Conversion of mesospheric HCl into active chlorine during the solar proton event in July 2000 in the northern polar region



This article is corrected by:

  1. Errata: Correction to “Conversion of mesospheric HCl into active chlorine during the solar proton event in July 2000 in the northern polar region” Volume 116, Issue D17, Article first published online: 9 September 2011


[1] For the large solar proton event of July 2000, the Halogen Occultation Experiment instrument observed a short-term decrease of mesospheric HCl in the northern polar region. Atmospheric chemistry and ion chemistry simulations show that HCl is converted into active chlorine species (ClO, Cl, and HOCl). Two main processes drive the transformation of HCl into active chlorine: reactions of negative chlorine species directly increase the concentrations of uncharged active chlorine compounds at the expense of HCl and the production of reactive O(1D) through N(2D) + O2 → O(3P, 1D) + NO has a considerable impact on the neutral chlorine chemistry.

1. Introduction

[2] Since the pioneering works of Swider and Keneshea [1973] and Crutzen et al. [1975], it is known that the absorption of energetic solar protons in the Earth's polar atmosphere can cause considerable chemical disturbances. During such a solar proton event (SPE) initially mainly N2+, O2+, N+, O+, N, and O are released. Subsequent ion chemical reactions and recombinations lead to the formation of NOx (= N, NO, NO2) [Crutzen et al., 1975; Porter et al., 1976; Rusch et al., 1981], and HOx (= H, OH, HO2) [Swider and Keneshea, 1973; Solomon et al., 1981]. Both NOx and HOx are involved in catalytic ozone destruction cycles. The increasing concentrations of NOx and HOx, and the subsequent decrease of O3 due to several major SPEs have been observed by satellites, and reproduced by atmospheric chemistry models quite well, see for instance [Crutzen et al., 1975; Solomon et al., 1983; Jackman et al., 1995, 2001; Rohen et al., 2005; Verronen et al., 2005].

[3] It has been pointed out by [Solomon and Crutzen, 1981] that the increasing NOx and HOx concentrations after a SPE interact with chlorine species. There is a formation of chlorine nitrate at the expense of reactive radicals

equation image

This affects the chemical SPE impacts in lower and middle stratosphere [Jackman et al., 2000; López-Puertas et al., 2005] but is of minor importance at higher altitudes.

[4] Another particularly important process is the transformation of hydrogen chloride into reactive chlorine

equation image

For the SPE in October/November 2003 short-term enhancements of ClO and HOCl concentrations have been observed in the northern polar region [von Clarmann et al., 2005]. The effect is most pronounced between 40 and 45 km, and has, to the authors' best knowledge, not been reproduced by atmospheric models yet. Recently, for instance, Jackman et al. [2008] have reported on considerable differences between their modeled and the measured response of chlorine species to the SPE of October/November 2003. von Clarmann et al. [2005] attribute the enhancements of active chlorine species to reaction (1), but they mention that possibly also ion chemistry reactions could lead to a conversion of HCl to reactive chlorine species.

[5] Negative chlorine species are indeed a significant fraction of the total anion density in the mesosphere [Chakrabarty and Ganguly, 1989; Fritzenwallner and Kopp, 1998]. Several ions such as O2, O, CO3, OH, NO2, and NO3 react with HCl to produce Cl, and Cl forms Cl (X) with X = (HCl, H2O, CO2) [Kopp, 1996; Kopp and Fritzenwallner, 1997; Wayne, 2000]. The most abundant chlorine ions in the mesosphere are Cl and Cl (H2O), and while reactions of both species with atomic hydrogen rerelease HCl, some of the recombination reactions of negative chlorine species with positive ions release Cl, ClO, ClNO2, and Cl2. The increasing ion production rates during a SPE will therefore influence the chemistry of both charged and uncharged chlorine species, and because of the rapidity of the ion reactions, the impacts are expected to occur without any significant delay.

[6] A large SPE occurred during 14–16 July 2000. For the period of 10–18 of that month, the Halogen Occultation Experiment (HALOE) instrument [Russell et al., 1993] onboard the Upper Atmosphere Research Satellite observed the northern polar cap region. In terms of NOx and O3 changes due to that SPE, simulation results of a 2D atmospheric model show good agreement with the HALOE measurements, however, the peak ozone depletion during the SPE seems to be underestimated by about 0.3 ppm at altitudes around 45–50 km [Jackman et al., 2001].

