of typically 0.35 × 10 16 molec cm −2 (Vigouroux et al., 2012;Duflot et al., 2013), IASI HCN measurements have to be rejected in these two types of atmosphere, and considered in the tropical belt for val-15 ues above 0.28 × 10 16 molec cm −2 . In order to broaden the exploitable latitude range, we take into account the IASI HCN measurements at subtropical latitudes with the same stability threshold (0.28 × 10 16 molec cm −2 ), assuming a 30 % confidence in the independence of the retrieval method to the atmospheric parameters -which is quite a prudent assumption. As a result, in the following, IASI HCN measurements is con-20 sidered in the ±35 • latitude band with a stability threshold of 0.28 × 10 16 molec cm −2 , and confidence in the stability of the method is 10 % at tropical latitudes ([±20 • ]) and 30 % at subtropical latitudes ([±35 • : ±20 • ]). Oppositely to HCN, for C 2 H 2 , the variation of HRI C 2 H 2 due to varying TC was found to be lower than 5 % for every C 2 H 2 abundances (data not shown). Consequently, in the following no IASI C 2 H 2 measurements ACPD 15,2015

Results
The goal of this section is to describe and evaluate the C 2 H 2 and HCN total columns as measured by IASI. We first compare HCN and C 2 H 2 total columns retrieved from IASI spectra and from ground-based FTIR spectra. We then depict the C 2 H 2 and HCN total columns at global and regional scales. IASI global and regional distributions are 5 finally compared with output from the Model for Ozone and Related Chemical Tracers, version 4 (MOZART-4) in order to evaluate the agreement between the model and the IASI distributions.

