Three-dimensional GCM modeling of nitric oxide in the lower thermosphere

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 111,, doi: /2005ja011543, 2006 Three-dimensional GCM modeling of nitric oxide in the lower thermosphere A. L. Dobbin, 1 A. D. Aylward, 1 and M. J. Harris 1 Received 21 November 2005; revised 17 March 2006; accepted 11 April 2006; published 26 July [1] The coupled middle atmosphere and thermosphere (CMAT) general circulation model has been used to predict the global distribution of nitric oxide (NO) in the lower thermosphere. This three-dimensional (3-D) model incorporates a complex ion and neutral chemical scheme, high-resolution solar flux data and variable auroral energy inputs. Comparison of simulated NO densities with those observed by the Student Nitric Oxide Explorer (SNOE) show that the CMAT model is able to recreate large-scale features in the observed NO distribution under differing geophysical conditions. While previous studies have used 1-D photochemical models to speculate on the latitudinal extent of aurorally produced NO, we have simulated NO production and transport in three dimensions under both geomagnetically stable and disturbed conditions. We have shown that the spatial and temporal distribution of aurorally produced NO is highly variable with location and local time. CMAT simulations suggest that under moderate geomagnetic conditions, the most equatorward geographic latitudes to be influenced by aurorally produced NO are 30 S and 45 N. Under conditions of high geomagnetic activity, aurorally produced NO is present at latitudes poleward of 15 S and 28 N. The asymmetry in latitudinal extent is attributed to the greater offset between geographic and geomagnetic poles in the southern hemisphere. NO density maxima are predicted to occur between 14 and 48 hours after a period of high geomagnetic activity, their spatial and temporal distribution depending on location with respect to the auroral oval and local time at which auroral forcing occurs. Citation: Dobbin, A. L., A. D. Aylward, and M. J. Harris (2006), Three-dimensional GCM modeling of nitric oxide in the lower thermosphere, J. Geophys. Res., 111,, doi: /2005ja Introduction 1 Atmospheric Physics Laboratory, Department of Physics and Astronomy, University College London, London, UK. Copyright 2006 by the American Geophysical Union /06/2005JA [2] In recent years, increasing attention has been paid to the abundance and variability of nitric oxide (NO) in the atmosphere. One reason for interest in this species arises from its low ionization threshold (ionization potential 9.26 ev). NO concentration has a direct impact on the ion composition of both the E region and the daytime D region, the latter being largely created by photoionization of NO by solar Lyman alpha [Nicolet and Aiken, 1960]. Second, radiative emission at 5.3 mm from NO is an important cooling mechanism in the thermosphere [Kockarts, 1980], capable of reducing the heating efficiency near 130 km by a factor of 4 in the summer hemisphere [Richards et al., 1982]. During the polar night, the lifetime of NO is sufficiently long that it can be transported downwards into the stratosphere [Solomon et al., 1982; Barth et al., 1999; Callis et al., 1998; Vitt et al., 2000] where it increases the abundance of odd-nitrogen species (NO y ) and can contribute to the catalytic destruction of ozone [Callis et al., 2001]. This could affect the temperature structure of the stratosphere and is thought to be one mechanism by which coupling of the upper and lower atmosphere occurs. NO in the MLT region displays a large degree of variability with season, solar cycle, solar rotation, and auroral activity. Owing to the chemical and radiative properties of this minor constituent, variability in NO abundance can have a significant impact on the structure and energetics of the atmosphere. [3] Both observational and numerical modeling studies have found a correlation between the flux of solar soft X rays in the 2 10 nm region and low-latitude NO concentrations [Barth et al., 1988; Siskind et al., 1990; Barth et al., 1999; Swaminathan et al., 2001]. X rays in this wavelength band are absorbed between 100 and 120 km altitude and the resultant ionization creates photoelectrons which contribute substantially to NO production. At high latitudes, auroral energy is the primary driver for producing NO [Barth et al., 2003; Baker et al., 2001; Solomon et al., 1999]. An increase in the flux of precipitating electrons leads to an increase in ionization that in turn leads to enhanced production of NO. A relationship therefore exists between peak NO abundances and auroral activity indicators, such as the Kp planetary geomagnetic index [Marsh et al., 2004]. Excess NO produced at auroral latitudes is thought to be transported by meridional winds, leading to increased concentrations of NO at middle and possibly even 1of14

2 low latitudes. Downward diffusion of excess NO, created above 140 km in response to enhancements in temperature and molecular oxygen density, is also thought to contribute to midlatitude NO increases following geomagnetic storms [Siskind et al., 1989a; Richards, 2004]. Through consideration of these mechanisms, NO abundance can be used as a diagnostic of the temporal and spatial distribution of solar and auroral energy deposition into the MLT region. [4] In a recent study by Barth and Bailey [2004], NO densities in the lower thermosphere measured with the SNOE satellite, were compared with those calculated by a one-dimensional (1-D) photochemical model that did not include auroral energy sources. While a good match existed between modeled and observed NO abundances at low latitudes, the SNOE data revealed the presence of excess NO concentrations poleward of ±30. The proposed source of this excess was aurorally produced NO that had been transported equatorward, although this could not be simulated with their 1-D model. Marsh et al. [2004] also proposed that aurorally driven variability in NO density exists poleward of ±30. Both the 1-D modeling study of Barth and Bailey [2004] and the observational study of Barth et al. [2003] suggested that at times of large geomagnetic storms, aurorally produced NO can be transported all the way to the