IASI-derived NH 3 enhancement ratios relative to CO for the tropical biomass burning regions

. Vegetation ﬁres are a major source of ammonia (NH 3 ) in the atmosphere. Their emissions are mainly estimated from "bottom-up" approaches which rely on uncertain emission factors. In this study, we derive new biome-speciﬁc NH 3 enhancement ratios relative to carbon monoxide (CO), ER NH 3 / CO – directly related to the emission factors, from the measurements of the IASI sounder on board the Metop-A satellite. This is achieved for large tropical regions and for a 8-year period (2008–2015). We ﬁnd substantial differences in the ER NH 3 / CO between the studied biomes with calculated values ranging 5 from 4.4 × 10 − 3 to 17 × 10 − 3 . For Evergreen Broadleaf Forest these are typically 75-100% higher than for Woody Savanna and Savanna biomes. This variability is attributed to differences in fuel types and size and is in line with previous studies. The analysis of the spatial and temporal distribution of the ER NH 3 / CO also reveals a (sometimes large) within-biome variability. On a regional level, Woody Savanna shows for example a mean ER NH 3 / CO for the region of Africa South of the Equator which is 50-100% lower than in the other ﬁve studied regions, probably reﬂecting regional differences in fuel type and burning conditions. 10 The same variability is also observed on a yearly basis with a peak in the ER NH 3 / CO observed for the year 2010 for all biomes. These results highlight the need for the development of dynamic emission factors that better take into account local variations in fuel type and ﬁre conditions. We also compare the IASI-derived ER NH 3 / CO with values reported in the literature, usually calculated from ground-based or airborne measurements. We ﬁnd a general underestimation over the referenced ER NH 3 / CO of about 40% for Woody Savanna and Savanna and up to a factor 1.5-4 for Evergreen Broadleaf Forest and Cropland. Beyond 15 a possible overestimation of the ER NH 3 / CO in the literature, the observed differences could also be related to various factors including instrumental limits, bias in the retrieval of the NH 3 columns, parameterization in the calculation of the ER NH 3 / CO or accumulation of CO in the studied regions during the ﬁre period.

carbon, vegetation fires also emit large amounts of reactive nitrogen species, among which ammonia (NH 3 ). With a contribution estimated to be about 13% (Galloway et al., 2004) of the total emissions, biomass burning is believed to be the second most important source of NH 3 after agriculture. From previous studies, it has been shown that biomass burning could affect significantly NH 3 concentrations in the atmosphere, especially in the tropics but also at higher latitudes (e.g., Bouwman et al., 1997;Coheur et al., 2009;Adon et al., 2010;Alvarado et al., 2011;Shephard et al., 2011;Adon et al., 2013;R'Honi et al., 5 2013; Whitburn et al., 2015Whitburn et al., , 2016aBenedict et al., 2017;Warner et al., 2017). Excess NH 3 in the environment is of great concern since it is responsible for many environmental issues such as eutrophication of terrestrial and aquatic ecosystems, soil acidification and loss of plant diversity (Aneja et al., 2001;Erisman et al., 2007). As the dominant alkaline species in the atmosphere, NH 3 rapidly combines with acid gases such as sulfuric acid (H 2 SO 4 ), nitric acid (HNO 3 ) and hydrochloric acid (HCl) resulting in the formation of secondary aerosols that are in turn impacting climate and human health (Bouwman et al.,10 1997; Aneja et al., 2001;Sutton et al., 2011;Behera et al., 2013;Lelieveld et al., 2015).