Comparison with ground-based observations
We compare in this section HCN and C 2 H 2 total columns retrieved from IASI spectra  6). IASI cloudy spectra were removed from the data set using a 10 % contamination threshold on the cloud fraction in the pixel. As  Total errors for ground-based measurements at Reunion Island are 17 % for both species, total error for HCN ground-based measurements at Wollongong is 15 %, total error for HCN ground-based measurements at Izaña is 10 %, and total error for C 2 H 2 ground-based measurements at Jungfraujoch is 7 %. Detailed description of groundbased FTIR data set, retrieval method and error budget can be found in Vigouroux   15,2015 Figure 7 shows the mean total column averaging kernels for the ground-based FTIR at each of the four sites. Similarly to IASI (Fig. 3), information content from groundbased instruments measurements is mostly in the middle-high troposphere for both species. The main difference can be observed for tropical C 2 H 2 : while IASI Jacobian peaks at 10 km for C 2 H 2 in a tropical atmosphere, ground-based FTIR averaging kernel 10 peaks at 15 km for C 2 H 2 at Reunion Island. Figure 8 shows the comparison between the IASI and the ground-based measurements. IASI retrieved total columns were averaged on a daily basis and on a 1 • ×1 • area around the observation sites. HCN retrieved abundances below 2.8 × 10 15 molec cm −2 have been removed from both ground-based and space measurements to allow com- 15 parison of both datasets (cf. Sect. 2.2.3). One can see that there is an overall agreement between the IASI and the ground-based FTIR measurements considering the error bars. An important result from this study is that IASI seems to capture the seasonality in the two species in most of the cases. This is best seen by looking at the IASI monthly mean retrieved total columns (black circles and lines in Fig. 8). The scatter of 20 the IASI daily mean measurements (red dots) are due to the averaging on a 1 • × 1 • area around the observation sites. At Reunion Island HCN and C 2 H 2 peak in October-November and are related to the Southern Hemisphere biomass burning season (Vigouroux et al., 2012). We find maxima of around 12 × 10 15 molec cm −2 for HCN and 10 × 10 15 molec cm −2 for C 2 H 2 . 25 The seasonality and interannual variability matches very well that of the ground-based FTIR measurements for HCN (correlation coefficient of 0.81 for the entire daily mean dataset, and of 0.98 for the monthly mean data set) but with the IASI columns being biased high by 0.79 × 10 15 molec cm −2 (17 %). For C 2 H 2 at Reunion Island, the sea-ACPD 15,2015  sonality and interannual variability matches reasonably well that of the ground-based measurements (correlation coefficient of 0.40 for the entire daily mean dataset, and of 0.72 for the monthly mean data set) but with the IASI columns being biased high by 1.10 × 10 15 molec cm −2 (107 %). Such a high bias between the two datasets could be due to the difference between space and ground-based instruments sensitivity (Figs. 3 5 and 7). One can also notice that the C 2 H 2 and HCN peaks are higher in 2010. As South American biomass burning plumes are known to impact trace gases abundance above Reunion Island (Edwards et al., 2006a, b;Duflot et al., 2010), these 2010 higher peaks are probably due to the 2010 great Amazonian fires (Lewis et al., 2011) influence. At Wollongong HCN peaks also in October-November due to the Southern Hemi-10 sphere biomass burning season (Paton-Walsh et al., 2010). We find maxima of around 11 × 10 15 molec cm −2 in October 2010, which is, similarly to Reunion Island, very likely to be a signature of the great Amazonian fires as South American biomass burning plumes are known to impact trace gases abundance above Australia (Edwards et al., 2006a, b). The seasonality and interannual variability matches well that of the ground- 15 based FTIR measurements (correlation coefficient of 0.55 for the entire daily mean dataset, and of 0.83 for the monthly mean data set), with the IASI columns being biased low by 0.48 × 10 15 molec cm −2 (10 %). At Izaña HCN peaks in May-July due to the biomass burning activity occurring in Northern America and Europe (Sancho et al., 1992). We find maxima of around 8 × 20 10 15 molec cm −2 . The seasonality and interannual variability matches poorly that of the ground-based FTIR measurements for HCN (correlation coefficient of 0.28 for the entire daily mean dataset, and of 0.64 for the monthly mean data set), with the IASI columns being biased high by 0.45 × 10 15 molec cm −2 (11 %). One can notice that HCN total columns as measured by ground-based FTIR are below the HCN stability threshold in 25 boreal winter, which may result in erroneous IASI measurements (because unstable) and explain this poor match between the two datasets. For C 2 H 2 at the Jungfraujoch site, the agreement between IASI and the groundbased retrieved columns is good (correlation coefficient of 0.70 for the entire daily mean ACPD 15,2015  dataset, and of 0.85 for the monthly mean data set), with the IASI columns being biased low by 0.15 × 10 15 molec cm −2 (12 %), opposite to the observations at Reunion. The larger columns observed in late winter are caused by the increased C 2 H 2 lifetime in that season (caused by the seasonal change in OH abundance) (Zander et al., 1991), and we find corresponding maxima of up to 4 × 10 15 molec cm −2 . 5