equator. The importance of downward diffusion of NO is discussed by Richards [2004] and Siskind et al. [1989a], whose 1-D modeling studies show large increases in midlatitude NO production at altitudes above 140 km following geomagnetic storms. Downward vertical diffusion of this enhanced NO was shown to cause an increase in modeled NO density at 110 km altitude. The increase at 110 km occurred approximately 1 day after a storm and lasted for more than a day. While there has been much speculation on the latitudinal extent of NO following periods of high geomagnetic activity, to date, very few 3-D GCM modeling studies covering this altitude region have been performed. [5] Despite considerable effort over the last decade, some fundamental questions about the creation, destruction, and transport of NO remain unanswered. The profile of NO at any given time evolves through a complex set of ion-neutral and neutral-neutral reactions, many of which are temperature dependent (e.g., N( 4 S) + O 2! NO + O) [Fuller- Rowell, 1993; Siskind and Rusch, 1992; Siskind et al., 2004]. Siskind and Rusch [1992] found that changing the temperature dependence of the reaction between ground state atomic nitrogen and molecular oxygen, in line with JPL90 recommendations, led to an over estimation of NO at 200 km by up to a factor of 3 compared to rocket measurements. Ambiguity of key reaction rates and branching ratios combined with sensitivity to highly variable auroral and solar energy inputs means that the modeling of NO production is a nontrivial task. Calculating the loss of NO can be problematic due to uncertainties about the opacity of NO to solar radiation, which can act as a primary destruction mechanism through the dissociation reaction, NO + hn! N + O. Indirect effects such as Joule heating at auroral latitudes can cause enhancements in NO through the temperature sensitivity of NO production reactions [Siskind et al., 1989a, 1989b]. Conflicting measurements of solar soft X-ray fluxes have lead to uncertainties in low-latitude production rates. Several sensitivity studies have been carried out which investigate how changes to key reaction rates, branching ratios and the state of the neutral atmosphere affect 1-D simulations of NO concentration [Siskind et al., 1989a, 1989b; Barth, 1992; Siskind et al., 1995]. Few modeling studies however, have been able to accurately reproduce the absolute magnitudes seen in satellite or rocket data. [6] Modeling of NO is notoriously difficult and despite recognized demand [e.g., Barth and Bailey, 2004; Siskind, 2000] comprehensive 3-D modeling of NO has not yet been accomplished. Without 3-D GCM type studies, questions about the spatial and temporal extent of NO transport, both horizontally and vertically, could remain unanswered. 3-D GCMs are the ideal tool for exploring the relative roles of horizontal advection and vertical diffusion in enhancing midlatitude NO around the peak altitude of 110 km. Questions about NO creation in key areas where measurement is difficult, such as the winter pole, could also be addressed with 3-D GCM studies, as could the timescales of thermospheric cooling following periods of high auroral activity. [7] In the following section the main production and loss mechanisms for NO are described. An outline of the CMAT GCM is then given along with key reaction rates and branching ratios used for the self consistent calculation of NO in the model. CMAT calculated NO density profiles produced under differing solar and geomagnetic activity conditions are presented and compared with data from the SNOE satellite. Two CMAT experiments are then described, the first of which explores the latitudinal range over which aurorally produced NO is distributed under conditions of moderate geomagnetic activity. Simulations were performed with and without the effects of advection in order to demonstrate the impact of winds on the distribution of aurorally produced NO. In the second experiment, a simple quasi-step geomagnetic storm was generated. The resultant profiles of NO density are presented and compared to profiles produced under conditions of moderate activity. An indication of the latitudinal and temporal response of NO to geomagnetic forcing is thus obtained. 2. Production and Loss of Nitric Oxide [8] Nitric oxide in the lower thermosphere is predominantly produced via the chemical reaction between excited atomic nitrogen and molecular oxygen N 2 D þ O2! NO þ O ð1þ This is the primary production mechanism at altitudes where thermospheric NO density peaks (around 110 km). In order for this reaction to proceed rapidly, the nitrogen atoms must have excess energy. Several paths exist for the production of excited nitrogen atoms, the most significant being dissociative recombination reactions that occur with ambient electrons NO þ þ e! N 2 D; 4 S þ O ð2þ N þ 2 þ e! 2N 2 D; 4 S ð3þ 2of14

3 Both of these reactions are temperature dependent. Reactions with energetic electrons (e*) (>10 ev) are also important for creation of N( 2 D) either directly via the reaction N 2 þ e! 2N 2 D; 4 S þ e* ð4þ where the strong N 2 bond is broken, or through a two-step process followed by N 2 þ e*! N þ 2 þ e þ e*; N þ 2 þ O! NOþ þ N 2 D At low latitudes these energetic electrons are primarily produced by photoionization reactions between high-energy photons and neutrals and are termed photoelectrons. At high latitudes, they are predominantly precipitating auroral electrons. Secondary electrons from processes such as auroral particle bombardment also make up a substantial proportion of the energetic electron population. It is interesting to note that the NO + produced in reaction (6) also produces N( 2 D) through reaction (2). [9] At altitudes above 120 km where the ambient temperature generally exceeds 400 K, the reaction between ground state atomic nitrogen, N( 4 S), and molecular oxygen becomes the dominant source of NO ð5þ ð6þ N 4 S þ O2! NO þ O ð7þ This reaction is slow and very sensitive to temperature. [10] Two major loss processes for NO occur through reactions with ground state nitrogen, and O 2 + NO þ N 4 S! N2 þ O ð8þ NO þ O þ 2! NOþ þ O 2 During daylight hours, photodissociation of NO by solar UV irradiance is also a significant loss mechanism ð9þ NO þ hn! N 4 