Validation of stratospheric nitric acid profiles observed by Improved Limb Atmospheric Spectrometer (ILAS)–II

Authors


Abstract

[1] The Improved Limb Atmospheric Spectrometer-II (ILAS-II) was launched aboard the Advanced Earth Observing Satellite-II (ADEOS-II) in December 2002. Stratospheric vertical profiles of nitric acid (HNO3) concentration observed by ILAS-II (version 1.4) are validated using coincident HNO3 measurements by balloon-borne instruments (MIPAS-B2 and MkIV) in March and April 2003. Further validation is performed by making climatological comparisons of lower stratospheric HNO3-ozone (O3) correlations obtained by ILAS-II and ILAS (the predecessor of ILAS-II) for specific potential vorticity-based equivalent latitudes and seasons where and when ILAS data showed very compact correlations in 1997. The reduced scatter of ILAS-II HNO3 values around the reference HNO3, which is derived from ILAS-II O3 using the ILAS HNO3-O3 correlation, shows that the precision of the ILAS-II HNO3 data is better than 13–14%, 5%, and 1% at 15, 20, and 25 km, respectively. Combining all of the comparisons made in the present study, the accuracy of the ILAS-II HNO3 profiles at 15–25 km is estimated to be better than −13%/+26%.

1. Introduction

[2] Polar stratospheric clouds (PSCs) play a central role in stratospheric ozone depletion over the Antarctic and Arctic in the winter/spring seasons. Nitric acid (HNO3) is a key component determining the chemical and physical properties of PSCs that form at low temperatures (<∼200 K) through condensation of HNO3 and water vapor (H2O) in the winter/spring polar stratosphere. Satellite measurements of HNO3 concentrations have been widely used to investigate not only spatial distributions and temporal variations of HNO3 [Santee et al., 1995, 1997, 1998, 1999, 2000, 2004; Irie et al., 2001] but also important details of the processes controlling stratospheric HNO3 concentrations, e.g., the formation of PSC particles and denitrification [Santee et al., 1998, 2002; Hayashida et al., 2000; Kondo et al., 2000a; Irie et al., 2001, 2004; Saitoh et al., 2002; Irie and Kondo, 2003; Rivière et al., 2003]. While these processes are not fully understood, the Improved Limb Atmospheric Spectrometer-II (ILAS-II) onboard the Advanced Earth Observing Satellite-II (ADEOS-II) provided recent measurements of vertical profiles of HNO3 concentration at high latitudes in both the Northern Hemisphere (NH) and the Southern Hemisphere (SH). Preoperational and operational observations by ILAS-II were performed in January–March and April–October 2003, respectively. In this paper, we assess the validity of the ILAS-II HNO3 profiles retrieved with the version 1.4 algorithm under PSC-free conditions by using comparisons with HNO3 profiles measured by balloon-borne instruments. Further assessment is performed from a climatological standpoint by comparing the lower stratospheric correlations between HNO3 and ozone (O3) mixing ratios obtained by ILAS-II in 2003 with those by ILAS (the predecessor of ILAS-II) in 1997. We choose O3 as a companion species, because it can generally be validated more precisely than other species because of extensive ozonesonde and satellite O3 data. We assume that HNO3-O3 correlations between 1997 and 2003 were invariant for the same latitudes and seasons. This assumption is supported from the multiyear data obtained by the Microwave Limb Sounder (MLS) onboard the Upper Atmosphere Research Satellite (UARS).

2. ILAS-II Measurements

2.1. Instrument

[3] The ILAS-II instrument, designed similarly to its predecessor ILAS [Sasano et al., 1999], is a solar occultation sensor to measure stratospheric vertical profiles of various constituents, including HNO3 and O3. The ILAS-II instrument consists of infrared (channel 1, 6.2–11.8 μm), midinfrared (channel 2, 3.0–5.7 μm), narrow-band (channel 3, 12.78–12.85 μm), and visible (753–784 nm) spectrometers, and a sun edge sensor. The ILAS-II was launched aboard the ADEOS-II satellite on 14 December 2002, and performed about 400 preoperational observations between January and March 2003. The operational observations were made with a frequency of about 14 times per day in each hemisphere for about 7 months from 2 April through 24 October 2003. The measurement latitudes ranged from 54° to 71°N and from 65° to 88°S, varying seasonally with the geometric relationship between the Sun, the Earth, and the satellite.