Until recently, most models of fire emissions were based on "bottom-up" approaches which rely on an estimation of the total burned biomass (BB, kg) combined with biome-specific emission factors (EFs), expressed as the mass of pollutant emitted per kilogram of BB (g kg −1 BB). Despite the numerous studies achieved in the past decades (e.g., Sinha et al., 2003;Yokelson et al., 2003;van der Werf et al., 2010;Wooster et al., 2011;Smith et al., 2014), the uncertainty on all parameters of these models 15 remain large. This is especially true for EFs, which have typical uncertainty of the order of 20-30% for frequently measured species (e.g. CO, CO 2 ) and up to 100% for species such as NH 3 which are not so well monitored (Langmann et al., 2009;Akagi et al., 2011). An accurate determination of the EFs is challenging partly because of the existence of a within-biome spatial and seasonal variability (van Leeuwen and van der Werf, 2011; Yokelson et al., 2011;Meyer et al., 2012;Mebust and Cohen, 2013; van Leeuwen et al., 2013;Castellanos et al., 2014;Schreier et al., 2014a). This variability is attributed to differences in fuel type 20 and burning conditions, the latter being itself controlled by climate, weather, moisture content, topography and fire practices (Andreae and Merlet, 2001;Korontzi et al., 2003;Yokelson et al., 2011;van Leeuwen and van der Werf, 2011;Castellanos et al., 2014). For nitrogen compounds, another main factor controlling the EFs is the Nitrogen content of the fuel (Andreae and Merlet, 2001;Jaffe and Wigder, 2012;Castellanos et al., 2014). Because it is generally not known to what extent EFs are based on a representative sample of a specific vegetation type (van Leeuwen and van der Werf, 2011;Castellanos et al., 2014), the 25 spatial and temporal variability in the EFs is usually not taken into account in the bottom-up approaches where EFs are taken from compilations of airborne and local measurements or from small fires burned under laboratory conditions (e.g., Andreae and Merlet, 2001;Akagi et al., 2011).
With their excellent spatial and temporal coverage, hyperspectral sounders on board satellites, directly measuring tropospheric concentration of trace gases in the atmosphere, offer a unique opportunity to determine EFs more accurately and to 30 capture their variability in time and space. Nowadays, the focus was principally on CO, nitrogen dioxide (NO 2 ) and aerosols (e.g., Pechony et al., 2013;Castellanos et al., 2014;Ichoku and Ellison, 2014;Mebust and Cohen, 2014;Schreier et al., 2014a, b). A recent study was also dedicated to formic acid (HCOOH) (Pommier et al., 2017). Until now, less attention has been given to NH 3 (Coheur et al., 2009;Alvarado et al., 2011;R'Honi et al., 2013;Luo et al., 2015). In this paper we derive biome-specific NH 3 enhancement ratios relative to CO (ER NH3/CO , also known as normalized excess mixing ratios) and relate them to EFs 35 (see section 2.2), over large tropical fire regions and long periods using the measurements of the Infrared Atmospheric Sounding Interferometer (IASI). The use of IASI is particularly suitable here because of its exceptional sampling (compared to other similar instruments, such as the Tropospheric Emission Spectrometer (TES) (Shephard et al., 2011)) and to our knowledge, it is the first time such a study focusing on biomass burning ERs is carried out at this scale for NH 3 . Section 2 hereafter briefly describes the datasets used and introduces the methodology for calculating the enhancement ratios. It also motivates the selec-5 tion of the regions studied. The results from our analyzes are presented and discussed in section 3, which is further divided in two main parts. The first part analyzes the variability of ER NH3/CO between and within the different biomes (an extensive comparison with ERs reported in the literature is also provided) while the second analyzes the interannual and seasonal evolution of ER NH3/CO . Summary and conclusion are given in section 4.

Instruments and data products
IASI is a nadir-looking high resolution Fourier Transform Spectrometer on board the polar orbiting sun-synchronous Metop (Meteorological Operational) satellites. The two first IASI sounders were launched in 2006 and2012 (Metop-A and-B, respectively). A third instrument is scheduled for launch in 2018 and will ensure at least 18 years of consistent measurements (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015)(2016)(2017)(2018)(2019)(2020)(2021)(2022)(2023). IASI covers the entire globe twice daily (09:30 and 21:30 local time when crossing the equator) with a relatively 15 small elliptical footprint on the ground varying from 12 × 12 km (at nadir) up to 20 × 39 km (off nadir); depending on the viewing angle (Clerbaux et al., 2009). Its large and continuous spectral coverage of the thermal infrared band region (645-2760 cm −1 ), its medium spectral resolution (0.5 cm −1 apodized) and its low instrumental