S þ O ð10þ Note that while reactions (9) and (10) destroy NO, they recycle odd nitrogen. If the ground state nitrogen produced in reaction (10) goes on to destroy more NO through reaction (8), the effectiveness of photodissociation as a loss mechanism is increased. Only reaction (8) truly destroys odd-n. After sunset, although NO is no longer destroyed by solar UV, its abundance is reduced. This is because, with the exception of auroral production, no excited state nitrogen is being created and thus NO production is greatly reduced. Any remaining N( 4 S) and O 2 + will react with NO to destroy it through reactions (8) and (9), but in the absence of sunlight, the densities of N( 4 S) and O 2 + quickly diminish. Once the loss processes are removed, NO may be transported to lower altitudes and latitudes, especially from auroral regions where NO production continues. [11] It is clear that the relative abundance of excited and ground state nitrogen is a key factor in both the production and loss of NO. The flux and energy of solar irradiance are fundamental in determining the abundance of these species and of O 2 +. Solar UV can also directly destroy NO through photodissociation. Changes in thermospheric heating can impact key reaction rates, again resulting in changes in NO density. Note that through consideration of the chemical and diffusive lifetimes of NO, Bailey et al. [2002] concluded that the abundance of NO at any one time is determined by the level of solar and auroral energy deposition over the previous day. Horizontal transport therefore plays an important role in determining the NO distribution. Although the key processes for the creation and loss of NO have been outlined above, many other reactions must be considered in order to represent the full picture. 3. CMAT GCM [12] CMAT is an extension of the University College London time-dependent Three-Dimensional Coupled Thermosphere Ionosphere Plasmasphere (CTIP) model [see Fuller-Rowell et al., 1996a; Millward et al., 1996]. The model has a vertical range from 10 mbar (30 km) to 7.6 pbar ( km) and solves the nonlinear equations of energy, momentum, and continuity on a model grid of 2 latitude, 18 longitude and a third scale height. A detailed description of CMAT is given by Harris [2001] and a summary of its key points is given by England et al. [2006]. Recent model developments are described in detail by Dobbin [2005]. Features of the model that directly influence the representation of NO are outlined below. [13] Thermospheric heating, photodissociation, and photoionization due to the absorption of solar X-ray, EUV, and UV radiation between 1.8 and 180 nm are calculated using fluxes from the SOLAR2000 empirical model [Tobiska et al., 2000]. A resolution of 1 nm is used for wavelengths above 6 nm. Below 6 nm, finer resolution is used. Hard X-ray fluxes at wavelengths between 0.2 and 1.8 nm are taken from the GLOW model [Solomon et al., 1988; Solomon and Abreu, 1989]. Absorption and ionization cross sections between 0.2 and 105 nm are based on Henke et al. [1993] and Fennelly and Torr [1992]. Between 105 and 1800 nm the cross sections are provided by R. Viereck (private communication, 2003). [14] Primary photon ionization rates of the major species are calculated using a combination of self-consistently calculated number densities, absorption and ionization cross sections, and solar flux data. Relative partitioning between the possible products of the ionization process are based on the model of Strickland and Meir [1982] as described by Fuller-Rowell [1993]. [15] Ionization and dissociation of each major species due to energetic photoelectrons is calculated using vertical profiles of the ratio of ionization due to photoelectrons, compared with primary photon ionization. For altitudes above 100 km, ratios based on calculations using the photoelectron code of Strickland and Meir [1982] [Fuller- Rowell, 1993] were found to result in peak photoelectron ionization rates that were notably less than those presented by Bailey et al. [2002]. This is thought to be related to the omission of solar photons with energies greater than 450 ev in the photoelectron spectrum calculation, and the absence of 3of14

4 Auger ionization, both of which are thought to be important in ionization rate calculations [Siskind et al., 1995]. In CMAT the ionization ratios have therefore been adjusted such that the major species peak photoelectron ionization rates match those presented by Bailey et al. [2002]. [16] Below 100 km, values of these ratios have been extrapolated (O 2, O) or estimated (N 2 ) based on the available energy from photons that would penetrate to that altitude, as described by Fuller-Rowell [1993]. Partitioning ratios for the possible products of major species ionization by both auroral electrons and photoelectrons are taken from Fuller-Rowell [1993]. [17] Radiative cooling due to 5.3 mm emission from excited NO is calculated as described by Kockarts [1980], with a collisional deactivation rate of cm 3 s 1 [Dodd et al., 1999]. [18] At large solar zenith angles (>75 ), calculation of the atmospheric column of each absorbing species is performed using an approximation of the Chapman Grazing incidence function due to Smith and Smith [1972]. [19] The CMAT neutral chemical scheme solves for major and minor neutral constituents (N 2,O 2,O x =O( 1 D) + O( 3 P) + O 3,N( 4 S), N( 2 D), NO x =NO+NO 2,HO x =OH+HO 2 +H, H 2 O, H 2, CO, CO 2, CH 4, NO 3, He) and ions species (NO +,N +,N 2 +,O 2 +,O +, and H + ). Photochemical equilibrium is assumed for O( 1 D), H 2 O 2, NO 3, NO +, N +, N 2 +, O 2 +, and N( 2 D). In order to improve the representation of NO in the model, several of the key reaction rates and branching ratios have been revised in line with recently published values. A list of reaction rates and branching ratios for processes forming NO and its precursor species is given in Appendix A. The full chemical scheme can be found in the work of Dobbin [2005]. [20] High-latitude auroral precipitation from the TIROS/ NOAA auroral precipitation statistical model is included [Fuller-Rowell and Evans, 1987, Codrescu et al., 1997] along with electric field strengths from Foster et al. [1986]. The effect of high-latitude small-scale electric field variability is included [Codrescu et al., 2000]. The vertical eddy diffusion coefficient used is a height dependent global mean based on the climatology of Garcia and Solomon [1983]. 