2.2. Version 1.4 Retrieval Algorithm

[4] The version 1.4 retrieval algorithm was used to obtain gas concentration profiles from the absorption lines measured with the channel-1 infrared spectrometer (T. Yokota et al., unpublished manuscript, 2006). The algorithm primarily used the strong absorption lines around 7.6 and 11.3 μm for the HNO3 retrieval and around 9.5–10.0 μm for O3. Spectroscopic data were adopted from the year 2000 edition of the High-Resolution Transmission (HITRAN) database, including updates through the end of 2001 [Rothman et al., 2003]. The HNO3 and O3 concentrations were retrieved simultaneously with other gases using a nonlinear least squares method [Yokota et al., 2002], after estimating the contribution of aerosol attenuation at 7.12, 8.27, 10.60, and 11.76 μm, where contributions of gas absorption are relatively much smaller as compared to the other wavelength regions. These four spectral elements are called “window elements.” The extinction coefficients of aerosols at the window elements were linearly interpolated and extrapolated with wave number to obtain transmittance due only to gaseous absorption over the whole range of infrared regions measured. This simple linear approximation may cause systematic errors in the HNO3 and O3 concentrations retrieved, as described below. Vertical profiles of concentration were retrieved by the so-called onion-peeling method [Yokota et al., 2002], with vertical resolutions of 1.3–2.9 km at tangent heights of 15–55 km. The tangent heights of the measurements were determined with a sun edge sensor method and a transmittance spectrum method for altitudes above and below 30 km, respectively (for more details see Nakajima et al. [2002], T. Tanaka et al. (unpublished manuscript, 2006), and http://www-ilas2.nies.go.jp/en/).

2.3. Errors

[5] To evaluate the precision of the ILAS-II concentration data, for every series of 100 occultation measurements (corresponding to about an 8-day time period) we calculated an average and a 1σ standard deviation of the retrieved concentration data at each 1-km altitude for each hemisphere. The minimum of the ratio (1σ standard deviation)/(average) for the whole of the ILAS-II measurement period gives an upper limit of the precision, because it should represent the precision of the data with the smallest contributions from the natural variability in the measurement period. The minimum value is referred to as the “repeatability error” for all of the ILAS-II version 1.4 data products. For HNO3 and O3, the repeatability errors are summarized in Table 1, showing that the estimated upper limits of the precision at 15–25 km are 1–23% and 2–14%, respectively. In addition to the repeatability error, the retrieved concentration data can further be affected by uncertainty in the gas concentrations assumed in the calculation of the aerosol attenuation contribution to the spectra at window elements, as mentioned in section 2.2, and by the uncertainty in temperature [Yokota et al., 2002]. The additional error due to these uncertainties is referred to as the “external error.” The median external errors calculated from all the ILAS-II data are listed in Table 2. In comparisons of HNO3 profiles obtained by ILAS-II and balloon-borne instruments below, we use the term “total error,” which is defined as the root-sum-square of the repeatability and external errors.

Table 1. Repeatability Errors (%) for HNO3 and O3
SpeciesHemisphere15 km20 km25 km
HNO3NH2351
HNO3SH1753
O3NH1446
O3SH1152
Table 2. External Errorsa (%) for HNO3 and O3
SpeciesHemisphere15 km20 km25 km
  • a

    Median values calculated from all the ILAS-II data.

HNO3NH1032
HNO3SH933
O3NH942
O3SH1051

[6] It should be noted that in the version 1.4 retrieval algorithm, the uncertainty in line parameters has not been considered in the calculation of the total error. While this omission may lead to an underestimate of the total error, the present paper empirically estimates the precision, which should contain all of the random error components, including random errors in line parameters. If the line parameters used contain systematic errors, they would be taken into account in future versions of the ILAS-II retrieval algorithm by using updated line parameter data.

[7] In the present paper, we will attempt to estimate the accuracy of ILAS-II HNO3 profiles. For O3, Sugita et al. [2006] have made statistically robust comparisons with a large amount of ozonesonde and satellite data (SAGE II/III, HALOE, and POAM III). The comparisons showed that most ILAS-II O3 profiles agreed with ozonesonde and satellite data to within 10% at 10–40 km.

[8] As described above, aerosol extinction at all the wavelengths measured with the channel-1 infrared spectrometer was simply estimated using a linear interpolation and extrapolation from the aerosol absorption at four window wavelengths. This simple technique sometimes yields a bias in the HNO3 and O3 concentration retrievals, depending on composition and volume of aerosols/PSCs that are present in air masses observed. Similar to the work of Yokota et al. [2002] for ILAS, we performed a theoretical simulation to quantify the magnitude of the HNO3 and O3 biases as a function of aerosol extinction coefficient (AEC) at 780 nm for several PSC compositions. As in the ILAS results, the biases of the ILAS-II HNO3 and O3 concentrations were both well correlated with AEC in the simulations. Table 3 summarizes PSC composition and size distribution parameters assumed in the simulations and the estimated biases of the HNO3 and O3 concentrations for AEC = 10−3 km−1. AEC values less than 10−3 km−1 are considered to include typical PSC conditions, because about 95% of all the ILAS-II AEC values at 20 km over the Antarctic were less than 10−3 km−1 for the period between April (late fall) and October (early spring) 2003. For AEC values less than 10−3 km−1, the largest positive bias was possible in the case of ice PSCs, but was less than 2.4 × 108 cm−3 (corresponding to about 0.14 ppbv HNO3 at 20 km) for HNO3 and less than 1.2 × 1012 cm−3 (about 0.74 ppmv O3 at 20 km) for O3. The negative bias in HNO3 concentration could reach 16.0 × 108 cm−3 (about 0.96 ppbv HNO3 at 20 km) in the case of nitric acid trihydrate (NAT), whereas no negative bias in O3 due to the presence of PSCs is predicted in any of the cases considered here.

