The IASI (Infrared Atmospheric Sounding Interferometer) (Clerbaux et al., 2009) instrument, on board the MetOp-A platform since October 2006, is a nadir-viewing Fourier transform spectrometer. It operates in the thermal infrared between 645 and 2760 cm-1 with an apodized resolution of 0.5 cm-1. The field of view of the instrument is composed of a 2 × 2 matrix of pixels with a diameter at nadir of 12 km each. IASI scans the atmosphere with a swath width of 2200 km and crosses the equator at two fixed local solar times (09:30, descending mode, and 21:30, ascending mode), allowing the monitoring of atmospheric composition twice a day at any location. The large spectral coverage, high radiometric sensitivity and accuracy, and rather high spectral resolution of the instrument allow it to measure the global distribution of several important atmospheric species (e.g. Boynard et al., 2009; George et al., 2009; Clarisse et al., 2011).

The IASI ozone profiles and partial columns considered in this paper have been retrieved using the method described in Eremenko et al. (2008). The retrieval is performed using the radiative transfer model KOPRA (Karlsruhe Optimised and Precise Radiative transfer Algorithm) and its inversion module KOPRAFIT (Stiller et al., 2000; Höpfner et al., 2001), both adapted to the nadir-viewing geometry. A constrained least squares fit method with an analytical altitude-dependent regularization is used (Kulawik et al., 2006). The applied regularization method is detailed in Eremenko et al. (2008). To summarize, the regularization matrix is a combination of first-order Tikhonov constraints (Tikhonov, 1963) with altitude-dependent coefficients. The coefficients are optimized both to maximize the degrees of freedom (DOFs) of the retrieval and to minimize the total error on the retrieved profile. Compared to previous studies using this algorithm (Eremenko et al., 2008; Dufour et al., 2010, 2012), several changes have been made. The emissivity of the surface is now taken into account based on a global monthly IASI-derived climatology (Zhou et al., 2011) allowing a better retrieval above arid regions. Different a priori and constraints are used depending on the tropopause height. This new scheme was introduced to reduce possible compensation effects during the retrieval procedure. An automatic detection of the tropopause height (calculated from the temperature profile retrieved from IASI using the definition based on the lapse rate criterion; WMO, 1957) has been introduced to discriminate between polar, midlatitudes, and tropical situations. If the tropopause is lower than 10 km, the polar constraint and a priori profile are used. If the tropopause is between 10 and 14 km, the midlatitude constraint and a priori profile are used. If the tropopause is higher than 14 km, the tropical constraint and a priori are used. The midlatitude and tropical regularization matrices are those already used in Eremenko et al. (2008) and Dufour et al. (2010, 2012) respectively. The polar constraint has been specifically developed following the same method as in Eremenko et al. (2008). The a priori profiles are compiled from the ozonesonde climatology of McPeters et al. (2007). The midlatitude a priori profile is set to the climatological profile of the 30–60∘ N latitude band for summer. The tropical a priori profile is set to the climatological profile of the 10–30∘ N latitude band over the year. The polar profile is set to the climatological profile of the 60–90∘ N latitude band for summer. As the version of the ozone product used in this study differs significantly from the version extensively validated in Dufour et al. (2012), a new validation against ozonesondes has been conducted and the results are presented in Sect. 3. The modifications of the algorithm do not influence the vertical sensitivity of IASI. As shown in Dufour et al. (2010, 2012), two semi-independent partial columns of ozone between the surface and 12 km can be considered: the lower-tropospheric column integrating the ozone profile from the surface to 6 km altitude – above sea level (a.s.l.) – and the upper-tropospheric column integrating the ozone profile from 6 to 12 km altitude. Note that the latter column can include stratospheric air masses depending on the tropopause height. The averaging kernels give information on the vertical sensitivity and resolution of the retrieval. The lower-tropospheric column shows a maximum sensitivity typically between 3 and 4 km with a limited sensitivity to the surface (Dufour et al., 2012). This implies that the ozone concentration profile in the lower troposphere is preferentially incremented at these altitudes during the retrieval process, independent of whether the true ozone profile is perturbed at other altitudes, especially at the surface. Moreover, it is worth noting that the partial columns are only semi-independent which means that they may include partial information from altitudes outside their altitude range. For example, the lower-tropospheric column includes information from altitudes higher than its upper limit (6 km). In order to estimate the fraction of contamination of the lower-tropospheric column by higher altitudes, we calculated the ratio between the integral of the averaging kernel of the lower-tropospheric column from 6 to 60 km and the integral from the surface to 60 km. Higher atmospheric layers contribute to about 20 to 30 % of the lower-tropospheric column in the midlatitude air masses (not shown). Note that only the morning overpasses of IASI are considered for this study in order to remain in thermal conditions with a better sensitivity to the lower troposphere.