4. CMAT Modeled Nitric Oxide Distributions [21] The CMAT model has been run for a period of 40 model days such that it is in a steady state, where the diurnal variability is repeated from day to day. Figure 1 shows two examples of the zonally averaged NO number density in the lower thermosphere as calculated by the CMAT model and as measured by the SNOE satellite [Solomon et al., 1996; Merkel et al., 2001]. SNOE observations are taken within 30 min of 1045 LT at a resolution of 3.3 km altitude, 5 geomagnetic latitude. The zonal averages shown here are from the Version 2, level 4 data set [Barth and Bailey, 2004]. [22] In the top panel of Figure 1, NO densities measured on day 266 of 1999 are shown along with CMAT calculated densities for a local time of This time was chosen as it is a standard output from the model and is within the range of local times at which SNOE measurements of NO were made. NO abundance at any one time is representative of the solar and geomagnetic conditions over the past day [Bailey et al., 2002, Marsh et al., 2004]. Values of the F10.7 solar activity proxy and geomagnetic activity index used as CMAT model inputs are therefore appropriate to the previous day (day 265 of 1999) when solar activity was moderate (F10.7 = 142) and geomagnetic activity was high (Kp = 5 ). The CMAT model is able to reproduce the large-scale features in the zonal mean NO distribution. High-latitude NO densities are clearly greater in magnitude than those at low latitudes due to enhanced production of NO in the auroral zones. In the northern hemisphere, CMAT predicts a peak NO density of m 3 occurring between 60 and 70 in good agreement with the observed peak of m 3 close to 60 latitude. A southern hemisphere NO peak of m 3 is observed approximately 10 poleward of the modeled peak that occurs at around 60 latitude and reaches m 3. At both high and low latitudes, the peak in thermospheric NO abundance occurs close to 110 km altitude. The CMAT model is able to reproduce this observed characteristic. The high-latitude peaks predicted by CMAT extend slightly further into the thermosphere than those observed by SNOE. This is thought to be related to the statistical nature of the highlatitude energy inputs used in CMAT which are not able to represent the detailed local response to auroral activity seen in instantaneous observations. The temperature sensitivity of the reaction between ground state atomic nitrogen and molecular oxygen may also play a role. Small increases in simulated thermospheric temperature above about 120 km can result in large increases in the production of NO. [23] Figures 1c and 1d show the zonal mean NO abundance measured by the SNOE satellite on day 80 of 1998 and as calculated by the CMAT model. A Kp index of 2 + was used in this model simulation along with an F10.7 of 126, in line with average geophysical conditions on the day preceding the observations. The CMAT NO distribution shown is for a local time of Both observed and modeled high-latitude NO densities are lower than in the previous example. This is the result of decreased auroral production. The small reduction in low-latitude NO abundance is indicative of a reduction in soft X-ray flux. The CMAT model is again able to reproduce the large-scale features present in the observed NO distribution. A northern hemisphere peak of m 3 is predicted by the model to occur close to 80 latitude, 12 poleward of the observed peak that occurs around 68 and reaches m 3. In the southern hemisphere, simulated and observed peak NO densities of m 3 and m 3, respectively, occur close to 80 latitude. Discrepancies between the simulated and observed NO distributions at high latitudes are again attributed to the statistical nature of the high-latitude auroral and electric field inputs used in the model. The high degree of variability in thermospheric NO, evident from extended observational datasets such as those provided by the SNOE satellite, should also be kept in mind when comparing GCM predictions with one-off measurements of NO abundance. 5. Study 1: Latitudinal Extent of Aurorally Produced NO [24] In a recent study by Barth and Bailey [2004], the latitudinal extent of aurorally produced NO was inferred by 4of14

5 Figure 1. Zonal mean nitric oxide number densities observed by the SNOE satellite and as calculated by the CMAT GCM. SNOE observations are from (a) day 266 of 1999 and (c) day 80 of Geophysical conditions for CMAT calculations are appropriate to (b) day 265 of 1999 and (d) day 79 of Contour intervals are every m 3. comparing results from a 1-D photochemical model with data from the SNOE satellite. Although the 1-D model used did not include the effects of horizontal transport or particle precipitation, it was able to adequately describe thermospheric NO at 110 km between 30 S and 30 N. NO concentrations observed at latitudes poleward of that region exceeded those predicted by the model. The authors concluded that electron precipitation and horizontal transport processes must therefore be taken into account when modeling NO concentrations poleward of 30. Richards [2004] and Siskind et al. [1989a] highlighted the importance of vertical diffusion in determining midlatitude enhancements in NO following geomagnetic storms. A model such as CMAT that includes auroral energy inputs and 3-D atmospheric transport is well suited to studies of the abundance, horizontal distribution and diurnal variability of high-latitude NO under variable geomagnetic conditions. Such detailed investigations of NO production and transport in the lower thermosphere are not possible with 1-D photochemical models and have not previously been performed with 3-D GCMs. [25] In this study the CMAT model has been used to investigate the spatial and temporal distribution of NO produced under conditions of moderate geomagnetic activity. Results from three CMAT model runs are presented, each for a total of 40 model days such that the model is in a stable state. The first simulation was run with no particle precipitation inputs. The second simulation included a