IASI Global distributions
We focus in this section on the description of the C 2 H 2 and HCN distributions retrieved from IASI spectra. For practical reasons, the figures used in this section also show simulated distributions that will be analyzed afterwards.
The left panels of Figs. 9 and 10 provide the seasonal global and subtropical distribu-10 tions of C 2 H 2 and HCN total columns, respectively, as measured by IASI and averaged over the years 2008 to 2010. Looking at IASI measurements (Figs. 9 and 10 -left panels), one can notice the following main persisting features for both C 2 H 2 and HCN: the hot spots mainly due to the biomass burning activity occurring in Africa and 15 moving southward along the year (Sauvage et al., 2005;van der Werf et al., 2006); the hot spot located in South East Asia being likely a combination of biomass burning and anthropogenic activities; the transatlantic transport pathway linking the African west coast to the South American east coast and moving southward along the year (Edwards et al., 2003(Edwards et al., , 20 2006aGlatthor et al., 2015).
The following seasonal features can also be observed: the transpacific transport pathway linking Eastern Asia to Western North America, especially in March-April-May (MAM) (Yienger et al., 2000); ACPD 15,2015  the transport pathway from Southern Africa to Australia in June-July-August (JJA) and September-October-November (SON) (Annegarn et al., 2002;Edwards et al., 2006a, b); the transport pathway linking South America (especially Amazonia) to Southern Africa and Australia during the SON period (Edwards et al., 2006a deep convection, and confinement by the strong anticyclonic circulation (Randel et al., 2010). The enhanced abundance of C 2 H 2 and HCN within the AMA in JJA observed by IASI is in accordance with previous studies (Park et al., 2008;Randel et al., 2010;Parker et al., 2011;Glatthor et al., 2015); however, one should keep in mind that this enhanced abundance measured by IASI is likely due to the 20 combination of this pollution uplift and confinement with the higher sensitivity of the method in the upper troposphere (Fig. 2).
One can also notice the very good agreement between the seasonal HCN distributions shown in our Fig. 10  In Northern America, Europe and Boreal Central Asia (Fig. 11 -Zones NAM, EUR and BCA), C 2 H 2 peaks in late boreal winter due to the increased C 2 H 2 lifetime as already noticed over Jungfraujoch (Fig. 8) In North Central America (Fig. 12 -Zone NCA), the annual HCN peak in April-June is driven by local fire activity (van der Werf et al., 2010).
In South America, Southern Africa and Australia (Figs. 11 and 12 -Zones SAM, SAF and AUS), the Southern Hemisphere biomass burning season clearly drives the C 2 H 2 10 and HCN peaks in September-November each year. The signature of the great 2010 Amazonian fires (Lewis et al., 2011) is visible on each of the these three Zones, South American fire plumes being known to impact Southern Africa and Australia (Edwards et al., 2003(Edwards et al., , 2006a. The February 2009 Australian bush fires (Glatthor et al., 2013) are also noticeable on Zone AUS for both species. 15 In Northern Africa (Figs. 11 and 12 -Zone NAF), C 2 H 2 and HCN peak in boreal winter because of the biomass burning activity occurring in the Zone, and peak also in boreal summer because of the European and South Mediterranean fires ( Van der Werf et al., 2010).
In South East Asia (Figs. 11 and 12 -Zone SEA), the observed C 2 H 2 and HCN 20 peaks in July-September and January-March are due to local fire activity (Fortems-Cheiney et al., 2011;Magi et al., 2012). Additionally, the July-September peaks are also likely due to the combination of the pollution uplift and confinement within the AMA with the higher sensitivity of the method in the upper troposphere. In Equatorial Asia (Figs. 11 and 12 -Zone EQA), local fire activity is visible in C 2 H 2 and HCN sharing important common sources (cf. Introduction), the same annual and seasonal features are observed for both species. However, biomass burning being the major source for HCN (while it is biofuel and fossil fuel combustions for C 2 H 2 ), one can notice the especially high increase in HCN abundance (up to 13 × 10 15 molec cm −2 ) in the Southern Hemisphere during the austral biomass burning 5 season (September to November). These observations are in accordance with previous studies (Lupu et al., 2009;Glatthor et al., 2009;Wiegele et al., 2012).

Comparison with model
In order to further evaluate the HCN and C 2 H 2 distributions retrieved from IASI spectra, they are compared in this section to the output of MOZART-4 for the years 2008-10 2010. We first describe the simulation set up before comparing simulated and observed distributions.

MOZART-4 simulation set up
The model simulations presented here are performed with the MOZART-4 global 3-D chemical transport model (Emmons et al., 2010a), which is driven by assimilated mete- Model emissions for HCN and C 2 H 2 used in this study are summarized in Table 1 and presented in Fig. 13. The majority of the emissions for both C 2 H 2 and HCN are 15 from anthropogenic source (about 80 and 55 % of the global source of C 2 H 2 and HCN, respectively; see Table 1). Averaged over the period 2008-2010, the highest HCN and C 2 H 2 anthropogenic surface emissions are observed over China, with elevated emissions over India, Europe and USA, due to intense industrialization, where values larger than 4 × 10 −12 kg(C 2 H 2 ) m −2 s −1 are entered in the model. The most intense HCN and 20 C 2 H 2 emissions due to biomass burning are observed over South East Asia, equatorial and southern Africa, South America, Siberia and Canada.