The CO data used here are retrieved from the IASI spectra within the 2143–2181.25 cm-1 spectral range using the FORLI-CO (Fast Optimal Retrievals on Layers for IASI for CO) retrieval code from the Université Libre de Bruxelles (ULB). FORLI-CO retrievals give CO concentration profiles using the optimal estimation method (Rodgers, 2000) and a single a priori profile. More details are given in Hurtmans et al. (2012). The IASI FORLI-CO product used in this study is the total column, publicly available from the Ether website (http://www.pole-ether.fr). Note that only half of the pixels are available for the year 2008. This explains the difference in measurement density between O3 and CO observations in the different figures. Carbon monoxide is often used as an indicator of biomass burning and anthropogenic pollution (e.g. Edwards et al., 2004; McMillan et al., 2010). In this study, we use the IASI CO columns as an anthropogenic pollution tracer.

Meteorological data from the ECMWF ERA-Interim reanalysis are used in our analyses. The reanalysis is based on a 4-D-Var assimilation system with a 12 h analysis window. The spatial resolution of the data set is approximately 80 km on 60 vertical levels from the surface up to 0.1 hPa (Dee et al., 2011). In our analyses, the meteorological parameters are taken at 00:00 UTC, corresponding roughly to the morning overpass time of IASI. The main variables considered in this study are the geopotential height, the potential vorticity (PV), and the horizontal wind field (u and v components), as well as the equivalent potential temperature, the vertical velocity and the convective available potential energy. The geopotential height associated with horizontal wind at 850 hPa gives a proxy for describing the weather situation and horizontal transport in the lower troposphere, whereas the same parameters at 300 hPa describe the situation in the UTLS. We also calculate the equivalent potential temperature at 850 and 300 hPa from temperature, relative humidity and specific humidity fields (Bolton, 1980) as an indicator of air mass origin (Holton, 2004). Potential vorticity (PV) is often used as a tracer of tropopause height and of air mass origin (e.g. Bethan et al., 1996). PV values between 1 and 1.6 PVU are representative of the upper troposphere, whereas PV values larger than 1.6 PVU are indicators of air mass origin above the dynamical tropopause. In this study, we consider mainly PV averaged at between 300 and 500 hPa with a 50 hPa interval as we are above all interested in the impact of stratospheric air masses on the free troposphere. In order to investigate the ascending motion of air masses, especially from the boundary layer towards the free troposphere within weather systems, we examine the vertical velocity at different pressure levels as well as the convective available potential energy (CAPE), which provides information on the capability of the low-pressure system to vertically transport air masses by convection.

Figure 1 describes the meteorological situation of this episode of high ozone. A large cold front extending from Mongolia to southern China on 4 May 2008, from north China to the southern Japanese islands on 5 May 2008, and moving eastward from Japan on 6 May 2008 characterizes the low-pressure system (Fig. 1a–c). The regions behind the frontal area and north of the polar jet, situated between 35 and 40∘ N on these dates (Fig. 1d–f), are strongly influenced by polar air masses with tropopause heights below 9 km (Fig. 1g–i). The 300–500 hPa mean PV values are larger than 1.6 PVU for the same regions, indicating that the upper troposphere is under the influence of lower-stratospheric air masses (Fig. 2j–l). The spatial correlation of low tropopauses and large PV values indicates that reversible subsiding ozone transfer affects the upper troposphere in this case. We then expect an ozone enhancement in the upper troposphere for these regions, and we will see in the following how IASI describes this ozone enhancement induced by ozone subsidence. The analysis of the upper-tropospheric columns shows that IASI observes columns larger than 40 DU in the regions affected by low tropopauses and large PV (Fig. 2g–i). A step gradient between 30 and 40 DU is observed in the upper-tropospheric ozone distribution reflecting the step gradient in the PV distribution. The very good spatial correlation of the high-UT ozone structures with those of high PV leads us to consider that the upper-tropospheric columns of ozone retrieved from IASI can be used as a proxy to determine the regions affected by subsiding ozone from the lower stratosphere. The threshold of 40 DU seems to be relevant to this identification.

