A nitric acid dataset from IASI for polar stratospheric denitrification studies

In this paper, we exploit the first 10-year data-record (2008-2017) of nitric acid (HNO3) total columns measured by the IASI-A/Metop infrared sounder, characterized by an exceptional daily sampling and a good vertical sensitivity in the midstratosphere (around 50 hPa), to monitor the causal relationship between the temperature decrease and the observed HNO3 loss that occurs each year in the Antarctic stratosphere during the polar night. Since the HNO3 depletion results from the formation of polar stratospheric clouds (PSCs) which trigger the development of the ozone (O3) hole, its continuous monitoring is of high 5 importance. We verify here, from the 10-year time evolution of the pair HNO3-temperature (taken from reanalysis at 50 hPa), the recurrence of specific regimes in the cycle of IASI HNO3 and identify, for each year, the day and the 50 hPa-temperature ("drop temperature") corresponding to the onset of denitrification in Antarctic winter. Although the measured HNO3 total column does not allow differentiating the uptake of HNO3 by different types of PSC particles along the vertical profile, an average drop temperature of ∼191 ± 3 K, consistent with the nitric acid trihydrate (NAT) formation temperature (close to 10 195 K at 50 hPa), is found. The spatial distribution and inter-annual variability of the drop temperature are briefly investigated and discussed in the context of previous PSCs studies. This paper highlights the capability of the IASI sounder to monitor the long-term evolution of the polar stratospheric composition and processes involved in the depletion of stratospheric O3.

below T N AT (∼ 195.7 K at 50 hPa (Hanson and Mauersberger, 1988)) depending on the meteorological conditions (Pitts et al., 2013;Hoyle et al., 2013;Lambert et al., 2016;Pitts et al., 2018), a threshold temperature of 195 K is considered in the sections below to identify the PSCs-containing regions. The potential vorticity is used to delimit dynamically consistent areas in the 90 polar regions. In what follows, we use either the equivalent latitudes ("eqlat", calculated from PV fields at 530 K) or the PV values to characterize the relationship between HNO 3 and temperatures in the cold polar regions. Uncertainties in ERA-Interim temperatures will also be discussed below.
From this figure, different regimes of HNO 3 total columns vs temperature can be observed throughout the year and from one year to another. In particular, we define here three main regimes (R1, R2 and R3) along the HNO 3 cycle. The full cycle and 100 the main regimes in the 70 • − 90 • S eqlat region are further represented in Fig. 1b that shows a histogram of the HNO 3 total columns as a function of temperature for the year 2011. The red vertical line in Fig. 1a and Fig. 1b represent the 195 K threshold temperature used to identify the onset of HNO 3 uptake by PSCs (see Section 2). The three identified regimes correspond to: -R1 is defined by the maxima in the total HNO 3 abundances covering the months of April and May (∼ 3×10 16 molec.cm −2 , R1 in Figure 1a and b), when the 50 hPa temperature strongly decreases (from ∼220 to ∼195 K). These high HNO 3 105 levels result from low sunlight, preventing photodissociation, along with the heterogeneous hydrolysis of N 2 O 5 Santee et al. (1999); Urban et al. (2009);de Zafra and Smyshlyaev (2001).
-R2 which extents from June to September is characterized by the onset of the strong decrease in HNO 3 total columns at the beginning of June, when the temperatures fall below 195 K, followed by a plateau of total HNO 3 minima. In this regime, the HNO 3 total columns average below 2 × 10 16 molec.cm −2 and the 50 hPa temperatures range mostly between 180 110 and 190 K.
-R3 starts in October when sunlight returns and the 50 hPa temperatures rise above 195 K. Despite the stratospheric warming with 50 hPa temperatures up to 240 K in summer, the HNO 3 total columns stagnate at the R2 plateau levels (around 1.5× 10 16 molec.cm −2 ). This regime likely reflects the photolysis of NO 3 and HNO 3 itself (Ronsmans et al., 2018) as well as the permanent denitrification of the mid-stratosphere, caused by the PSCs sedimentation, despite the renitrification of the 115 lowermost stratosphere (Braun et al., 2019) where the IASI sensitivity to HNO 3 is lower (Ronsmans et al., 2016). The   and temperature in the mid-stratosphere. The results (not shown here) exhibit a similar HNO 3 -temperature behavior at the 135 different levels with, as expected, lower and larger temperatures in R2, respectively, at 30 hPa (180 and 185 K) and at 70 hPa (∼190 K), but still below the NAT formation threshold at these pressure levels (T N AT ∼193 K at 30 hPa and ∼197 K at 70 hPa) (Lambert et al., 2016). Therefore, the altitude range of maximum IASI sensitivity to HNO 3 (see Section 2) is characterized by the PSCs formation and the denitrification process. Furthermore, the consistency between the 195 K threshold temperature taken at 50 hPa and the onset of the strong total HNO 3 depletion seen in IASI data (see Fig. 1a and Fig. 1c) is in agreement 140 with the largest NAT area that starts to develop in June around 20 km (Spang et al., 2018), which justifies the use of the 195 K temperature at that single representative level in this study. Despite the limited vertical resolution of IASI which does not allow to investigate the HNO 3 uptake by the different types of PSCs during their formation and growth along the vertical profile, the HNO 3 total column measurements from IASI constitute an important new dataset for exploring the polar denitrification over the whole stratosphere. This is particularly relevant considering the mission continuity, which will span several decades with 145 the planned follow-on missions.