noise (∼0.2 K at 950 cm −1 and 280 K) make it an invaluable instrument for monitoring atmospheric composition (Clerbaux et al., 2009). CO is retrieved from IASI measurements using the FORLI (Fast Optimal Estimation Retrievals on Layers for IASI) software . The 20 retrieval of NH 3 is based on a new and flexible retrieval algorithm, which relies on the calculation of a so-called Hyperspectral Range Index (HRI) and subsequent conversion to a NH 3 total column (molec.cm −2 ) using a neural network (Whitburn et al., 2016b). For a detailed description of the retrieval methods and parameters, we refer the reader to Whitburn et al. (2016b) for NH 3 and Hurtmans et al. (2012) for CO. The validation of FORLI-CO profiles and columns have shown good agreement overall using in-situ, aircraft and satellite observations (Pommier et al., 2010;De Wachter et al., 2012;Kerzenmacher et al., 25 2012;George et al., 2015). For NH 3 columns, the validation has started but is more difficult considering the important spatial and temporal variability of NH 3 and the scarcity of correlative ground-and airplane-based measurements in many regions of the world (Van Damme et al., 2015). Two studies, based on a previous NH 3 retrieval algorithm also using the HRI but relying on a two-dimensional look-up tables for the conversion into a NH 3 total column (molec.cm This work makes use of 8 years (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) of daily global NH 3 and CO total columns (molec.cm −2 ) from the measurements of IASI on board Metop-A. Only daytime satellite observations have been considered as they usually show a better sensitivity, especially to NH 3 . We also have assumed a similar sensitivity for IASI to NH 3 and CO in the lower layers of the atmosphere, which is not expected to introduce a significant bias in the studied regions due to a generally positive thermal contrast prevailing during daytime (Clarisse et al., 2010;Van Damme et al., 2014). A more important bias may result from the use of a unique vertical profile shape in the retrieval scheme of NH 3 total columns which is therefore not representative of the large variety of profiles observed above biomass burning plumes. Whitburn et al. (2016b) have calculated that the use of an 5 alternative profile could affect the retrieved column up to 50%. This is important to keep in mind for the analyses presented next.
We also used in support to the selection of the studied regions and the NH 3 and CO columns, active fires detection data and nitrogen dioxide (NO 2 ) total columns (molec.cm −2 ). Detected active fires are taken from the Global Monthly Fire Location Product (MCD14ML, Level 2, Collection 5) developed by the University of Maryland from the measurements of the MODerate 10 resolution Imaging Spectroradiometer (MODIS) on board the NASA Terra and Aqua satellites (Justice et al., 2002;Giglio et al., 2006). Active fires are monitored at a resolution of 1×1 km 2 with fires as small as 100 m 2 detected. NO 2 total columns are retrieved from the measurements of the Global Ozone Monitoring Experiment (GOME-2) also on board the Metop satellites and working in the UV-Vis spectral band region (Valks et al., 2011). 15 From the IASI NH 3 and CO total columns (molec.cm −2 ), we have derived NH 3 enhancement ratios relative to CO (ER NH3/CO ) defined as the ratio of the number of emitted molecules of NH 3 (here the NH 3 total column) over the emitted molecules of the reference species CO (here the CO total column) (Andreae et al., 1988;Lefer et al., 1994;Hobbs et al., 2003). The choice of CO as reference species is here particularly suitable as it is a dominant species emitted by fires and has a lifetime of several weeks in the free troposphere. One main advantage of the ERs compared to the EFs is that ERs calculation only requires simultaneous 20 measurements of the studied (NH 3 ) and the reference species (CO) while EFs calculation requires fuel information that are not always available or completely reliable (Andreae and Merlet, 2001). In fire plumes, ERs can be estimated following (Goode et al., 2000;R'Honi et al., 2013):

Enhancement ratios
When a lot of measurements are available, which is often the case for IASI-derived measurements owing to its excellent 25 spatial and temporal resolution, average ER NH3/CO can be estimated from the slope of the linear regression of NH 3 versus CO (Andreae and Merlet, 2001;Coheur et al., 2009). The ERs can also be derived directly from the EFs by multiplying the ratio EF NH3 /EF CO with the ratio of the molar masses M CO /M NH3 (Andreae and Merlet, 2001). This will be used here to convert the reported EF values from ground-based and airborne studies into ERs in order to allow the comparison with our IASI-derived ER NH3/CO . 30

Selection of the areas and biomes and calculation of the ER NH3/CO