Table 3. Systematic Biases of HNO3 and O3 Caused by Optically Thick PSCs (AEC = 10−3 km−1) at 20 km
 ICENATSTSa (a)STSa (b)STSa (c)STSa (d)
  • a

    Supercooled ternary solution of HNO3/H2SO4/H2O.

  • b

    Mode radius.

  • c

    Standard deviation of a lognormal size distribution.

  • d

    Mixing ratios were calculated assuming the typical air number density at 20 km [Yokota et al., 2002; T. Yokota et al., unpublished manuscript, 2006].

r,b μm100.50.50.50.50.5
σc1.51.51.51.51.51.5
HNO3, wt%371530.5
H2SO4, wt%5334760
 
HNO3 bias, 108 cm−32.4−16.0−0.091.22.02.0
HNO3 bias,d ppbv0.14−0.96−0.0050.070.120.12
O3 bias, 1012 cm−31.20.150.0050.230.250.22
O3 bias,d ppmv0.740.090.0030.140.150.13

3. Comparison With Balloon-Borne Measurements

3.1. Approach

[9] The ILAS-II HNO3 vertical profiles are first compared with those measured by two balloon-borne instruments: (1) the Michelson Interferometer for Passive Atmospheric Sounding-Balloon-borne version 2 (MIPAS-B2) instrument [Oelhaf et al., 1996; Friedl-Vallon et al., 2004], which is a cryogenic Fourier transform infrared (FTIR) spectrometer measuring thermal emission from the limb of the atmosphere, and (2) the MkIV instrument [Toon, 1991; Sen et al., 1998], which is a solar occultation FTIR spectrometer. For both MIPAS-B2 and MkIV, the HNO3 profile used in the present study was retrieved using the HITRAN2000 spectroscopic database as used in the ILAS-II version 1.4 algorithm. The accuracy and precision of these balloon-borne HNO3 measurements at 20 km are about 10–12% and 5%, respectively (Table 4). Vertical resolutions estimated from the instantaneous fields of view of the instruments are about 1.2 and 2.0 km for MIPAS-B2 and MkIV, respectively. As used in previous ILAS HNO3 validation studies by Koike et al. [2000] and Irie et al. [2002], these measurement characteristics are sufficient to validate HNO3 measurements made by satellite-borne solar occultation sensors.

Table 4. List of Balloon Measurements
InstrumentDateAccuracy,a %Precision,a %Distance,b kmTime Difference,b hourPV Difference,c %
  • a

    One-σ values at 20 km.

  • b

    Values between the ILAS-II and balloon measurement locations at 20 km.

  • c

    Values between the ILAS-II and balloon measurement locations at the same potential temperature level near 20 km.

  • d

    Values between the ILAS-II and MkIV air masses using trajectories at 20 km.

MIPAS-B220 Mar.1051045.84
MkIV1 Apr.1252500 (518–585)d33.6 (<1.0)d1

[10] The measurements by MIPAS-B2 and MkIV instruments were performed near Kiruna, Sweden (67.9°N, 21.1°E), during the night on 20 March 2003, and at sunrise on 1 April 2003, respectively. On 20 March 2003, ILAS-II measurements were made very close to the locations of the MIPAS-B2 measurements; the distance between the ILAS-II and MIPAS-B2 measurement locations at an altitude of 20 km was only about 100 km (Table 4). The time difference was only 6 hours at 20 km (Table 4). In addition, the potential vorticity (PV) difference, based on United Kingdom Met Office (UKMO) data, was only 4% at the potential temperature corresponding to the 20-km altitude of the MIPAS-B2 measurements. On the other hand, no ILAS-II measurement was made on 1 April 2003, when the MkIV measurement was made. However, forward isentropic trajectories calculated from the time and location of the MkIV measurements using UKMO wind fields (Figure 1) encountered an ILAS-II measurement on 2 April, within a 600-km distance and a 1-hour time difference from the MkIV air mass trajectory at 20 km (Table 4). For the potential temperature level corresponding to the 20-km altitude of the MkIV measurements, the PV difference between the ILAS-II and MkIV measurement times and locations was only 1% (Table 4). These meteorological conditions indicate that the HNO3 profiles measured by MIPAS-B2 and MkIV should be appropriate for validating the corresponding ILAS-II HNO3 profiles. To make precise comparisons with MIPAS-B2 and MkIV measurements, the ILAS-II HNO3 values are linearly interpolated to the potential temperature levels corresponding to the balloon measurement altitudes. In addition, we assume that O3 is a long-lived tracer in order to carefully search pairs of air masses having the same HNO3 concentrations. This assumption is valid because little or no chemical ozone loss likely occurred during the short time periods between the ILAS-II and balloon measurements (<∼34 hours (Table 4)). Detailed comparisons are then made only for altitudes where ILAS-II O3 concentrations agree with those of balloon measurements to within the combined error ranges for ILAS-II and the balloon data.

