JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, …dust.ess.uci.edu/ppr/ppr_BTJ07.pdf · 118 119...
Transcript of JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, …dust.ess.uci.edu/ppr/ppr_BTJ07.pdf · 118 119...
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 1 of 54
Numerical simulations of the three-dimensional distribution of meteoric dust in the 1
mesosphere and upper stratosphere 2
3
C. G. Bardeen1, O. B. Toon1, E. J. Jensen2, D. R. Marsh3, and V. L. Harvey4 4
5
1 Department of Atmospheric and Oceanic Sciences, Laboratory for Atmospheric and 6
Space Physics, University of Colorado, Boulder, Colorado, USA. 7
2 NASA Ames Research Center, Moffet Field, California, USA. 8
3 National Center for Atmospheric Research, Boulder, CO, USA 9
4 Laboratory for Atmospheric and Space Physics, University of Colorado, Boulder, 10
Colorado, USA. 11
12
13
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 2 of 54
Micrometeorites that ablate in the lower thermosphere and upper mesosphere are thought 13
to recondense into nanometer-sized smoke particles and then coagulate into larger dust 14
particles. Previous studies with one-dimensional models have determined that the 15
meteoric dust size distribution is sensitive to the background vertical velocity and have 16
speculated on the importance of the mesospheric meridional circulation to the dust spatial 17
distribution. We use a three-dimensional general circulation model with sectional 18
microphysics to study the distribution and characteristics of meteoric dust in the 19
mesosphere and upper stratosphere. We find that the mesospheric meridional circulation 20
causes a strong seasonal pattern in micrometeorite concentration in which the summer 21
pole is depleted and the winter pole is enhanced. This summer pole depletion of dust 22
particles results in fewer dust condensation nuclei (CN) than has traditionally been 23
assumed in numerical simulations of polar mesospheric clouds (PMCs). However, the 24
total number of dust particles present is still sufficient to account for PMCs if smaller 25
particles can nucleate to form ice than is conventionally assumed. During winter, dust is 26
quickly transported down to the stratosphere in the polar vortex where it may participate 27
in the nucleation of sulfate aerosols, the formation of the polar CN layer, and the 28
formation of polar stratospheric clouds (PSCs). Dust distributions in our model are 29
compared to results from the one-dimensional models, to observations of charged dust in 30
the mesosphere and to observations of meteoric dust content in stratospheric aerosols. 31
32
INDEX TERMS: 0305 (Aerosols and particles), 0340 (Middle atmosphere: composition 33
and chemistry), 0341 (Middle atmosphere: constituent transport and chemistry), 0545 34
(Modeling), 3332 (Mesospheric dynamics) 35
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 3 of 54
36
KEYWORDS: meteoric, dust, smoke, mesosphere, microphysics, model 37
38
39
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 4 of 54
1. Introduction 39
Estimates of the flux of meteoric material into the Earth’s atmosphere vary considerably 40
[Gabrielli, et al., 2004], but commonly used values are 16±5 kt yr-1 [Hughes, 1978] and 41
40±20 kt yr-1 [Love and Brownlee, 1993]. Most of these micrometeorites are thought to 42
ablate in the lower thermosphere and upper mesosphere [Hunten, et al., 1980; 43
Kalashnikova, et al., 2000], resulting in atomic metal layers observed by lidar between 80 44
and 110 km [Kane and Gardner, 1993; Plane, 2003]. The ablation products are likely to 45
recondense into nanometer-sized smoke particles [Rosinski and Snow, 1961], which are 46
extremely small and difficult to detect [Gelinas, et al., 1998; Havnes, et al., 2001; Lynch, 47
et al., 2005; Rapp, et al., 2005]. These small particles can coagulate into larger particles 48
[Hunten, et al., 1980] which may act as heterogeneous ice nuclei for polar mesospheric 49
clouds (PMC) [Rapp and Thomas, 2006; Turco, et al., 1982] and polar mesospheric 50
summer echoes (PMSE) [Rapp and Lubken, 2001]. 51
52
Murphy et al. [1998] and Cziczo et al. [2001] have observed meteoric dust in ~50% of 53
their samples of stratospheric aerosols. Meteoric dust may affect the size distribution and 54
composition of the sulfate aerosol layer and subsequently affect the nucleation of polar 55
stratospheric cloud (PSC) particles [Prather and Rodriguez, 1988]. Curtius et al. [2005] 56
observed an increased fraction of stratospheric aerosols with a non-volatile content inside 57
the Arctic polar vortex (67%) compared to outside the vortex (24%). They concluded that 58
meteoric dust is transported downward from the mesosphere inside the vortex and the 59
increased fraction is therefore an indicator of mesospheric air inside the polar vortex. 60
Meteoric dust descending in the polar vortex may also contribute to enhancements in the 61
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 5 of 54
polar condensation nuclei layer observed in the winter around 30 km [Hofmann, et al., 62
1988]. 63
64
Previous numerical simulations of meteoric dust formation and transport have been one-65