One of the key steps in this study is the selection of the areas of interest for the calculation of the ER NH3/CO . To be relevant, ER NH3/CO need to be calculated for areas were fires are the dominant source of emissions of NH 3 and CO. The selection has been done on a pixel basis. We have first calculated the linear regressions, globally on a 1°× 1°grid, between the monthly means of the couples NH 3 -CO total columns (molec.cm −2 ), NH 3 -NO 2 total columns, and NH 3 total columns-number of 5 active fires (#fires). We have next selected the pixels for which a correlation coefficient (r) higher than 0.3 was found for the three couples of regression (NH 3 -CO, NH 3 -NO 2 and NH 3 -#fires). These are shown in Fig. 1 (colored pixels) and constitute the areas considered for the calculation of the ER NH3/CO . Pixels with a r higher than 0.3 for the considered couple but not for (at least) one of the two other couples are shown in gray. The idea behind this selection procedure is that a good correspondence between the monthly means of NH 3 , CO and NO 2 total columns provides an indication of a dominant contribution of the hours in the studied regions (Dentener and Crutzen, 1994;Aneja et al., 2001;Whitburn et al., 2015Whitburn et al., , 2016a, NH 3 is more likely 20 to be transported over longer distances. This can be seen on the NH 3 -CO correlation map where positive correlations are also found over seas downwind of the source areas. For each of the selected pixels, we have next calculated an ER NH3/CO per year between 2008 and 2015 from the slope of the linear regression between NH 3 and CO retrieved columns (molec.cm −2 ). To take into account the NH 3 and CO columns most likely related to fire emissions, we have only considered IASI measurements located within 50 km from a fire. We have also 25 included a quality filter on the NH 3 and CO measurements: only total columns with a relative error lower than 100% for NH 3 and 25% for CO were retained for the regression. Finally, as a post-filtering, we have only kept for the analysis the ER NH3/CO for which the linear regressions between NH 3 and CO columns show a correlation coefficient larger than 0.3 and for which we have more than 10 measurements. The impact of these pre-and post-filters on the calculated ER NH3/CO are discussed in section 3.1. An example of a linear regression between NH 3 and CO for one of the selected pixels (Evergreen Broadleaf Forest 30 (EBF) in Indonesia) is given in Fig. 2.
For this study, we focus on the four dominant biomes in the selected pixels. These were identified using the MODIS Land Cover Type product (MCD12Q1) with the 17-class International Geosphere-Biosphere Program classification (IGBP) (Friedl et al., 2010) (Fig. 3). The four selected classes are 1) the Evergreen Broadleaf Forests (EBF), 2) the Woody Savannas (WS), 3) the Savannas (S) and 4) the Crops together with the Crop and Natural Vegetation Mosaic (C+CNVM), here denoted C. Fig. 3 also shows the distribution of the mean yearly ER NH3/CO averaged over the time period 2008-2015 for the selected pixels.
A first analysis of the distribution of the ER NH3/CO reveals a variability between the four biomes, especially in Africa North of the Equator and in Central South America where a gradient is observed between EBF and WS and between EBF and S, respectively, with higher ER NH3/CO found for EBF. A clear gradient is observed as well in Africa South of the Equator from 5 the northwest to the southeast.
The pixel-based ER NH3/CO have next been grouped by biome to analyze their regional and temporal variability. In addition, to facilitate the study of the spatial distribution of the ER NH3/CO , we have defined six main regions which include the majority of the pixels of interest (see Fig. 1 Yokelson et al., 2011) and is mainly attributed to differences in fuel size and density: EBF, characterized by dense fuel, is indeed dominated by smoldering combustion, which emits more reduced or incompletely oxidized products (among them NH 3 and CO) than grassland (van Leeuwen and van der Werf, 2011). One should note, however, that Kaiser et al. (2012) reported higher ER NH3/CO for Savannas than for tropical forests. For Crop (C), the mean ER NH3/CO (9.0×10 −3 ) is close to the ER NH3/CO (EBF) but is more difficult to interprete because the biome probably includes different types of fuel. Figure 5, 25 representing the cumulative frequency of the pixel-based yearly ER NH3/CO per biome, also shows the biome-trends in the ER NH3/CO . EBF and C have for example about 35% of the calculated ER NH3/CO above 0.01 while this value corresponds to only about 10-15% for S and WS. These differences in the ER NH3/CO between biomes are, however, not necessarily found when looking at the average ER NH3/CO at the region scale. For Central America (C.AM.) and South-East Asia (SE.ASIA) in particular, the differences between ER NH3/CO are low (of the order of 5-10%). For South America (S.AM.), ER NH3/CO (EBF) 30 is about twice higher than ER NH3/CO (S) but close to ER NH3/CO (WS) (within 10%).