Figure 1.

Locations of the MkIV and ILAS-II measurements (stars). These measurements were made on 1 and 2 April 2003, respectively. Two-day isentropic forward trajectories from the MkIV measurement locations at 15, 20, and 25 km are shown with different curves. Red symbols indicate the locations of the trajectories at times within 1 hour of the ILAS-II measurements.

3.2. Results and Discussion

[11] In Figures 2 and 3, the HNO3 profiles measured by ILAS-II are compared with those measured by MIPAS-B2 and MkIV, respectively. For each balloon HNO3 profile, one coincident ILAS-II profile is shown. In Figures 2 and 3, the dotted curves represent the total errors of the ILAS-II HNO3 data. The horizontal bars show the total errors in HNO3 (random + systematic errors) for the balloon measurements. As seen in Figure 2a, the HNO3 mixing ratios derived from ILAS-II measurements agree with the MIPAS-B2 values to within their combined errors at 15–22 and 25–26 km. The relative differences in HNO3 (Figure 2b), defined as ((ILAS-II data) – (other data))/(other data), are shown with solid symbols for 15–21, 24–28, and 31 km where the difference between O3 mixing ratios obtained by ILAS-II and MIPAS-B2 was within their combined error ranges. Thirteen comparison pairs selected using O3 in this way show that the relative differences in HNO3 range from 0% to +14% at 15–21 km and from +9% to +35% at 24–28 and 31 km (Figure 2b).

Figure 2.

(a) Vertical profiles of HNO3 mixing ratio measured by ILAS-II and MIPAS-B2 on 20 March 2003 and (b) the relative differences ((ILAS-II value) – (MIPAS-B2 value))/(MIPAS-B2 value). The dotted curves in Figure 2a represent the total errors of the ILAS-II HNO3 measurements. Total errors of the balloon measurements are shown with error bars. These combined errors are shown with two thin curves in Figure 2b. The relative differences are shown with solid and open symbols when the differences in O3 between ILAS-II and MIPAS-B2 are within and out of their combined error ranges, respectively.

Figure 3.

(a and b) Same as Figure 2 but for the comparison of HNO3 vertical profiles measured by ILAS-II and MkIV on 2 and 1 April 2003, respectively.

[12] For the comparison with MkIV measurements, the ILAS-II HNO3 mixing ratios agreed with those of MkIV to within the combined errors at 12–13, 15–18, 20–21, and 28–31 km (Figure 3a). Below the altitude of peak HNO3 mixing ratio, the stratospheric HNO3 and O3 mixing ratios generally show a positive correlation, similar to NOy-O3 correlations [Murphy et al., 1993; Fahey et al., 1996; Michelsen et al., 1998]. For altitudes below 22 km, where MkIV HNO3 values are at their maximum, the O3 mixing ratios derived from the ILAS-II measurements were systematically greater than MkIV O3 data by 0.3–0.5 ppmv at 14–16 km and lower by 0.3–0.5 ppbv at 19–22 km. These differences in O3 exceeded the combined O3 error ranges for the ILAS-II and MkIV measurements. Correspondingly, the ILAS-II HNO3 mixing ratios are greater than the MkIV data by 0.8–1.2 ppbv at 14–16 km and lower by 0.4–1.2 ppbv at 19–22 km (Figure 3a), indicating that O3 is useful for optimizing HNO3 comparison conditions in the lower stratosphere. Such significant differences in O3 were found at 14–16 and 19–33 km, and thus the comparisons for these altitudes are excluded from the following analyses, resulting in only 4 coincident pairs of ILAS-II/MkIV air masses. For these coincident pairs, found at altitudes of 12–13 and 17–18 km, where no significant differences in O3 occurred, the relative differences in HNO3 ranged between −17% and +13%, as shown with solid symbols in Figure 3b.

[13] We note here that in the present study the vertical profiles of HNO3 mixing ratio derived from ILAS-II observations have been compared with only two balloon HNO3 profiles. It is unlikely that the calculated differences give a quantitative estimate of the accuracy of the ILAS-II HNO3 measurements, because the number of balloon data available is small and the calculated differences may have uncertainties of at least 15–17%, due to the total errors of the balloon measurements used (Table 4). Therefore we make additional estimates from the standpoint of climatology below, using the HNO3-O3 correlations.