dimensional [Gabrielli, et al., 2004; Hunten, et al., 1980; Megner, et al., 2006]. Megner 66
et al. [2006] found that the size distribution of dust particles is sensitive to vertical 67
velocity; however, the overall impact of transport is difficult to evaluate from a one-68
dimensional simulation. Three-dimensional simulations of meteoric metals [Prather and 69
Rodriguez, 1988] and chemical tracers [Fischer and O'Neill, 1993; Plumb, et al., 2002] 70
have shown that the meridional mesospheric circulation will transport material from the 71
summer pole to the winter pole where it will be transported downward inside the polar 72
vortex. For this study, we created a three-dimensional general circulation model with 73
sectional microphysics (WACCM/CARMA) to investigate the combined impact of 74
transport and coagulation on the distribution of meteoric dust in the mesosphere and 75
upper stratosphere. 76
77
2. Model Description 78
WACCM/CARMA is a combination of the Whole-Atmosphere Community Climate 79
Model (WACCM) general circulation model [Garcia, et al., 2007] and a sectional 80
microphysics model based upon the Community Aerosol and Radiation Model for 81
Atmospheres (CARMA) [Toon, et al., 1988; Turco, et al., 1979]. As is shown in Figure 82
1, a single column version of CARMA is implemented as a WACCM physics package. 83
This means that the model’s state is passed to CARMA one column at a time, that 84
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 6 of 54
CARMA calculates tendencies on constituents within these columns, and that these 85
tendencies are then passed back to WACCM where they are used to adjust the model’s 86
state. Each CARMA size bin is added as a unique WACCM constituent. WACCM is 87
responsible for advection, vertical diffusion and wet deposition of all constituents, while 88
CARMA is responsible for the production, sedimentation and coagulation of the meteoric 89
dust constituents. 90
91
2.1 WACCM 92
We used WACCM3 version 3.1.9 tag 9 at 4˚x5˚ resolution with 66 vertical levels and a 93
model top near 150 km, which provides a vertical spacing of ~3.5 km in the mesosphere. 94
For one simulation, we used a vertical grid with 125 levels and ~0.5 km spacing near the 95
mesopause. WACCM’s vertical diffusion algorithm handles turbulence, gravity waves, 96
and molecular diffusion. The molecular diffusion code was designed for gases and is 97
unstable for the larger mass of the dust particles. Therefore it was disabled for the dust 98
particles. The default tuning of the gravity wave algorithm [Beres, et al., 2005; Garcia, et 99
al., 2007; Sassi, et al., 2002] is a compromise between generating a good climatology in 100
the stratosphere and the mesosphere. We conducted sensitivity tests with parameter 101
choices for the gravity wave parameterization that create a more realistic mesopause. 102
WACCM has water vapor and prescribed sulfate aerosols; however, in these experiments 103
we did not let the dust particles nucleate ice particles or interact with the sulfate aerosols. 104
Dust particles are removed in the troposphere via wet deposition using the existing 105
scheme [Barth, et al., 2000]. 106
107
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 7 of 54
2.2 CARMA 108
We used CARMA 2.3 configured for one-dimensional columns using the same vertical 109
grid as WACCM. Most runs used 21 size bins for dust particles with a mean radius from 110
1 to 100 nm and a doubling of particle volume between adjacent bins. Some runs used 28 111
size bins with radii from 0.2 to 100 nm. The initial particle size is a function of the 112
coagulation [Hunten, et al., 1980; Rosinski and Snow, 1961] and polymerization [Plane, 113
2003] that may occur in the region of the meteor trail. We base our cases on the 114
approximate size of an iron or silicon oxide molecule of 0.2 nm and a slightly larger size 115
of 1 nm that assumes more initial recombination. The density of the dust particles is 116
assumed to be 2.0 g cm-3 [Hunten, et al., 1980]. 117
118
The dust source function uses a production rate from Kalashnikova et al. [2000], which 119
has emissions from 75 to 110 km with a peak around 83 km (figure 2). All particles are 120
placed into the smallest size bin and the production rate is scaled over the area of the 121
Earth and the production altitude range to preserve the global mean mass influx at a rate 122
of ~16 kt yr-1. This global mass influx is based on the Kalashnikova et al. [2000] 123
production rate for particles with a radius of 1.3 nm and a density of 2.0 g cm-3 and is 124
similar to the rates measured by Hughes et al. [1978] and Gabrielli et al. [2004]. If 125
instead, the mass influx from Love and Brownlee [1993] were used, approximately twice 126
as many particles would be created. This would have a significant effect on the 127
coagulation rates and the resulting particle distributions. The dust source is assumed to be 128
temporally and horizontally uniform. Meteoric particles that do not ablate because of 129
their size are not included in the model. These particles represent about 20% of the total 130
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 8 of 54
mass [Kalashnikova, et al., 2000], are larger (5-50 µm radius) [Kalashnikova, et al., 131