When comparing the ER NH3/CO by biome between the six regions in Figure 4 (solid error bars), we find good agreements but also large differences, in line with what has already been reported by, for example, van Leeuwen and van der Werf (2011) observed, but to a lesser extent, for the C biome ranging between 7.3×10 −3 for SE.ASIA and 10.5×10 −3 for C.AM. Note that 5 this intra-biome variability is also found within a given region, as observed in Fig. 3 and as evidenced by the sometimes large standard deviation (std) associated with the mean ER NH3/CO (e.g. EBF in the AFR.NEQ. region with a std of about 0.01). As mentioned in section 1, these differences can be explained by changes in the fuel type (size and density) but also the climate, weather, topography, moisture and N content, and fire practices. In addition for EBF, different regional deforestation practices could also lead to variation in the ER NH3/CO (van Leeuwen and van der Werf, 2011). It should finally be mentioned that for the 10 AFR.NEQ. region, the measured NH 3 columns at the end of the fire period probably originate from the combination of both biomass burning emissions and another source, possibly agriculture as suggested in Whitburn et al. (2015); this might therefore introduce a bias in the ER NH3/CO . Overall, these results clearly highlight the need for developing new regional-dependent EFs, in order to improve the representativeness of estimations from bottom-up inventories.
The comparison of the IASI-derived ER NH3/CO with the values reported in the literature from ground-based or airborne 15 studies (see Table 1) shows a general overestimation of the latter (or an underestimation from IASI), especially for the biomes EBF and C where a factor 1.5-4 difference is observed. The only exception is for Kaiser et al. (2012) for EBF for which a better correspondence (within 50% difference) is found. Note that the largest difference for EBF is with an ER NH3/CO reported for tropical dry forest (Yokelson et al., 2011), but the latter is likely not representative for the complete Evergreen Broadleaf Forest class. For S and WS, the agreement is much better (with a maximum IASI underestimation of about 40%) except for Andreae 20 and Merlet (2001), Bertschi et al. (2003) and Kaiser et al. (2012) reporting ER NH3/CO up to a factor 4 higher. Note that, in Bertschi et al. (2003) ER NH3/CO are derived from smoldering logs for which higher values are logically expected. Note also that for WS, the ER NH3/CO are compared here to values reported for Savannas, which are usually included with S into the same biome. While an overestimation of the average ER NH3/CO (or EF NH3 ) in the literature is possible, other reasons are likely to play a role. First, the differences with the IASI-derived ER NH3/CO could also be (at least partly) explained by the consideration 25 in our work of IASI measurements within 50 km of an active fire, while ground and airborne measurements are done in the direct vicinity of the fire. Second, another possible reason might lie in the difficulty for MODIS to detect smoldering fires, causing the IASI-derived ER NH3/CO to reflect preferentially the flaming phase of the vegetation fires. Third, an accumulation of CO in the region during the fire period (due to its much longer lifetime compared to NH 3 ) might introduce a bias in the calculated ER NH3/CO . Finally, the differences with the reported ER NH3/CO could also be due to the chosen methodology for 30 the calculation of the ER NH3/CO . To verify this, we have recalculated mean biome-specific ER NH3/CO for the six regions (not shown) by varying one by one the pre-and post-filters considered before (see section 2.3). We have performed four tests: 1) with a maximum distance of the NH 3 total column to a detected fire of 30 km and 2) 100 km (instead of 50 km), 3) with a maximum error on the NH 3 total column of 75% (against 100%) and 4) by filtering out the ER NH3/CO for which the linear regressions between NH 3 and CO columns show a correlation coefficient (r) of the linear regression lower than 0.6 (instead 35 7 Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-331, 2017 Manuscript under review for journal Atmos. Chem. Phys. Discussion started: 4 May 2017 c Author(s) 2017. CC-BY 3.0 License. of 0.3). We find a very limited impact of the distance to a fire and the error on the NH 3 column allowed, with differences of only about 3-8% (interestingly, an increase -decrease-of the tolerance on the maximum distance to a fire systematically slightly decrease -increase-the mean ER NH3/CO ). In contrast, an increase to 0.6 of the threshold on the correlation coefficient introduces a large increase in the mean ER NH3/CO of about 25-35% (and up to 40% for WS in the AFR.SEQ. region). Taking into account this increase, we find mean ER NH3/CO closer (but still below) to what is reported in the literature, especially for WS 5 and S. The agreement would become even better if we consider in addition a possible bias due to the use of a non-representative NH 3 vertical profile considered for the retrieval of the NH 3 , as mentioned in section 2.1. Note that despite the impact of the pre-and post-filters chosen, the analysis on the regional and inter-biome variability in the ER NH3/CO remains valid. At a regional level (all biomes combined), a comparison with the satellite-derived ER NH3/CO based on TES measurements (Luo et al., 2015) shows again higher values of about 50% compared to our calculated ER NH3/CO (Table 1). However, higher compared to our AFR.SEQ. region.