[7] The HALOE instrument also provides HCl data up to about 67 km. In this paper we present modeling results of the July 2000 SPE, focusing on chlorine species, and compare it with the HALOE HCl measurements.

2. Model Description

2.1. Atmospheric Model

[8] The atmospheric model is based on the two-dimensional meteorological module THIN AIR [Kinnersley, 1996], and the chemistry model SLIMCAT [Chipperfield, 1999]. The former provides zonally averaged temperature, pressure, and wind fields, and calculates the transport of 38 chemical species on a grid of 29 isentropic, and 19 latitudinal levels. The corresponding vertical resolution is about 3 km up to approx. 100 km height. The latitudinal resolution is 9.5°. The model has a gravity wave scheme, and parameterizations for the eddy fluxes. A detailed description of the meteorological properties is given by Kinnersley [1996].

[9] The model used here is an improved version of the one used by Winkler et al. [2008]. The main difference is that for a proper simulation of the chemistry at mesospheric heights, the equilibrium family approach of the original SLIMCAT code has been replaced by an independent treatment of the chemical species. This means that, for instance, O3, O(1D), and O(3P) are integrated independently, and no longer by the means of the pseudospecies Ox. The same applies for HOx, NOx, ClOx, and BrOx. The chemical integration time step is 90 s (while it was 15 min in the former model versions). For the purpose of this study, Cl2 and CH3CN have been added to the model. Besides these modifications, the chemistry code has not been changed, and a more detailed description of it is given by Chipperfield [1999].

2.2. Ion Chemistry Model

[10] The University of Bremen Ion Chemistry (UBIC) model is used to simulate the time evolution of 138 species due to about 550 reactions (plus recombinations). It accounts for photoionization of NO by Lyman-α radiation, photodissociations of charged species, and photodetachment of electrons. Table 1 lists the species included in UBIC. The set of UBIC reactions is a combination of the reactions given by Brasseur and Chatel [1983], Böhringer et al. [1983], Brasseur and Solomon [1986], Viggiano et al. [1994], Kopp [1996], Rees [1998], Kazil [2002], and Verronen [2006].

Table 1. UBIC Species
CationsN+, N2+, NO+, NO2+, O+(4S), O+(2D), O+(2P), O2+, O2+(a4), O4+, O5+, H+, CO+, CO2+, HCO+, H2O+, O2+(H2O), H+(H2O)n=1…7, H+(H2O)(OH), H+(H2O)(CO2), H+(H2O)2(CO2), H+(H2O)(N2), H+(H2O)2(N2), H+(CH3CN), H+(CH3CN)(H2O)n=1…6, H+(CH3CN)2, H+(CH3CN)2(H2O)n=1…4, H+(CH3CN)3, H+(CH3CN)3(H2O)n=1,2, NO+(H2O), NO+(H2O)2, NO+(H2O)3, NO+(CO2), NO+(N2), NO+(H2O)(CO2), NO+(H2O)2(CO2), NO+(H2O)(N2), NO+(H2O)2(N2), NO2+(H2O)n=1,2
Anionse, O, O2, O3, O4, OH, NO2, NO3, CO3, CO4, CH3, HCO3, O(H2O), O2(H2O)n=1,2, O3(H2O)n=1,2, OH(H2O)n=1,2, NO2(H2O)n=1,2, NO3(H2O)n=1,2, CO3(H2O)n=1,2, NO3(HNO3)n=1…4, NO3(HNO3)(H2O), NO3(HNO3)2(H2O), H2SO4, HSO4(H2SO4)n=1,2, HSO4(H2SO4)(H2O), HSO4(HNO3)n=1,2, HSO4(HNO3)(H2O), HSO4(HNO3)2(H2O), HSO4(H2SO4)(HNO3), HSO4(H2SO4)(HNO3)(H2O), Cl, Cl2, Cl3, ClO, ClO(HCl), ClO(H2O), ClO(CO2), ClO(HO2), NO3(HCl)
NeutralsN(4S), N(2D), N2, O(3P), O(1D), O2, O3, H, H2, OH, HO2, NO, NO2, NO3, N2O, H2O, CH4, CH3, CO2, CO, HCO3, HNO3, HNO2, N2O5, H2SO4, CH3CN, Cl, Cl2, ClO, ClNO2, ClONO2, HCl, HOCl

[11] For the purpose of this study, the UBIC model has been operating on line with the neutral chemistry model. After each chemical time step Δt, UBIC is updated with the actual concentrations of the uncharged species, performs the ion chemistry calculations for the time period of Δt, and returns the new concentrations to the model's neutral chemistry.