4. Climatological Comparison

4.1. Approach

[14] We next compare the ILAS-II HNO3 data obtained in 2003 with those obtained by ILAS (version 6) in 1997 to further evaluate the ILAS-II HNO3 data from the standpoint of climatology. As shown in Figures 4 and 5, both the ILAS-II and ILAS observations were made between April and June and covered very similar geographic latitudes and equivalent latitudes (ELs). EL is defined as the geographic latitude enclosing the area where PV values are greater than or equal to a given PV value [Butchart and Remsberg, 1986]. Vertical profiles of HNO3 in 2003 may be different from those in 1997, especially for the seasons when diabatic descent of stratospheric air generally occurs. To minimize the effect of the year-to-year variations in HNO3 profiles on the climatological ILAS-II/ILAS comparison, we use O3 as a long-lived tracer for the specific latitudes and seasons where and when the ILAS data show very compact HNO3-O3 correlations in 1997. The climatological comparison method using O3 is described in detail below.

Figure 4.

Locations of the measurements made by ILAS and ILAS-II at 20 km for the Northern Hemisphere. (a) Measurement latitudes for ILAS and ILAS-II are shown in black and red, respectively. Equivalent latitudes for (b) ILAS and (c) ILAS-II. The measurements were categorized into regions inside the vortex (red), in the boundary region (green), and outside the vortex (blue), using the method of Nash et al. [1996]. The equivalent latitudes are shown in black only for the periods when the polar vortex was much weaker relative to the winter/spring season.

Figure 5.

(a–c) Same as Figure 4 but for the Southern Hemisphere.

[15] It is well recognized that HNO3 is the predominant component of the total reactive nitrogen (NOy) in the mid- and high-latitude lower stratosphere (at ∼10–25 km) [Kawa et al., 1992; Sen et al., 1998; Kondo et al., 2000b; Wetzel et al., 2002]. The HNO3/NOy ratio varies depending primarily on altitude, season, and aerosol loading [Mills et al., 1993; Koike et al., 1994; Sen et al., 1998]. On the other hand, because of intense production of NOy and O3 in the upper tropical stratosphere and their longer lifetimes relative to horizontal mixing [Plumb and Ko, 1992], the correlation between NOy and O3 mixing ratios are usually compact throughout the lower stratosphere, even at mid and high latitudes [Murphy et al., 1993; Fahey et al., 1996; Michelsen et al., 1998], except inside the polar vortex where severe chemical ozone depletion and denitrification occasionally occur. The observations of these NOy-O3 correlations have revealed a slight seasonal dependence in the slope of the correlations, with a large latitudinal dependence, especially in terms of the tropical, midlatitude, and polar vortex regimes. It can thus be expected that the mid- and high-latitude HNO3-O3 correlations at altitudes of 10–25 km would usually be tight and the slope of the correlations would be dependent not only on season and aerosol loading but also on EL.

[16] In Figures 4 and 5, the ELs of the ILAS and ILAS-II measurements at 20 km are shown in black only for the periods when the polar vortex was much weaker relative to the winter/spring season. For these periods, the daily maximum of the value (dPV/dEL) × (average wind speed along PV isolines) was less than 30% of the wintertime maximum at potential temperature levels corresponding to about 20 km. Considering the typical seasonal evolution of the polar vortex, the Arctic and Antarctic vortex strengths were least in June and April, respectively, in the months when ILAS-II and ILAS measurements overlapped (April–June). The polar vortex might no longer exist during the overlap months, especially over the Arctic in June. Under these meteorological conditions, little or no chemical ozone depletion or denitrification likely occurred over the Arctic in June or over the Antarctic in April. For June and April, we focus on the HNO3-O3 correlations obtained at ELs of 75° ± 5°N and 60° ± 5°S, respectively, because ILAS observed the tightest HNO3-O3 correlations at these ELs in 1997. The 1σ HNO3 standard deviations of the ILAS correlations for the above ELs and seasons were less than ∼0.7 ppbv for every 0.5-ppmv O3 range. The value of 0.7 ppbv is comparable to the precision of the ILAS HNO3 measurements (∼0.8 ppbv [Koike et al., 2000; Irie et al., 2002]).