ER NH3/CO interannual and seasonal variability
In this second part, we focus our analysis on the temporal variability in the ER NH3/CO . Fig. 6 shows the mean ER NH3/CO averaged by biome and by year (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015). The solid line represents the 8-years average for each biome. We find an interannual variability in the mean ER NH3/CO up to a factor two for the four studied biomes. Interestingly, the minimum ER NH3/CO is 20 found in 2012 for all biomes. Similarly the highest mean ER NH3/CO is observed in 2010 for all biomes (especially marked for EBF) except for S for which the maximum is found in 2008 (despite an ER NH3/CO for 2010 also above the 8 years average).
When analyzing the variability in the yearly averaged ER NH3/CO for each region separately (Fig. 7), we find that the high mean ER NH3/CO of 2010 for EBF is exclusively carried by the AFR.NEQ. region with a mean ER NH3/CO of 40×10 −3 (against about 15×10 −3 for the other years in the region). For the WS biome, the peak of 2010 is mainly due to the S.AM., AFR.NEQ. and 25 SE.ASIA regions with an ER NH3/CO about a factor 1.5-2 higher compared to the other years. This important variability in the ER NH3/CO are probably due to differences in the burning conditions from one year to another. One possible reason to explain the high mean ER NH3/CO for 2010 in the different regions is the El Niño Southern Oscillation (ENSO) event that occurred that year and that was responsible for severe droughts and increased fire activity in the studied regions (Whitburn et al., 2015). This is however probably not sufficient to explain the 3 times increase for EBF for 2010 in the AFR.NEQ. region but no 30 clear evidence of other processes influencing the ER NH3/CO were found for that year. Surprisingly, the same increase in the ER NH3/CO are not observed for the year 2015, which was the strongest El Niño year since 1997 (Chisholm et al., 2016). For WS, high ER NH3/CO are in addition observed for 2011 for South and Central America (S.AM. and C.AM.). However, this has small impact on the global yearly ER NH3/CO , which is mainly driven by the two regions in Africa (AFR.NEQ. and AFR.SEQ.),

8
Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2017-331, 2017 Manuscript under review for journal Atmos. Chem. Phys. representing about 33% and 40% of all the calculated ER NH3/CO for WS, respectively (see Fig. 4). For the S biome, the yearly ER NH3/CO is largely dominated by the AFR.SEQ. and the S.AM. (47% and 33% of the ER NH3/CO , respectively). Despite being not as pronounced on the global yearly ER NH3/CO (Fig. 6)  fire season in the AFR.NEQ. region (for the four biomes) than for the beginning of the fire period. The interannual variability in the ER NH3/CO was also found important (up to a factor 2), with a peak for 2010 for each biome, possibly related to the severe droughts that have occurred that year in the studied regions consequently to an important El Niño event. The important variability of the ER NH3/CO both in time and space clearly shows the need for developing dynamic datasets of EFs which better take into account the fuel type and fire conditions.

5
In comparison to the values reported in the literature, mainly from ground-based and airborne studies, we found a general underestimation of the mean IASI-derived ER NH3/CO of about 40% for S and WS and up to a factor 1.5-4 for the EBF and C biomes. These differences may be explained by various factors including 1) the parametrization (pre-and post-filtering of the data) considered for the calculation of the ER NH3/CO , 2) a bias towards the flaming phase due to the selection of IASI Acknowledgements. IASI has been developed and built under the responsibility of the Centre National d'Études spatiales (CNES, France). It is flown on board the Metop satellites as part of the EUMETSAT Polar System. The IASI L1 data are received through the EUMETCast near real-time data distribution service. We thank the NASA for providing MODIS fire radiative power data. We also acknowledge the use of the MODIS global land cover map. We thank EUMETSAT for the use of the operational EUMETSAT O3MSAF NO2 product. The algorithm for the retrieval of the NO2 total columns used in this work has been developed in the context of the Satellite Application Facility on Ozone and Andreae, M. and Merlet, P.: Emission of trace gases and aerosols from biomass burning, Global Biogeochemical Cycles, 15, 955-966, 2001. plumes, Atmospheric Chemistry and Physics, 9, 5655-5667, doi:10.5194/acp-9-5655-2009. Crutzen, P. and Andreae, M.: Biomass burning in the tropics: impact on atmospheric chemistry and biogeochemical cycles., Science, 250, 1669Science, 250, -1678Science, 250, , doi:10.1126Science, 250, /science.250.4988.1669Science, 250, , 1990.