2000] and are expected to sediment rapidly [Hunten, et al., 1980; Turco, et al., 1981]. 132
133
CARMA includes routines for sedimentation and coagulation of particles. The fall 134
velocities are calculated by assuming a Stokes flow with size corrections from Fuchs 135
[1964]. The vertical transport is solved using the piecewise parabolic method of Colella 136
and Woodward [1984]. No additional diffusion of aerosol particles is added by CARMA. 137
Coagulation coefficients are calculated to include Brownian, convective and gravitational 138
effects. A sticking coefficient of 1 is used, which means that particles are assumed to 139
stick together once they collide. The sticking coefficient and coagulation coefficients 140
could be affected by particle charging [Hunten, et al., 1980; Jensen and Thomas, 1991; 141
Reid, 1997], intermolecular forces like Van der Wals forces [Hunten, et al., 1980] and 142
magnetic dipole interactions [Saunders and Plane, 2006], which could either reduce or 143
enhance coagulation. Megner et al. [2006] identified coagulation rates as a significant 144
source of uncertainty in their meteoric dust model and it remains so in these simulations. 145
CARMA calculates the effect of coagulation using the numerical approach described in 146
Jacobsen et al. [1994]. 147
148
3. Results 149
We conducted 7 simulations including a control run with a 10-year duration and 6 150
sensitivity tests with 4-year durations. The definition of each simulation is shown in 151
Table 1. All simulations except simulation 7, the 125 level case, start with the same initial 152
condition and none of them start with any meteoric dust. The 125 level case has an initial 153
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 9 of 54
condition interpolated from a slightly different reference simulation; however, these 154
differences should not have an influence on the results. Figure 3 shows the total mass of 155
dust in the atmosphere as a function of height and time for the control run. A semiannual 156
pattern in dust mass exists in the mesosphere, while an annual pattern exists in the 157
stratosphere and troposphere. The model nearly reaches steady state in the entire 158
atmosphere after 10 years and reaches steady state above 30 km within 3 years (figure 4). 159
Since we are not modeling the interaction of dust with sulfate aerosols and water vapor, 160
we do not expect the model to properly represent the state of the dust particles in the 161
troposphere or lower stratosphere. Therefore, the analysis that follows is restricted to 162
heights above 30 km. In the control run, we have averaged the last 7 years of data, but in 163
the sensitivity tests, we only use the last year of the 4-year run. Comparisons between the 164
control run and the sensitivity tests are made using the fourth year of each simulation. 165
166
3.1 Control 167
Our control simulation uses the default WACCM configuration and an initial dust particle 168
radius of 1 nm. The simulation was run for 10 years and the average of the last 7 years for 169
January and July is shown in Figure 5. There is a large seasonal variation in dust 170
concentrations caused by the mesospheric circulation that goes from the summer pole to 171
the winter pole. These two seasons are the extremes cases, with a more uniform 172
distribution occurring in-between. It can also be seen that the mass density is highest in 173
the winter pole and increases toward the stratosphere. Figure 6 shows polar plots at 174
several altitudes for the month with the peak in the dust mass mixing ratio at that altitude. 175
Other than at 90 km, the dust is largely confined to the vortex. The descent through the 176
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 10 of 54
mesosphere takes about a month from April to May; however, it takes an additional four 177
months, out to September, to get the peak to 30 km. 178
179
To study the vortex in more detail, the vortex identification technique of Harvey et al. 180
[2002] was used to locate the vortex in the last year of the data. This technique calculates 181
a scalar quantity Q that is an indicator of rotation (negative) and strain (positive) in the 182
wind field and attempts to find streamlines around areas of cyclonic rotation where the 183
contour integral of Q is zero. Figure 7 shows the mass mixing ratio of all the meteoric 184
dust inside and outside of the vortex for both hemispheres. The seasonal transport of dust 185
between hemispheres because of the meridional circulation can be seen clearly both 186
inside and outside of the vortex between potential temperatures of 4000 and 5000 K. At 187
lower potential temperatures, dust can be seen descending into the stratosphere mostly 188
inside the vortex starting in the fall. Figure 8 shows the ratio of the dust mass mixing 189
ratio inside the vortex to that outside the vortex. This figure highlights that the dust is 190
well confined while it descends in the vortex, particularly in the southern hemisphere. In 191
the southern hemisphere, a vortex with enhanced dust concentrations persists throughout 192
the year between 500 and 1000 K. Curtius et al. [2005] measured the fraction of 193