moderate level of auroral activity with particle precipitation and electric field inputs defined according to a single power index (PI). This index quantifies both the spatial extent and intensity of auroral particle precipitation based on observations made by the TIROS/NOAA satellites [Fuller-Rowell and Evans, 1987; Codrescu et al., 1997] and describes the total amount of particle energy delivered to an entire auroral hemisphere over half an orbit. Average patterns of ionospheric convection electric fields, derived from Millstone Hill radar observations of plasma convection data, have been keyed into this index as described by Foster et al. [1986]. In this case, the power index has been set to 5, which is representative of a Kp of 2 + and an estimated hemispheric power input of between 10 and 16 GW. In order to demonstrate the effects of transport on aurorally produced NO, the third simulation was run with auroral inputs appropriate to a Kp of 2 +, but with horizontal and vertical advection terms set to zero in the minor species 5of14

6 Figure 2. Latitude versus height plots of nitric oxide number density at a longitude of 0, at 1200 UT. Profiles are as calculated by CMAT (a) with auroral energy inputs appropriate to a Kp of 2 + and (b) with no auroral inputs. Conditions are appropriate to northern spring equinox at an F10.7 of 105. Contour intervals are every m 3. continuity equation. All other geophysical conditions in the three simulations are the same and are appropriate to low solar activity (F10.7 = 105) during northern spring equinox (day 79 of the year). [26] Figure 2 shows latitude versus height profiles of NO number density as calculated by CMAT for the first two cases described above. Plots are for a longitude of 0 at 1200 UT and demonstrate the characteristic morphology of NO abundance resulting from each model simulation. Comparison of the NO distributions created with (Figure 2a) and without (Figure 2b) auroral forcing reveals enhanced levels of NO when auroral inputs are included. The high-latitude peaks occur at 110 and 118 km altitude in the northern and southern hemispheres respectively, the northern hemisphere peak being slightly greater in magnitude than that in the south. Differences in the locations and magnitude of these peaks arise from asymmetries in the auroral energy inputs to each hemisphere, associated with the offset magnetic field. [27] A more detailed appreciation of the equatorward spread of aurorally produced NO can be gained from Figure 3 which shows difference plots of NO number density as calculated in simulations 1 and 2 described above. The plots show the difference in calculated NO density on a constant pressure level, at approximately 110 km altitude where the peak NO abundance is expected to occur. The profiles shown are from four different universal times, illustrating that aurorally produced NO is highly variable on both temporal and spatial scales. When auroral forcing is included in CMAT, enhanced NO concentrations are present in the high-latitude regions of the northern and southern hemispheres. Differences in dayside and nightside magnetic reconnection result in different auroral forcing throughout the diurnal cycle. This is reflected in the high-latitude NO distribution. [28] At 0000 UT significant amounts of excess NO are present at geographic latitudes above approximately 50 N (geomagnetic latitude 48 ) and poleward of 30 S (geomagnetic latitude 44 ). The largest enhancement is seen close to 75 S (geomagnetic latitude 82 ) where an excess NO density of m 3 is predicted. The regions of enhanced NO illustrate a hemispheric asymmetry in auroral energy input that arises as a result of the offset between the geographic and geomagnetic poles. Aurorally produced NO reaches further toward the equator in the southern hemisphere where the offset between geographic and geomagnetic poles is largest. At 0600 UT, the northern hemisphere high-latitude NO is more varied and extends slightly further equatorward than at 0000 UT. In the southern hemisphere however, the maximum in aurorally produced NO is less pronounced, as is the degree of spatial variation. At 1200 UT the maximum NO enhancement occurs in the northern hemisphere where a peak of m 3 is predicted to occur close to 75 latitude (geomagnetic latitude 80 ). A similar situation is seen at 1800 UT although the high-latitude NO enhancement in the southern hemisphere is greater than at midday. Over the whole day, the latitudes furthest toward the equator to be affected by significant amounts of aurorally produced NO, under conditions of moderate geomagnetic activity, are approximately 30 S (geomagnetic latitude 39 ) and 45 N (geomagnetic latitude 43 ). [29] The mechanisms that contribute to midlatitude and low-latitude NO increases following geomagnetic activity are not yet fully understood. Barth and Bailey [2004] suggested that NO produced at auroral latitudes could be transported equatorward and account for observed increases in midlatitude NO density. In the 1-D modeling study of Siskind et al. [1989a], the importance of horizontal transport of NO was questioned and downward diffusion was identified as the cause of midlatitude NO increases. Richards [2004] also showed that downward diffusion is an important source of NO at 110 km altitude. While it is beyond the scope of the current study to define the relative role of each process, we will demonstrate that advection does have an 6of14