[12] UBIC uses the semi-implicit symmetric method [Ramaroson, 1989] to integrate the chemical rate equations. Because of the high reactivity of some of the ion species, UBIC's time step has to be as small as 10−2 s. This makes the simulations quite time consuming. Therefore the ion chemistry has not been simulated globally but only at the 2-D model latitude 66° North corresponding to the locations of the HALOE measurements (see section 3), and at some altitudes of interest (38, 41, 46, 49.5, 54, 58, and 63.5 km).

2.3. Ionization Rates Due to Particle Precipitation

[13] The atmospheric ionization rates for the SPE in July 2000 originate from Monte-Carlo simulations of ionizing and dissociative interactions of energetic protons with air molecules. A description of the method is given by Schröter et al. [2006]. The input data for the calculations are proton counting rates provided by Geostationary Operational Environmental Satellites (GOES). The resulting ionization rates have a time resolution of 1 h, and are distributed on the main atmospheric constituents according to their abundance and ionization cross sections [Rusch et al., 1981]. The release of atomic nitrogen and oxygen is accounted for by the production rates from Porter et al. [1976], Zipf et al. [1980], and Rusch et al. [1981]. Additionally, ionizations due to galactic cosmic rays are parametrized after Heaps [1978].

2.4. Model Runs

[14] In order to be able to compare the simulation results with the occultation measurements from HALOE, model outputs during sun rise have been used. The HALOE instrument provides about 15 sun rise measurements per day at basically constant latitudes but at different longitudes [Russell et al., 1993]. The longitudinal dependency of the satellite's data is accounted for by 24 model runs performed with hourly shifted ionization rates. From the 24 simulation results a weighted average has been calculated according to the longitudes of the HALOE measurements. This approach is not fully consistent in terms of transport processes but the introduced errors are small during only a few days of model time because in the altitude region of interest here the time constants for transport are much larger than the photochemical lifetimes.

[15] Simulations have been undertaken with the atmospheric model and UBIC, but the model was additionally run without ion chemistry, using parameterizations for the production of NOx and HOx due to the SPE. Following Porter et al. [1976] and Rusch et al. [1981] a release of 1.25 NOx (45% N(4S), 55% N(2D)) per ionization is assumed. N(2D) rapidly forms nitric oxide via the reaction N(2D) + O2 → NO + O. The SLIMCAT chemistry does not account for N(2D), and therefore it is assumed that N(2D) instantaneously becomes NO. This is the commonly used way to parametrize the SPE caused NOx production in atmospheric chemistry models [see, e.g., Rohen et al., 2005; Semeniuk et al., 2005; Jackman et al., 2005b, 2008]. Consequently, for atomic oxygen the same production rate as for nitric oxide is applied. Additionally, a production of two HOx per ionization in the height region of interest here is assumed [Solomon et al., 1981].

3. Results

[16] During the July 2000 SPE, HALOE measured at latitudes 62°…69° North. For the analysis presented here, only measurements inside the polar cusp (geomagnetic latitude > 60°) are considered. This is the region in which usually a homogeneous ionization due to precipitating solar protons is assumed [Vitt and Jackman, 1996; Jackman et al., 2000, 2005a]. HALOE level 3AT data have been used which provide HCl profiles up to about 67 km. Figure 1 shows the observed HCl mixing ratios around the SPE. There is a short-term HCl decrease, largest on 14 and 15 July, and most pronounced for altitudes >45 km.

Figure 1.

Daily mean HCl mixing ratios for 10–18 July 2000 as measured by the HALOE instrument in sun rise mode at latitudes between 62° and 69°N (only data inside the polar cusp are used).

[17] The simulations with the UBIC model yield significantly larger HCl losses than the model runs with parameterized production of NOx and HOx, (hereafter called PARAM), and they agree better with the observations (see below). (PARAM stands for the parameterized production rates for HOx and NOx.) By performing several UBIC model runs it was possible to identify the following two main processes (besides reaction 1) which drive the transformation of HCl into active chlorine species:

[18] 1. Reactions of negative chlorine species directly increase the concentrations of uncharged active chlorine compounds at the expense of HCl. This is the dominating process above about 55 km. At lower altitudes, chlorine ions are only a small fraction of the total charge density, and therefore the impacts on the neutral chemistry are of minor importance.