The question now is to determine to what extent IASI is able to provide information about the low-pressure system's influence on the lower-tropospheric ozone distribution. The physical processes, which may affect the lower-tropospheric ozone distribution, are (i) the reversible ozone subsidence associated with low-tropopause heights, which induces an enhancement of ozone in the upper and free troposphere and then partly in the lower troposphere; (ii) irreversible stratosphere–troposphere exchanges, which also lead to ozone enhancement. The first process is expected to affect ozone distribution on a synoptic scale, whereas the second process is more localized. Figure 2a–c show that lower-tropospheric ozone columns larger than 28 DU are observed with IASI in the vicinity of the low with similar spatial patterns to the UT columns and the PV distribution. The observed enhancement in the lower-tropospheric column (surface 6 km) arises from (1) the actual (reversible) transfer of ozone to the free troposphere, (2) the definition of the LT columns by itself, (3) the limited vertical resolution of the retrieval and the associated smoothing of the vertical profile. Indeed, the LT columns are defined as the columns from the surface up to 6 km. Consequently, when the tropopause is low (below 9 km), the LT column arithmetically includes layers with upper-tropospheric characteristics. Moreover, due to the limited vertical resolution of the retrieval and the associated smoothing of the vertical profile, the lower-tropospheric column is partly contaminated by ozone from outside the column altitude boundaries, as discussed in Sect. 2.2. This may contribute to an overestimation of the lower-tropospheric column. However, it is difficult to estimate this overestimation in our case because no ozonesonde observations were available along the path of the low.

We show with this case study that having the IASI UT and LT ozone columns allows us to determine those regions affected by the subsiding transfer of ozone occurring behind the frontal area and whether the lower troposphere is affected. We will now examine if and how IASI can be used to characterize irreversible STEs. The analysis of the PV distribution at different pressure levels allows the identification of the region in the vicinity of the low where STE occurs. In the case study from 4 to 6 May 2008, we identify two regions with high PV values down to 600 or 500 hPa in the path of the low (not shown): one on the east coast of Korea on 5 May (∼ 39∘ N, 128∘ E) and one offshore to the northeast of Tokyo (∼ 39∘ N, 142.5∘ E). The STE is situated on the southeast flank of the low-pressure system, behind and in the southern part of the cold front in the two cases. Figure 3 displays the longitudinal and latitudinal vertical section of ozone at 128∘ E and 39∘ N for 5 May (top) and at 142.5∘ E and 39∘ N for 6 May (bottom). On 5 May, strong stratospheric intrusion of ozone is observed between 38 and 39∘ N and between 125 and 130∘ E. The vertical section at 128∘ E shows that the free and lower troposphere are still connected to the polar UTLS reservoir that day. On 6 May, a stratospheric intrusion is observed between 36 and 38∘ N and between 140 and 143∘ E. The vertical section along 142∘ E shows that the enhanced ozone in the lower troposphere (below 7 km) is partly disconnected from the polar UTLS reservoir. Backtrajectories performed with the HYSPLIT (Hybrid Single Particle Lagrangian Integrated Trajectory) model (Draxler and Rolph, 2015; Rolph, 2015) show that the 3 km altitude air masses located in this area originate at altitudes between 5 and 7 km the day before in north China and Inner Mongolia (Fig. S1 in the Supplement). The tropopause height was around 7–8 km on 5 May 2008 for these regions (Fig. 1h). This means that the air masses reaching the area northeast of Tokyo at 3 km on 6 May have a UTLS origin and transport ozone-rich air into the lower troposphere. Thus, we show that the downward transport from the UTLS affects ozone concentrations in the lower troposphere for specific regions on the southeastern flank of the weather system, whereas the perturbation of the tropopause associated with this system influences upper- and lower-tropospheric ozone over larger areas in the vicinity of the low.