[17] For ELs of 75° ± 5°N in June and 60° ± 5°S in April, the only factor producing a difference between HNO3-O3 correlations in 1997 and 2003 should be aerosol loading, because the correlation depends primarily on season, EL, and aerosol loading, as discussed above. The stratospheric aerosol loading has reached very low levels from the late 1990s to the present because of low volcanic activity, according to the integrated backscatter from the Fraunhofer Institute for Atmospheric Environmental Research lidar and Stratospheric Aerosol and Gas Experiment (SAGE) II measurements [World Meteorological Organization (WMO), 2003]. Thus the lower stratospheric HNO3-O3 correlations at ELs of 75° ± 5°N in June 1997 and at ELs of 60° ± 5°S in April 1997 should most likely be similar to those for 2003, and therefore the HNO3 measurements made by ILAS-II can be evaluated by comparing the HNO3-O3 correlations obtained by ILAS-II and ILAS for these EL ranges and seasons. To confirm that the HNO3-O3 correlations at those ELs and seasons are invariant over a timescale of years, we also analyze the UARS/MLS HNO3 (version 6) and O3 data (version 5) [Santee et al., 2004]. To perform a quantitative evaluation as a function of altitude, we compare HNO3 mixing ratios measured by ILAS-II with the reference HNO3 (referred to as HNO*3) values, which are calculated from the O3 mixing ratios measured by ILAS-II using the tight correlations between HNO3 and O3 observed by ILAS. All of the mixing ratios required in this calculation and the following analyses have been scaled according to their estimated bias (for NH at 20 km, for example, +2% for ILAS HNO3 [Irie et al., 2002], +3% for ILAS O3 [Sugita et al., 2002], and −2% for ILAS-II O3 [Sugita et al., 2006].

4.2. Results and Discussion

[18] In Figure 6a, mean vertical profiles of HNO3 and O3 mixing ratios observed by ILAS-II at ELs of 75° ± 5°N in June 2003 are compared with those observed by ILAS in 1997. For all the altitudes between 12 and 25 km, ILAS-II O3 mixing ratios agree with those of ILAS to within their standard deviations. However, for ELs of 60° ± 5°S in April, when diabatic descent of stratospheric air generally occurs especially at higher altitudes, the comparison between O3 profiles observed by ILAS-II and ILAS show larger differences at higher altitudes (Figure 6b). The differences are presumably due to a difference in diabatic descent between 2002 and 1997. To reduce such an effect on the comparison of HNO3 profiles measured by ILAS-II and ILAS, we use HNO3-O3 correlations below.

Figure 6.

Mean vertical profiles of HNO3 and O3 obtained (a) at ELs = 75° ± 5°N in June and (b) at ELs = 60° ± 5°S in April. HNO3 and O3 profiles are indicated by lines without and with symbols, respectively. Solid and dashed lines indicate data obtained by ILAS-II in 2003 and by ILAS in 1997, respectively. One-σ standard deviations are shown with error bars.

[19] For ELs of 75° ± 5°N in June 1997 and ELs of 60° ± 5°S in April 1997, the HNO3-O3 correlations obtained by ILAS are shown with open symbols in Figures 7a and 7b, respectively. The mean and 1σ standard deviations of ILAS HNO3 values in each 0.5-ppmv O3 range are shown. The numbers of data points used for each 0.5-ppmv O3 range are 22–62 and 35–147 for ELs of 75° ± 5°N and 60° ± 5°S, respectively. The ILAS HNO3 values plotted in Figures 7a and 7b are the predefined HNO*3 values as a function of O3 mixing ratio.

Figure 7.

HNO3-O3 correlations obtained by ILAS (open symbols) and MLS (solid symbols) at (a) ELs = 75° ± 5°N and (b) ELs = 60° ± 5°S. Time periods during which the data were obtained are described. For ILAS, the averages of the HNO3 mixing ratios for each 0.5-ppmv O3 range are plotted. One-σ standard deviations are shown with error bars. For MLS, correlations between daily averaged HNO3 and O3 mixing ratios measured at each of four potential temperature levels between 420 K (∼17 km) and 585 K (∼24 km) are shown. Measurements made in 1995, 1996, and 1997 are shown in green, blue, and red, respectively. One-σ standard deviations of HNO3 and O3 data are shown only for 1997, for clarity.

[20] Long-term, simultaneous measurements of stratospheric HNO3 and O3 mixing ratios were made by UARS/MLS from 1991 through 2000. In Figures 7a and 7b, the correlations between the daily averages of the HNO3 and O3 mixing ratios derived from the MLS observations are also shown for ELs of 75° ± 5°N in May and July and for 60° ± 5°S in April, respectively. Averages of the MLS data were calculated for each of four different potential temperature levels, namely 420 K (∼17 km), 465 K (∼19 km), 525 K (∼22 km), and 585 K (∼24 km). It should be noted that for ELs of 75° ± 5°N we plotted the MLS data for May and July, instead of June, because MLS seldom or never made measurements in the NH in June [Santee et al., 2004]. We also note that Figures 7a and 7b show the HNO3-O3 correlations only for 1995–1997, because variations in stratospheric aerosol loading [WMO, 2003] could affect the correlations in 1991–1994, and the MLS measurements in 1998–2000 were much more sparse relative to 1995–1997 [Santee et al., 2004].