IASI vs. model global distributions
Figures 9 and 10 provide the seasonal global and subtropical distributions of C 2 H 2 and HCN total columns, respectively, as measured by IASI and as simulated by MOZART- information, we rather applied on each of the MOZART-4 simulated profiles the Jacobians of the used forward model (cf. Sect. 2.2.3 and Fig. 3) to take into account the sensitivity of both the radiative transfer model and IASI. Note that here again HCN abundances below 2.8 × 10 15 molec cm −2 have been removed from both space measurements and simulated columns to allow comparison of both datasets (cf. Sect. 2.2.3).

5
MOZART-4 simulations can be evaluated by looking at Figs. 9 and 11 for C 2 H 2 , and Figs. 10 and 12 for HCN. Figures 11 and 12 show the simulated C 2 H 2 and HCN total columns time series, respectively, for each of the zones defined in Fig. 6 superimposed to IASI observations. Table 2 summarizes the biases and correlation coefficients resulting from the comparison between model and observations. Looking at these Table   10 and Figures In Table 2, for C 2 H 2 , the correlation coefficients are good (≥ 0.6) to very good (≥ 0.9) except for the zones SAM (South America), SEA (South East Asia) and EQA (Equatorial Asia). For HCN, the correlation coefficients are good (≥ 0.6) except for the zones NCA (North Central America), NAF (Northern Africa), SEA and EQA. For South America (Zone SAM), correlation coefficient is not as good for C 2 H 2 (R = 5 0.54) due to a backward shift of the species abundance peaks in years 2008 and 2009: in the model, this increase occurs from July to October while observations (and previous studies, e.g. van der Werf et al., 2010) show an increase from August to December. This backward shift is also visible for HCN (Fig. 12), but to a lesser extent. For South East Asia and Equatorial Asia (Zones SEA and EQA), the low correlation 10 coefficients (cf. Table 2) can be attributed to the difficulty of locating precisely with the model the intercontinental convergence zone (ITCZ) which drives the long-range transport of C 2 H 2 and HCN-loaded plumes into the zone. Additionally, for Equatorial Asia, the too low fire emissions considered in the model for Indonesia from July to December 2009 may also be a cause for these low correlation coefficients. 15 For HCN in northern Africa (Zone NAF), correlation coefficient is very low (R = 0.07) because the model sets the abundance peaks around August while observations show peaks occurring around December, which is in accordance with previous studies (van der Werf et al., 2010). This inadequate timing for HCN in the model simulations could be due to an overestimation of the Southern African contribution to the Northern African 20 loading and is visible on Fig. 10 (JJA).

Conclusions
We have presented a fast method to retrieve HCN and C 2 H 2 total columns from IASI spectra. The sensitivity of this method to the two species is mostly in the mid-upper troposphere. With this method, C 2 H 2 total columns can be retrieved globally with 25 5 % precision, while HCN abundances can be retrieved for abundances greater than 14377 ACPD 15,2015  Total columns have been retrieved globally for a three year period and compared to routine FTIR measurements available at Reunion Island (HCN and C 2 H 2 ), Wollongong (HCN), Jungfraujoch (C 2 H 2 ), and Izaña (HCN). The comparison between IASI 5 and FTIR retrieved total columns demonstrates the capabilities of IASI to capture the seasonality in HCN and C 2 H 2 in most cases.
Global seasonal distributions, as well as regional time series of the total columns, have been shown for the two species. IASI is able to capture persisting, seasonal and exceptional features for both species, and the observed patterns are in a general good 10 agreement with previous spaceborne studies ( ACE-FTS and MIPAS).

ACPD
are probably due to the 2010 great Amazonian fires [Lewis et al., 2011] influence.
At Wollongong HCN peaks also in October-November due to the Southern Hemisphere biomass burning season [Paton-Walsh et al., 2010]. We find maxima of around 11 415 Figure 8. Time series of HCN (left panel) and C 2 H 2 (right panel) measurements for Reunion Island (HCN and C 2 H 2 ), Wollongong (HCN only), Izaña (HCN only), and Jungfraujoch (C 2 H 2 only). IASI measurements are shown as daily and 1 • × 1 • means (red dots) with associated SDs (light red lines), and as monthly and 1 • × 1 • means (black circles and line) with associated SD (vertical black lines). Ground-based FTIR measurements are shown as daily means with associated total error by green crosses and lines. Correlation coefficients are given on each plot for daily means in red and for monthly means in black.