[19] 2. The reaction N(2D) + O2 → NO + O is believed to produce both ground state as well as electronically excited oxygen atoms

equation image

Though this is not an ion reaction, it is included in the UBIC scheme to resolve the rapid N(2D) chemistry. The production of O(1D) leads to a release of OH, initially mainly through O(1D) + H2O → 2OH which in turn depletes HCl via reaction (1).

[20] It turned out that the simulation results are very sensitive to changes of the O(1D) branching ratio, β = ΔO(1D)/(ΔO(1D) + ΔO(3P)), from reaction (2). Miller and Hunter [2004] found an upper limit of β = 2%.

[21] Figure 2 shows the modeled HCl losses, both from UBIC simulations as well as from simulations using parametrized HOx and NOx production rates, in comparison with HALOE measurements. The UBIC results are much closer to the HALOE observations, and at lower altitudes, a variation of the O(1D) branching ratio has a large impact on the resulting HCl losses. With β = 2%, the HCl loss is overestimated but with values 0.5% or 0.2% the model results agree much better with the measurements.

Figure 2.

Differences of zonally averaged HCl mixing ratios at three different altitudes for sun rise conditions (differences with respect to the mean HCl values of 10–12 July). Shown are HALOE data and simulation results from the atmospheric model at 66.5°N. The HALOE error bars represent 1 standard deviation. PARAM indicates the model with parametrized production rates for HOx and NOx. The UBIC simulations differ in the branching ratio for O(1D) production from reaction N(2D) + O2 (2, 0.5, and 0.2%); see text for details.

[22] A β of the order 0.2% to 0.5% is also supported by the comparison of modeled ozone losses with the HALOE ozone data. While the ozone destruction during the SPE below about 50 km is significantly overestimated with β = 2%, with β = 0.5% or β = 0.2% the modeled ΔO3 agrees much better with the HALOE measurements; see Figure 3. Figure 3 shows that the model with parameterized NOx and HOx production underestimate the ozone losses in that altitude region. This is in agreement with the simulation results of Jackman et al. [2001]. At higher altitudes the UBIC modeled ozone losses are also larger than in the simulations with parameterized HOx and NOx production rates but the differences are smaller, for instance at 58 km the peak ΔO3 are UBIC = −0.75 ppm, HALOE = −0.65 ppm, and PARAM = −0.55 ppm.

Figure 3.

Differences of zonally averaged ozone mixing ratios (differences with respect to the mean HCl values of 10–12 July) at 46 and 49 km. Shown are HALOE data and simulation results from the atmospheric model at 66.5°N. The HALOE error bars represent 1 standard deviation. PARAM indicates the model with parametrized production rates for HOx and NOx. The UBIC simulations differ in the branching ratio for O(1D) production from reaction N(2D) + O2 (2, 0.5, and 0.2%).

[23] Figure 4 shows the observed and the modeled HCl losses as a function of altitude for the 15 July. While at altitudes below about 46 km a small β yields best agreement with the HALOE measurements, the observed ΔHCl peak around 50 km seems to be underestimated, and is better reproduced with higher values of β. Therefore, no final conclusion can be drawn here regarding the best value for β, however, considering the result of the ozone comparison as well, the authors tend to favor a smaller β. The HALOE HCl uncertainties increase with altitude (Figure 4), and this could partly explain the difference between model results and observations, especially above about 50 km.

Figure 4.

Observed and modeled zonally averaged HCl differences at sun rise on 15 July (with respect to the mean HCl mixing ratio of 10–12 July) as a function of altitude. Shown are HALOE ΔHCl and model results with UBIC (with branching ratios for O(1D) production 0.2% and 2%) and with parameterized production rates for HOx and NOx, respectively. The HALOE error bars represent 1 standard deviation.