[21] Considering that the precision of the individual HNO3 profile measurements by MLS is as good as 1.0–1.5 ppbv [Santee et al., 2004], larger error bars on the MLS HNO3-O3 correlations relative to those for ILAS (Figure 7) can be attributed to the poorer vertical resolution of the MLS measurements (5–10 km) compared to that of the ILAS measurements (1.9–3.5 km), because HNO3 and O3 mixing ratios vary with altitude. As seen in Figure 7, the differences in MLS HNO3 mixing ratios among the different years 1995–1997 are all less than 1.5 ppbv at the same O3 mixing ratio. This indicates that the lower stratospheric HNO3-O3 correlations were almost invariant over years 1995–1997 for the above ELs and seasons. The results support the assumption made in the present study that the lower stratospheric HNO3-O3 correlations were invariant for these ELs and seasons over several years. Furthermore, the ILAS HNO3-O3 correlations agree well with those of MLS in 1997 for these ELs and seasons, confirming that the ILAS correlations are suitable for making climatological comparisons with ILAS-II HNO3 data.

[22] In Figures 8a and 8b, the HNO3-O3 correlations obtained by ILAS-II (solid symbols) are plotted together with those of ILAS (open symbols), for ELs of 75° ± 5°N in June and for ELs of 60° ± 5°S in April, respectively. It can be readily seen from Figures 8a and 8b that the HNO3-O3 correlations measured by ILAS-II were compact, as well as those measured by ILAS. However, the ILAS-II correlations systematically depart from ILAS correlations, especially at higher altitudes for ELs of 60° ± 5°S in April (Figure 8b). Similar systematic differences are also seen from the comparison with MLS data; while the daily averages of HNO3 mixing ratios obtained by MLS in 1995–1997 were all less than ∼10 ppbv (Figure 7b), the ILAS-II HNO3 values reached ∼11.5 ppbv (Figure 8b) and were systematically greater than the MLS values when we consider the precision of the MLS measurements (1.0–1.5 ppbv).

Figure 8.

HNO3-O3 correlations obtained by ILAS-II (a) at ELs = 75° ± 5°N in June 2003 and (b) at ELs = 60° ± 5°S in April 2003. The correlations are shown with solid symbols. Colors indicate the altitudes of the measurements. For comparison, the correlations obtained by ILAS in 1997 at the same ELs and seasons are also shown with open symbols.

[23] To make quantitative estimates of the difference as a function of altitude, the mean HNO3 vertical profile measured by ILAS-II at ELs of 75° ± 5°N in June is compared with the mean profile of HNO*3 values, which were calculated from ILAS HNO3-O3 correlations and ILAS-II O3 data. In Figure 9a, the dotted curves and error bars represent the 1σ standard deviations of the ILAS-II HNO3 and HNO*3 values, respectively. The mean relative differences between ILAS-II HNO3 and HNO*3 values ranged from +9 to +26% at 13–25 km (Figure 9b). Similarly, in the comparison at ELs of 60° ± 5°S in April, ILAS-II HNO3 values are systematically greater than HNO*3 by 17–24% at 15–24 km (Figure 10b). These results suggest that the ILAS-II HNO3 data could have some positive biases at 15–25 km altitude, although the exact magnitude of the bias cannot be quantified only from these comparisons because the results are fully based on climatological comparisons.

Figure 9.

(a) Vertical profiles of ILAS-II HNO3 values (line) and reference HNO3 (HNO*3) values (symbols) for ELs = 75° ± 5°N in June. The associated dotted curves and error bars represent 1σ standard deviations for ILAS-II HNO3 and HNO*3, respectively. (b) Vertical profile of the mean relative differences between the ILAS-II HNO3 and HNO*3 values is shown. Error bars represent 1σ standard deviations.

Figure 10.

(a and b) Same as Figure 9 but for ELs = 60° ± 5°S in April.

[24] The standard deviations of the relative differences between the ILAS-II HNO3 and HNO*3 values are shown with error bars in Figures 9b and 10b. These standard deviations are the result of the scatter of the ILAS-II HNO3 values around the HNO*3 values. The HNO*3 values were calculated from O3, the spatial distribution of which was most likely controlled mainly by dynamical processes for the altitudes, latitudes, and seasons used in the present study. The use of HNO*3 values thus accounts for some dynamical effects on the HNO3 spatial distribution, so that these effects are considered to be reduced in the above calculations of the standard deviations. Therefore it is expected that the standard deviations of the differences between the ILAS-II HNO3 and HNO*3 values give an upper limit of the precision of the ILAS-II HNO3 measurements.