7 Figure 3. Difference plots of NO number density as a function of latitude and longitude at an altitude of 110 km at 4 different universal times. Plots show the difference between NO densities calculated in two CMAT simulations, the first with auroral inputs appropriate to a Kp of 2 + and the second with no auroral forcing. Geophysical conditions are appropriate to northern spring equinox with an F10.7 of 105. Contour intervals are every m 3. influence on the horizontal distribution of NO. Plots of the difference in calculated NO density between simulations 2 and 3 described above are shown in Figure 4. Both of these model runs included auroral energy inputs appropriate to a Kp of 2 +. Simulation 2 included the effects of horizontal and vertical species advection whereas simulation 3 did not. The plots are derived from calculated NO number densities at a longitude of 0 at four different universal times. Positive values (shown in white) illustrate regions where NO densities are enhanced by transport, whereas negative values (shown in grey) indicate areas of depleted NO. In each of the plots, midlatitude NO concentrations are enhanced when advection terms are included in the species continuity equation. Depending on time of day, southern hemisphere NO densities at the peak altitude of approximately 110 km, are reduced at latitudes poleward of about 60 to 70. At all universal times shown, NO densities are enhanced between 50 S and 60 S, extending to 30 S ( 25 geomagnetic latitude) at 0600 UT. In the northern hemisphere a similar situation occurs although the region of enhanced NO is further poleward. The plots show that aurorally produced NO is being transported away from the region of production, to latitudes equatorward of the auroral oval. [30] At low latitudes, the NO number density may be enhanced or diminished depending on the altitude and time of day. The cellular structure is indicative of tidal influence, as is the vertical distance between peaks and troughs which matches the wavelength of the diurnal tide (approximately 25 to 30 km). The perturbations are strongest at the equator where tidal variations in the vertical wind tend to maximize. The influence of vertical tidal motions on the distribution of NO has been investigated by Marsh and Roble [2002] who confirmed that sunrise/sunset asymmetries in satellite measurements of NO are predominantly due to tidal forcing. 7of14

8 Figure 4. Difference plots of nitric oxide number density at four different universal times. Plots show the difference between NO densities calculated in two CMAT simulations, the first with horizontal and vertical advection included in the NO density calculation and the second without. Conditions are appropriate to northern spring equinox with an F10.7 of 105. Auroral inputs are appropriate to a Kp of 2 +. Contour intervals are every m 3. [31] The vertical, horizontal, and temporal extent of NO production and transport will be strongly dependent on the position and structure of the auroral oval. Variation in the distribution and magnitude of high-latitude energy inputs will influence the production and transport of species in and from that region. Joule heating and associated winds will depend on the nature and location of auroral energy inputs and as such so will the distribution of NO. The profiles shown here will therefore be highly variable with auroral energy inputs. A general appreciation of the extent of NO transport and horizontal distribution under conditions of moderate geomagnetic activity can however be gained from this simple study, and the importance of transport illustrated. The relative importance of each mechanism involved in aurorally induced midlatitude and low-latitude NO enhancements will be the subject of a future CMAT GCM study. 6. Study 2: Model Response to High Auroral Activity [32] The CMAT model study described in the following section was conducted in order to simulate global changes in the NO distribution that would arise from a period of high auroral energy influx. As in study 1 described above, the CMAT model was first run to a steady state condition over a period of 40 days using conditions appropriate to northern spring equinox (day 79). An F10.7 of 105 was used, along with statistical particle precipitation and high-latitude electric field patterns associated with a power index (PI) of 5 (Kp = 2 + ). This represents conditions of moderate auroral activity. From this starting point, a 6 day run was performed that used auroral energy inputs appropriate to a PI of 5 for the first 24 hours, after which the PI was increased linearly from 5 to 10, over a period of 1 hour. Activity level 10 is representative of high auroral activity. Whilst using this PI, the hemispheric power input was set to 125GW after Fuller- Rowell et al. [1994, 1996b], resulting in auroral energy inputs equivalent to a Kp of approximately 6. Auroral inputs remained at this level for 12 hours, after which the PI was relaxed back to level 5. This level of activity was then held constant at a PI of 5 for the remainder of the 6-day period. Table 1 summarizes the time history of auroral power inputs to the model as a function of hours from the start of the 6-day run. In order to give some indication of the 8of14

9 Table 1. Time History of Auroral Power Inputs to the Model as a Function of Hours From the Start of the 6-Day Simulation Hour PI Kp N S N S 0to to 25 5 to to 6 67 to to to to to 5 6 to to to to Oval Centroid Geomagnetic Latitude 9of14 Max Production Rate, cm 3 s 1 geomagnetic latitude at which maximum energy deposition occurs, the average position of the auroral oval centroid at 110 km altitude is given for the northern (N) and southern (S) hemispheres. Maximum production rates at this point are also given as an indication of energy input at each activity level. Note that CMAT model days start at 1200 UT so hour 0 to 1of the run occurs from 1200 to 1300 UT and so on. From the same starting point, a control simulation was also performed using auroral energy inputs characteristic of PI 5 throughout the full 6-day period. [33] The quasi-step variation of auroral power input utilized here is considerably simpler than the complex small-scale variations that occur in reality. The bulk energy inputs are however representative of those seen in storm periods. This approach is therefore sufficient to simulate the first order characteristics of geomagnetic storm response and has been employed in numerous other modeling studies [e.g., Fuller-Rowell et al., 1994; Fuller-Rowell et al., 1996b; Field et al., 1998; Fujiwara et al., 1996; Burns et al., 2004]. [34] Calculated nitric oxide number densities at 110 km as a function of time and geographic latitude are shown in Figure 5. The timescale refers to hours from the start of the 6-day model simulation, hour 0 representing 1200 UT on the first day. Profiles are shown for 0, 90, 180, and 270 geographic longitude. Clearly seen are high-latitude enhancements that increase and decrease with time as the high-latitude geographic grid points move through