[24] The contribution from the O(1D) channel to the impact on HCl becomes less important at higher altitudes. While in the stratosphere the released Cl is converted rapidly through Cl + O3 → ClO + O2, there is a regeneration of HCl by reactions of Cl with CH4, H2, and HO2 at higher altitudes. This limits the impact on HCl. The same is true for the ion chemical HCl depletion but its effect is much larger than the O(1D) impact. Therefore, above about 54 km the HCl loss is mainly due to the direct negative ion chemical conversion of HCl into Cl. The UBIC modeled HCl decrease is significantly larger than in the model runs with parametrized HOx and NOx production rates but it is still smaller than the measured ΔHCl. As the rate coefficients of the negative chlorine reactions are not known with very high accuracy, additional model runs have been performed with scaled reaction rate coefficients of all the negative chlorine reactions. With a scaling of a factor of two, the ΔHCl increases significantly and reaches almost the peak values measured around 58 km but then the HCl decrease at 64 km is overestimated; see Figure 5. It is not clear whether the uncertain rate coefficients are responsible for the differences between modeled and observed ΔHCl or whether ion chemical processes are missing in the UBIC model. Therefore, further comparisons of simulation results with measurements are needed. Figure 5 shows also the results of a simulation without any negative chlorine chemistry yielding significantly smaller HCl losses.

Figure 5.

Observed and modeled zonally averaged HCl differences at sun rise on 15 July (with respect to the mean HCl mixing ratio of 10–12 July) as a function of altitude. Shown are UBIC model results with different scalings of the negative chlorine chemistry, (×0) equals all reactions deactivated, (×1) equals default reaction rates, and (×2) equals reaction rates doubled. For all runs the branching ratios for O(1D) production was 0.2%. The HALOE error bars represent 1 standard deviation.

[25] If all negative ion chemistry reactions involving chlorine species are switched off, and the O(1D) production from reaction (2) is deactivated, the simulation results do not differ significantly from the results of model runs with parameterized HOx and NOx production rates.

[26] The decreasing HCl mixing ratios during the SPE correspond to increasing amounts of other chlorine species, in the first place ClO, Cl, and HOCl. The effect is short lived, and there is a clear diurnal cycle: During night the HCl loss is largest; see Figure 6.

Figure 6.

Modeled ΔHCl at 46 and 58 km in comparison with mixing ratios of other chlorine species during the SPE (UBIC run with 0.2% branching ratio for O(1D) production).

[27] While the simulations predict that the transformation of HCl into active chlorine is a short-lived effect, the observations seem to indicate that the HCl recovery is somewhat slower (Figures 2 and 3). But because of the fact that the HALOE observations only last until the 18 July, no final conclusion can be drawn about the recovery here.

4. Summary and Conclusions

[28] Atmospheric chemistry and ion chemistry simulations of the solar proton event in July 2000 have been compared with measurements of the HALOE instrument in the northern polar region. The simulations indicate that the observed short-term decrease of mesospheric HCl is due to a conversion of HCl into active chlorine species (ClO, Cl, and HOCl). The magnitude of the HCl loss cannot be explained solely by the gas phase reaction HCl + OH → Cl + H2O (and assuming a production of 2 HOx per ionization). If reactions of negative chlorine species and the production of O(1D) from the reaction N(2D) + O2 are taken into account, the modeled HCl decreases agree significantly better with the observations. While the release of O(1D) during the solar proton event of July 2000 has the largest impact on the chlorine chemistry below about 55 km, the ion chemical conversion of HCl into Cl is the governing process at higher altitudes. A critical, and yet not very well known parameter is the branching ratio for the release of O(1D) from the reaction N(2D) + O2. The upper limit of 2% causes both too large HCl as well as O3 losses. Smaller values, of the order 0.2%–0.5% yield much better results. Further comparisons with measurements will be needed to test our understanding of chlorine activating processes and their representation in ion chemistry models in detail. The O(1D) production should be implemented in the parameterizations describing the neutral atmospheric effects of particle impact ionizations.

[29] As the SPEs of July 2000 and October/November 2003 were of about the same size [Jackman et al., 2005a], it is not totally unreasonable to compare their impacts on chlorine species, though one should be aware that both the atmospheric ionization profiles as well as the photochemical conditions differed. The modeled ClO and HOCl increases after the SPE in July 2000 are of the same order as the enhancements observed for the October/November 2003 SPE, e.g., modeled peak ΔClO and ΔHOCl at 41 km (not shown) are 0.3 ppb and 0.18 ppb, respectively, whereas about +0.3 ppb have been observed for both species at 40 km by von Clarmann et al. [2005]. Therefore, we suggest that the observed chlorine activation in October/November 2003 might be caused by the same processes as identified here for the July 2000 solar proton event.


[30] This work is financially supported by the German Research Council (Deutsche Forschungsgemeinschaft (DFG)) within its priority programme CAWSES (Climate and Weather of the Sun-Earth System).