[25] For the ILAS-II measurements made in NH in June (in SH in April), the standard deviations of the differences between the ILAS-II HNO3 and HNO*3 values were estimated to be 14% (13%), 6% (6%), and 2% (1%) at 15, 20, and 25 km, respectively, as shown in Figure 9b (10b). We have used repeatability errors to derive upper limits of the precision of the ILAS-II HNO3 measurements, as described above. The derived precision for the NH (SH) measurements were 23% (17%), 5% (5%), and 1% (3%) at 15, 20, and 25 km, respectively (Table 1), and were very similar to those calculated from the standard deviations. These results indicate that the spatial inhomogeneity of HNO3 can be substantially reduced by using HNO*3 at the measurement times and locations for which O3 can be regarded as a long-lived tracer. Considering that both of the estimates give upper limit values, the precision of the ILAS-II HNO3 measurements for NH (SH) is estimated to be better than 14% (13%), 5% (5%), and 1% (1%) at 15, 20, and 25 km, respectively.

[26] Figure 11 summarizes the results from all the comparisons of HNO3 profiles derived from ILAS-II observations with those derived from balloon observations and the HNO*3 values. At 15–25 km, where comparisons are more robust than at other altitudes, the maximum and minimum of the relative differences of the ILAS-II HNO3 data from balloon or HNO*3 values reached −13% and +26%, respectively (Figure 11). Since this range of relative differences cannot be explained only by the above estimates of the precision of the ILAS-II HNO3 measurements and errors in the data compared with ILAS-II data, part of the differences should reflect the accuracy of the ILAS-II measurements. Using the maximum and minimum of the difference, the accuracy of the ILAS-II HNO3 measurements at 15–25 km is estimated to be better than −13%/+26%. The most plausible magnitude of the bias in ILAS-II HNO3 data at 15–25 km is +14%, which is the average of all the differences given at 15–25 km. Because the wintertime variations of the HNO3 concentration at these altitudes in the Antarctic vortex are generally much larger than the estimated precision and accuracy of HNO3 measurements by ILAS-II [e.g., Santee et al., 2004], the ILAS-II HNO3 data would be useful for investigating the processes leading to PSC formation and denitrification, as well as the spatial distribution and temporal variation of HNO3.

Figure 11.

Vertical profiles of the relative differences of ILAS-II HNO3 mixing ratios with those measured by MIPAS-B2 (solid circles) and MkIV (open circles) and HNO*3 values for NH (ELs = 75° ± 5°N) in June (solid line) and for SH (ELs = 60° ± 5°S) in April (dotted line).

5. Conclusions

[27] We have compared vertical profiles of ILAS-II HNO3 data (version 1.4) with those obtained by balloon-borne instruments (MIPAS-B2 and MkIV) over the Arctic in March and April 2003 to assess the validity of the ILAS-II HNO3 data. Additional assessments were made by comparing ILAS-II HNO3 data with HNO*3 values for ELs = 75° ± 5°N in June 1997 and for ELs = 60° ± 5°S in April 1997, where HNO*3 was calculated from the O3 mixing ratios measured by ILAS-II using the ILAS HNO3-O3 correlations that were likely maintained from 1997 until 2003 for these ELs and seasons. In support of this assumption, UARS/MLS data showed that the lower stratospheric HNO3-O3 correlations for these ELs and seasons were very compact and almost invariant over several years between 1995 and 1997.

[28] For the lower stratosphere, the standard deviation of the differences between the ILAS-II HNO3 and HNO*3 values provided a good quantitative estimate of the precision of the ILAS-II HNO3 mixing ratio data. This is because the spatial inhomogeneity of HNO3 mixing ratios at a given altitude in the lower stratosphere is largely reduced by using HNO*3 as a reference. Using this technique, the precision of the HNO3 measurements by ILAS-II at 15, 20, and 25 km was estimated to be better than 13–14%, 5%, and 1%, respectively.

[29] For all of the comparisons with balloon data and HNO*3 values, differences in HNO3, defined as ((ILAS-II data) – (other data))/(other data), ranged between −13% and +26% at 15–25 km. This indicates that the accuracy of the ILAS-II HNO3 data is better than −13%/+26% at 15–25 km. The ILAS-II HNO3 values tended to be systematically greater than balloon data and HNO*3 values. The most plausible bias in the ILAS-II HNO3 data is +14% at 15–25 km. The results presented here thus provide the quantitative basis for investigating polar stratospheric chemistry and dynamics using ILAS-II HNO3 data.

Acknowledgments

[30] The authors are grateful to all the members of the ILAS-II science team for their contributions to the ILAS-II project. We also thank the Fujitsu FIP Corporation for processing the ILAS-II data at the Data Handling Facility (DHF) of the National Institute for Environmental Studies (NIES). Support from the Ministry of the Environment is gratefully acknowledged. Part of this work was carried out when H.I. was at NIES. Work at the Jet Propulsion Laboratory, California Institute of Technology, was done under contract with NASA.

Ancillary