On 5 May 2008, an anticyclone forms over central east China and the North China Plain. The northwesterly winds reaching the NCP change progressively to southwesterly winds from 4 to 6 May with low winds and then a stagnant situation on 5 May (Fig. 1a–c). This, associated with low cloud coverage and increasing radiation, is a situation favourable to the accumulation of primary pollutants over the NCP and to the photochemical production of ozone due to both local emissions and the regional transport of pollutants. The question here is how IASI is able to describe this situation. The CO columns observed with IASI show enhanced values over the NCP on 5 and 6 May (Fig. 2e and f). Considering CO as a pollution tracer, enhanced IASI CO columns can be used to evaluate the build-up of pollutants. Concomitantly, lower-tropospheric ozone columns as large as 30 DU are observed (Fig. 2b and c). A significant spatial correlation (r= 0.6) is calculated between CO and lower-tropospheric ozone columns for a square region including the NCP (35–41∘ N, 114–122∘ E) on 5 May. In addition, the upper-tropospheric ozone column does not show enhanced values over the NCP for these 2 days (Fig. 2h and i). The analysis of the vertical section of ozone distribution on 5 May shows that the large ozone concentrations in the Beijing region (Fig. 3, ∼ 39∘ N and ∼ 116∘ E) and across the NCP are retrieved below 6 km and are disconnected from the UTLS region. The maximal values of ozone are retrieved between 2 and 3 km over Beijing (Fig. 3) in agreement, considering the vertical sensitivity and resolution of IASI, with in situ measurements, which frequently report high ozone concentrations at an altitude of 1.5–2 km above Beijing during April–May (Huang et al., 2015). This associated with the correlation with CO suggests that the enhanced ozone observed with IASI is mainly due to the photochemical transformation of primary pollutants emitted over the NCP. To evaluate the degree of the photochemical production of ozone, we calculate the equivalent or mean mixing ratio corresponding to the CO and LT O3 columns. This allows us to estimate a relative enhancement ratio of O3 to CO of 0.14 and 0.08 on 5 and 6 May respectively. These values are in agreement with the typical values ranging between 0 and 0.3 reported over east Asia by Tanimoto et al. (2008). The estimated enhancement ratio remains quite low, suggesting an early stage of ozone production.

On 11 May 2008, a cut-off low forms over Inner Mongolia (Fig. 4a). The cut-off low is not yet completely dissociated from the polar reservoir. A band of upper-tropospheric columns larger than 40 DU is observed by IASI between 35 and 45∘ N (Fig. 5g). As seen in Sect. 4, it indicates that the region is under the influence of subsiding ozone. The lower-tropospheric ozone columns do not show a clear enhancement for the same latitude band. On that day, subsiding ozone affects the lower-tropospheric ozone only moderately.

On 12 May 2008, the cut-off low is well dissociated from the western current and its centre reaches the Bohai Sea (Fig. 4b). Upper-tropospheric ozone columns larger than 45 DU are retrieved all around the cut-off low. Lower-tropospheric ozone columns larger than 32 DU are observed especially in the southwestern part of the low, just above the NCP (Fig. 5b). The analysis of the vertical section of ozone at 117∘ E (Fig. 7a) shows that the subsiding transfer of ozone due to the tropopause perturbation strongly affects lower-tropospheric ozone north of 33∘ N. At 32∘ N, the ozone enhancement observed in the lower troposphere is not connected to the UTLS reservoir, suggesting a possible photochemical origin for this enhancement. IASI CO columns are also enhanced in the NCP region and partly correlated with the enhanced ozone columns (Fig. 5e). This indicates that pollution likely plays a concomitant role in the ozone enhancement in that case.

On 13 May 2008, the centre of the cut-off low moves slightly to the east and reaches the Yellow Sea (Fig. 4c). As for the previous day, large upper- and lower-tropospheric ozone columns are observed with IASI in the vicinity of the low (Fig. 5c and i). The two columns are slightly smaller than the day before over the NCP. The analysis of the vertical section of ozone at 115∘ E (Fig. 7b) shows that the subsiding transfer of ozone due to the tropopause perturbation is less effective than the previous day. Even if the lower-tropospheric ozone remains partly connected to the UTLS reservoir north of 33∘ N, secondary maxima are observed at ∼ 4 km of altitude, suggesting that an additional source of ozone may contribute to the LT ozone enhancement. South of 33∘ N the ozone enhancement is clearly located in the lower troposphere. The good spatial correlation of LT ozone enhancement and the strong CO enhancement observed all over the NCP (Fig. 5f) confirms that pollution plays a concomitant role in explaining the ozone distribution in the lower troposphere over the NCP for this day.

On 14 May 2008, the cut-off low shifts to the Sea of Japan (Fig. 4d). A large area including north China, Korea, and reaching Japan shows upper-tropospheric columns larger than 40 DU (Fig. 6g), which indicates the region is under the influence of subsiding ozone. Within this area, the largest LT ozone columns are observed in an area less extended and situated on the southeastern flank of the low, mainly over the Sea of Japan (Fig. 6a). The lower troposphere is then certainly under the influence of the UTLS.