Onset of HNO 3 depletion and drop temperature detection
To go beyond the vertically integrated view of denitrification and to identify its spatial and temporal variability, the daily time evolution of HNO 3 during the first 10 years of IASI measurements and the temperatures at 50 hPa are explored. In particular, the second derivative of HNO 3 total column with respect to time is calculated to detect the strongest rate of decrease seen in J a n 0 8 0 6 J u n 0 8 J a n 0 9 1 0 J u n 0 9 J a n 1 0 1 7 M a y 1 0 J a n 1 1 2 0 M a y 1 1 J a n 1 2 0 8 J u n 1 2 J a n 1 3 1 7 M a y 1 3 J a n 1 4 2 3 M a y 1 4 J a n 1 5 2 0 M a y 1 5 J a n 1 6 0 4 J u n 1 6 J a n 1 7 1 7 M a y 1 7 Note that the HNO 3 time series has been smoothed with a simple spline data interpolation function to avoid gaps before calculating the second derivative of HNO 3 total column with respect to time as the daily second-difference HNO 3 total column.

160
The horizontal red line shows the 195 K threshold. As already illustrated in Fig. 1a and Fig. 1c, the strongest rate of HNO 3 depletion (i.e. the second derivative minimum) is found around the 195 K threshold temperature, within a few days (4 to 23 days) after total HNO 3 reaches its maximum, i.e. typically between the 17 th of May (  in HNO 3 depletion occurs on average a few days before June. The delay of 4-23 days between the maximum in total HNO 3 and the start of the depletion (see Fig. 2) is also visible in Fig. 3a. The yearly zonally averaged time series over the ten years of IASI can be found in Fig. 4; it shows the reproducibility in the HNO 3 depletion measured from IASI from year to year.

Distribution of drop temperatures
To explore the capability of IASI to monitor the onset of denitrification at a large scale from year to year, Figure 5 shows the 180 spatial variability of the drop 50 hPa temperatures (based on the second derivative minima of total HNO 3 averaged in 1 • × 1 • grid cells) inside a PV region of ≤ −8× 10 −5 K.m 2 .kg −1 .s −1 , for each year of the IASI period. The red contour represents the PV isocontour of ≤ −10× 10 −5 K.m 2 .kg −1 .s −1 that delimits our region of interest. The dates corresponding to the onset of HNO 3 depletion inside that region are found to range between mid-May and early-July (not shown here). The calculated drop temperatures vary significantly between ∼ 180 and ∼ 210 K. These high extremes are only found in very few cases and should 185 be considered with caution as they correspond to specific regions above ice shelves with emissivity features that are known to yield errors in the IASI retrievals (Hurtmans et al., 2012;Ronsmans et al., 2016). Note also that these spatial variations might partly reflect the range of maximum sensitivity of IASI to HNO 3 (hence, the use of temperature at a single pressure level might be restrictive to some extent) and biases in ECMWF reanalysis. Reanalysis data sets are, indeed, known to feature large uncertainties. In particular, they do not always capture small-scale fluctuations due to the limited spatial resolution, especially in and stratospheric temperatures at 50 hPa, taken from the ECMWF ERA Interim reanalysis. That single representative pressure level has been considered in this study given the maximum sensitivity of IASI to HNO 3 around that level over a range where the PSCs formation/denitrification process occur.
The annual cycle of total HNO 3 , as observed from IASI, has first been characterized according to the temperature evolution. K (191±3 K on average over the 10 years), which demonstrated the good consistency between the 50 hPa drop temperature and the PSCs formation temperatures in that altitude region. Finally, the annual and spatial variability (within 1 • × 1 • ) in the drop temperature was further explored from IASI total HNO 3 and shown to range between ∼180 and ∼210 K. While recurrent patterns of extreme high drop temperatures were found from year to year and suspected to result from unfiltered poor quality retrievals in case of emissivity issues above ice, the range of drop temperatures is interestingly found in line with the PSCs 225 nucleation temperature that is known, from previous studies, to strongly depend on a series a factors (e.g. meteorological conditions, HNO 3 vapour pressure, temperature threshold exposure, presence of meteoritic dust). The results of this study highlighted the ability of IASI to measure the variations in total HNO 3 and, in particular, to capture and monitor the rapid denitrification phase over the whole polar regions.
To the best of our knowledge, it is the first time that such a large satellite observational data set of stratospheric HNO 3 230 concentrations is exploited to monitor the evolution HNO 3 versus temperatures. Thanks to the three successive instruments (IASI-A launched in 2006 and still operating, IASI-B in 2012, and IASI-C in 2018) that demonstrate an excellent stability of the Level-1 radiances, the measurements will soon provide an unprecedented long-term dataset of HNO 3 total columns. It could constitute a new accurate climatological parameter that could be inserted in the PSCs classification schemes. Further work