the auroral oval. The largest nitric oxide densities are seen in the southern hemisphere, between 40 S and 50 S geographic latitude ( 45 and 55 geomagnetic latitude), at a longitude of 180 where concentrations of approximately m 3 are predicted. This maximum appears approximately 42 hours after the end of the storm period (24 38 hours). The geographic latitude of the auroral oval centroid at this time is approximately 62 S ( 67 geomagnetic latitude), over 12 poleward of the NO density maximum. This suggests the NO maximum arises though processes other than direct chemical production. In the northern hemisphere, maximum densities of about m 3 occur between 50 and 60 geographic latitude (57 and 67 geomagnetic latitude) at a longitude of 270. This maximum occurs approximately 38 hours after the storm forcing has ended. The amount of NO present at any point on the globe is clearly highly dependent on the geographic location in relation to the region of increased auroral energy input. [35] Figure 6 shows difference plots of calculated NO density on a constant pressure surface (at approximately 110 km) derived from the simulation including 12 hours storm forcing, and the control simulation that used constant medium activity forcing throughout the 6-day period. Positive values indicate increased NO abundance as a result of storm forcing while shaded areas indicate a depletion in NO density. During the first 24 hours, auroral forcing in both runs was appropriate to a PI of 5. There is therefore no difference between the NO densities calculated in each simulation for the first day. Throughout the storm forcing period that follows, depletion in NO concentration at high latitudes is visible in both hemispheres. Contributing factors will be the equatorward expansion of the auroral oval and high-latitude winds associated with pressure gradients. Modeled meridional winds (not shown) for the simulation that included storm forcing show that at geographic latitudes poleward of approximately 60, winds of up to 75 m s 1 are present during and after the storm period. These enhanced winds are the result of steep pressure gradients created by high-latitude Joule heating. The greatest effect is seen in the dawn and dusk sectors where the winds act to blow NO rich air both toward lower latitudes and over the pole. At altitudes above 130 km, high-latitude winds can reach over 300 m s 1 and act to rapidly transport NO to middle and low latitudes. A pole to equator circulation is set up with upwelling at high latitudes and downwelling at low latitudes [Maeda et al., 1989]. [36] Middle latitudes undergo an increase in NO density as the auroral oval expands to lower latitudes. For example, at 90 longitude, the auroral oval centroid in the southern hemisphere moves from approximately 64 S ( 74 ) to 56 S ( 68 ) geographic (geomagnetic) latitude between hours 24 and 36 of the simulation. Excess NO resulting from the sudden influx of auroral energy starts to be transported equatorward by meridional winds, associated with high-latitude Joule heating. At latitudes equatorward of about 60, the direction of the wind is largely determined by pressure gradients associated with solar heating. During the night, midlatitude and low-latitude meridional winds blow equatorward at velocities of up to 40 m s 1. These winds can act to transport NO rich air approximately 1730 km in a 12-hour period, equivalent to approximately 16 latitude. This explains the rapid temporal change in NO density at midlatitude locations. [37] By the end of the storm forcing period (38 hours into the simulation), increased concentrations of NO are visible within 40 of the equator at 180 longitude. Over the following 36 to 48 hours, the region affected by aurorally produced NO extends to latitudes even closer to the equator. The timescales over which midlatitude and low-latitude NO enhancements are seen differ significantly with longitude, as does the spatial extent. At 0 longitude, aurorally produced NO is limited to geographic latitudes poleward of approximately 35 N and 45 S. The maximum enhancements appear around 30 hours after the storm forcing has ended. At 180 longitude, significant amounts of excess NO

10 Figure 5. Nitric oxide number density at approximately 110 km altitude as a function of time and latitude at 0, 90, 180, and 270 geographic longitude. The time axis refers to hours from the start of a CMAT run that includes high-activity auroral forcing between 2400 and 3800 hours. Auroral forcing before and after this period is characteristic of moderate geomagnetic activity. Conditions are appropriate to northern spring equinox with an F10.7 of 105. Contour intervals are every m 3. are present within 15 and 28 of the equator in the southern and northern hemispheres respectively. These maxima occur up to 40 hours after the end of the storm forcing period. In general, the midlatitude NO enhancements start to die away between 24 and 48 hours after storm forcing ends. At certain locations however, significant quantities of excess NO are present throughout the remainder of the 6 day period. Southern hemisphere enhancements of nearly m 3 are present at approximately 40 S, 180 longitude, 90 hours after relaxation of the storm. The studies of Siskind et al. [1989a] and Richards [2004] showed that significant amounts of NO are produced at altitudes above 140 km during periods of enhanced geomagnetic activity. Downward diffusion was found to lead to substantial increases in NO density below 120 km after approximately 1 day, continuing for several days after the disturbed period. Downward vertical diffusion, high-latitude circulation and direct horizontal transport will all contribute to the increased concentrations of modeled NO at middle and low latitudes over different timescales. [38] One factor that is likely to have an impact on both the timescale and magnitude of changes in high-latitude NO density following high-energy auroral inputs is the local time at which the auroral forcing occurs. If particle precipitation leads to the production of excess NO in the sunlit portion of the atmosphere, photodissociation by solar UV will act to destroy that NO. Barth et al. [2001] suggested the lifetime of an NO molecule to chemical destruction under illuminated conditions is 19 hours. This will be the case at 0 longitude in the simulation performed here. Conversely, if excess NO is created in the nighttime, as is the case for 180 longitude here, that excess NO will be longer lived and available for transport to lower latitudes. 10 of 14