Over China, an anticyclonic situation starts to develop south of the NCP inducing a change in the wind regime and warmer conditions from 14 May (Fig. 4d). Enhanced CO columns and lower-tropospheric ozone columns are retrieved with IASI over the NCP (Fig. 6a and d) with moderate UT ozone columns. The analysis of the vertical section of ozone concentrations at 35∘ N shows that very large ozone concentrations are retrieved for the entire free and upper troposphere in the eastern part of the section (Fig. 7c). This corresponds to the region over the Sea of Japan under the direct influence of the cut-off low and then greatly influenced by the UTLS. The situation is different over the NCP: ozone concentrations in the upper troposphere are moderate and a distinct maximum in the lower troposphere is clearly visible. This, associated with CO enhancement over the NCP in good spatial correlation with LT ozone, indicates that the ozone enhancement observed with IASI over the NCP is of anthropogenic origin and related to the photochemical production of ozone. On that day, the estimated enhancement ratio of O3 to CO is 0.11, in agreement with the enhancement ratio calculated for the previous case study.

On 15 and 16 May 2008, strong enhancements of CO and lower-tropospheric ozone are observed with IASI over the entire NCP (Fig. 6b–c and e–f). Both CO and O3 increase compared to the previous day. The anticyclone is firmly settled over China, leading to a stagnant situation with low winds all over the NCP (Fig. 4e). This situation is favourable to the accumulation of pollutant and then to the photochemical production of ozone. Figure 7d shows the vertical section of ozone concentrations retrieved with IASI at 37∘ N. The ozone enhancement is located below 4 km, especially between 115 and 116∘ E, in agreement with the findings of Sect. 4.2. This, with CO enhancement, indicates that the ozone enhancement is due to photochemical production from pollutants emitted in the NCP. In this case of stronger CO enhancement, the enhancement ratio of O3 to CO (0.09 on 15 May and 0.06 on 16 May) decreases compared to the previous days.

On 13 and 14 May, large CO and O3 columns are retrieved from IASI over the Yellow Sea and over the Sea of Japan on the southern flank of the cut-off low-pressure system (Figs. 5c and 6a). Fairly strong westerly winds are present at 850 hPa in the same region, suggesting a possible advection of air masses from the NCP towards Japan associated with the weather system (Fig. 4c and d). In order to assess whether the weather system may have contributed to transporting the pollutants (O3 and CO), we perform backtrajectories on 13 May for an area south of Korea (Fig. S2). The 3 km air masses originate from the boundary layer over the NCP on 11 May. They are uplifted and transported at an altitude of between 3 and 4 km on subsequent days (Fig. S2). In order to investigate whether the pollutant uplifting on 11 May occurs over a region more extended than those shown in Fig. S2, we examined two meteorological variables that indicate a possible ascending motion of air masses: the convective available potential energy (CAPE) and the vertical velocity. Figure 8 shows that CAPE is significant on the inside eastern flank of the cut-off low and that negative vertical velocities, i.e., ascending winds, are present from the surface up to 300 hPa (Fig. 8 shows only the vertical velocity at 700 hPa as an example). In addition, backtrajectories performed on 11 May indicate that most of the air masses between 38–40∘ N and 116–117∘ E at 3 km originate from the atmospheric layers below 1 km and circulate over the NCP during the previous 24 h (Fig. S3). This shows that pollutants (CO and O3) have been uplifted from the boundary layer into the free troposphere over the NCP and then exported towards Japan by the cut-off low. This transport pathway is relatively well known. Very recently, Ding et al. (2015) studied the uplifting and transport of CO in east Asia in detail. They show that the vertical transport of anthropogenic CO originating from the NCP is mainly carried out by frontal lifting, which can be, but is not necessarily, associated with WCB. They also pointed out the topography's additional role in the CO lifting over the NCP.

To complete the study, we calculate the enhancement ratio of O3 to CO over the NCP on 12 May, over the Yellow Sea and Korea (32–36∘ N, 122–130∘ E) on 13 May, and over the Sea of Japan (30–38∘ N, 128–140∘ E) on 14 May. The ratios are respectively 0.16, 0.21 and 0.28. The increase in the ratio indicates possible photochemical processing during transport. Part of the large amount of lower-tropospheric ozone is then due to the transport of ozone produced over the NCP but also to ozone produced during the transport.