11 Figure 6. Difference plots of nitric oxide number density at approximately 110 km altitude, at 0, 90, 180, and 270 geographic longitude. NO densities were calculated in two CMAT runs, the first with high-activity auroral forcing between 2400 and 3800 hours and moderate forcing for the remaining period and the second with constant moderate activity forcing throughout the full 6 days of the simulation. Conditions are appropriate to northern spring equinox with an F10.7 of 105. Contour intervals are every m 3. [39] Without knowledge of the spatial and temporal changes in atmospheric energy inputs it is extremely difficult to assess the lifetime of NO following production in, and transport from the auroral zones. While this is the first three dimensional modeling study of lower thermospheric NO transport, previous attempts to determine the relationship between NO abundance and geomagnetic activity have been made. Marsh et al. [2004] noted a 1 day lag in the maximum correlation between changes in observed highlatitude NO density and the daily average Kp index of geomagnetic activity. A similar delay was noted by Barth and Bailey [2004] in relation to the transport of aurorally produced NO to the equator following periods of intense geomagnetic activity. Both of these studies used zonal averages of data from the SNOE satellite, which comprises NO measurements from a single local time (approximately 1030 LT). Levels of geomagnetic activity were determined from daily averages of commonly used activity indices. While activity indicators such as daily Kp can give a good indication of the total auroral energy influx over a day, they cannot be used to describe the diurnal variation in highlatitude energy inputs. As previously mentioned, the number of hours for which aurorally produced NO is exposed to daylight can have an important effect on its lifetime and thus available transport time. Attempts to assess the effects of auroral forcing on NO distribution that do not take into account the spatial and temporal variability in global energy inputs will therefore be of limited accuracy. While the results presented here confirm that there is a lag in the maximum response of NO density to high-energy auroral 11 of 14

12 Table A1. CMAT Chemical Reaction Rates and Branching Ratios for Processes Involving the Production or Loss of NO, N( 2 D), or N( 4 S) Reaction Rate Coefficient Reference HO 2 +NO! NO 2 + OH exp(250/t) JPL97, DeMore et al. [1997] NO 2 +O! NO + O exp(120/t) JPL97, DeMore et al. [1997] NO + O 3! NO 2 +O exp( 1370/T) Roble [1995] N( 4 S) + NO! N 2 + O T > 400 K: Swaminathan et al. [1998] T < 400 K: exp(160/t) N( 2 D) + NO! N 2 + O Swaminathan et al. [1998] N( 4 S) + O 2! NO + O exp( 3600/T) Siskind et al. [2004] N( 2 D) + O 2! NO + O (T/300) Duff et al. [2003] N( 2 D) + O! N( 4 S) + O Bailey et al. [2002] N( 2 D) + O 3! NO + O T 1/2 exp( 1200/T) Banks and Kockarts [1973] N( 2 D)! N( 4 S) + hn s 1 Fuller-Rowell [1993] N( 4 S) + OH! NO + H Roble [1995] N 2 O+O( 1 D)! 2(NO) Roble [1995] NO + CLO! NO 2 + CL exp(294/t) Brasseur and Solomon [1986] O+N + 2! NO + +N( 2 D) T < 1500 K: (300/T R ) 0.44 Fuller-Rowell [1993] T > 1500 K: (T R /300) 0.2 NO + +e! O+N( 2 D), (300/T e ) 0.85 Bailey et al. [2002] N( 4 S) f(n( 2 D), N( 4 S)) = (0.85, 0.15) N + 2 +e! N( 2 D) + N( 4 S) (T e /300) 0.37 Guberman [1991] f(n( 2 D), N( 4 S)) = (1.12, 0.88) N + +O 2! O + + NO Swaminathan et al. [1998] N + +O 2! O + 2 +N( 2 D), Swaminathan et al. [1998] N( 4 S) f(n( 2 D), N( 4 S)) = (0.0, 1.0) Fuller-Rowell [1993] N + +O! O + +N( 4 S) Fuller-Rowell [1993] O + 2 +N( 4 S)! NO + + O Fuller-Rowell [1993] O + 2 +NO! NO + +O Fuller-Rowell [1993] N 2 +O +! NO + +N( 4 S) 300 K <T 2 < 1700 K: (T 2 /300) (T 2 /300) 2 Fuller-Rowell [1993] 1700 K <T 2 < 6000 K: (T 2 /300) (T 2 /300) 2 N( 2 D) + e! N( 4 S) + e (T e ) 0.81 Swaminathan et al. [1998] NO 2 +hn! NO + O s 1 Roble [1995] NO 3 +hn! NO + O s 1 Magnotta and Johnston [1980] N( 4 S) + hn! N + +e J N4S = s 1 Barth [1992] Nighttime J N4S =0. NO + hn! N( 4 S) + O J NO = exp[ 10 8 Nicolet [1979] (N O2 /cos(sza)) 0.38 ] exp[ N O3 ]s 1 Nighttime J NO =110 6 [daytime J NO at sec(sza)=1] NO + hn! NO + +e Ji NO =(Ly) Brasseur and Solomon [1986] exp[ N O2 /cos(sza)] s 1 Nighttime Ji NO =110 6 [daytime Ji NO at sec(sza)=1] forcing, they show for the first time that the lag is dependent on geographic location in relation to the auroral oval and possibly the local time at which auroral forcing occurs. [40] By zonally averaging the CMAT NO data presented in this study and isolating concentrations from a single local time of 1100 LT, the CMAT data can be presented in a format comparable with the zonally averaged SNOE dataset. When this is done, the maximum NO number densities are seen at high latitudes on day 3 of the simulation (not shown). The average PI on day 2 is 7, corresponding to a Kp of 3 +, whereas the average PI on day 3 of the simulation is 5, corresponding to a Kp of 2 +. It is therefore reasonable to conclude that when zonally averaged NO data is considered, at a single local time, the maximum response to changes in the daily Kp geomagnetic index occurs with a 1-day lag. At present, observational datasets of NO concentration do not have global coverage and it is only through using a 3-D GCM such as CMAT that detailed studies of the spatial and temporal distribution of aurorally produced NO can be performed. [41] Following periods of high geomagnetic activity, Barth and Bailey [2004] observed that NO enhancements in the lower thermosphere could remain at low latitudes for several days after the storm period. This is in keeping with the results of Richards [2004] and Siskind et al. [1989a]. The CMAT simulations performed here suggest that at equinox, auroral energy can have an effect on high-latitude and midlatitude NO concentrations for several days after relaxation of a storm. The lifetime of excess NO concentrations will be highly dependent on the number of hours in 12 of 14