stratospheric aerosols with a non-volatile component and found a ratio of 3:1 between the 194
inside and the outside of the vortex. Their measurements were made between 10 January 195
2003 and 19 March 2003 at potential temperatures of 300 to 500 K near Kiruna, Sweden. 196
While not measuring exactly the same thing, in Figure 8 we get a ratio ~3:1 at a similar 197
time and location. Figure 8 indicates that much larger dust gradients should be 198
measurable at other times and locations. 199
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 11 of 54
200
Hunten et al. [1980], did some of the fundamental modeling work on meteoric dust using 201
one-dimensional simulations with an earlier version of CARMA using conditions from 202
mid-latitude. In Figure 9, we compare our results for the 1 nm case to theirs presented in 203
their Figures 4 and 5. Hunten et al. [1980] significantly overestimates the concentration 204
and surface area density below ~75 km during the summer at mid to high latitudes 205
relative to our results. Our results also show an enhancement during the winter at high 206
latitudes. In Figure 10, we compare our size distributions at 30, 60 and 90 km to those of 207
Hunten et al. [1980]. At 90 km, we have fewer large particles than did Hunten et al. 208
[1980] in the summer because of the mesospheric circulation, but overall the distributions 209
are similar. At 60 km, the summer size distributions show a large depletion in small 210
particles, but a slight enhancement in large particles relative to Hunten et al. [1980] 211
because of ascending air in the region during the summer. At 30 km, the distribution is 212
shifted towards larger particles than Hunten et al. [1980] found because of the enhanced 213
dust concentrations from the mesospheric circulation and the confinement within the 214
polar vortex. 215
216
It is very difficult to measure meteoric dust particles, because of their small sizes and 217
their remote location. Several groups have attempted to detect charged dust particles with 218
rocket soundings. Figure 11 shows a comparison of our data from the control run and 219
simulation 7 with the dust observations from Lynch et al. [2005] made on 07 March 2002 220
and 15 March 2002 from Poker Flats, Alaska. Simulation 7 has a higher vertical 221
resolution than the control run and also has a smaller initial particle radius. To compare 222
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 12 of 54
their charged particle measurements to our particle concentrations, we have scaled up 223
their results assuming that only 5% of the total particles were charged and that the 224
particles they detected were ~1 nm in radius. We see good agreement in the overall shape 225
and approximate quantities, with the higher vertical resolution simulation showing a 226
better fit to the sharp gradients in the observations. Our results show more particles than 227
observed below 80 km, however the amount of free electrons is dropping in this region 228
making charging less efficient and the instrument’s measurements below 77 km are 229
suspect [Lynch, et al., 2005]. 230
231
3.2 Sedimentation 232
Figure 12 shows the results of simulation 2, which does not include the sedimentation of 233
dust particles. Sedimentation plays an important role in defining the sharp drop off in 234
particle concentration above 90 km, since particle fall velocities (Figure 13) are larger 235
than the vertical wind (Figure 14) in this region. Sedimentation also plays a role in the 236
descent of particles in the mesosphere and stratosphere. Without sedimentation, the 237
particles take longer to descend in the polar vortex, which leads to increased mass 238
densities in the mesosphere and decreased mass densities in the upper stratosphere. The 239
differences between the simulations are about 20% of the total mass density, so 240
sedimentation is not as important as advection for the transport of meteoric dust to the 241
stratosphere. The model assumes that the dust particles are spherical. If the larger 242
particles have a fluffier shape [Saunders and Plane, 2006], then the fall velocities will be 243
reduced and sedimentation may play a lesser role in the lower mesosphere and 244
stratosphere than indicated by these results. 245
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 13 of 54
246
3.3 Coagulation 247
The results of simulation 3, which does not include coagulation, are shown in Figure 15 248
and are similar to the results from simulation 2 (Figure 12) in the mesosphere and 249
stratosphere. This result indicates that it is the sedimentation of large particles formed by 250
coagulation that is important in this region, while it is the sedimentation of small particles 251
that is important for defining the sharp gradient in number concentration near 90 km 252
(Figure 5). Coagulation is enhanced in the Polar Regions because of the increased 253
concentration of dust from the meridional circulation and the confinement of dust in the 254
polar vortex. We assume a sticking coefficient of 1.0 in the model; however, particle 255
charging in the lower thermosphere and upper mesosphere [Jensen and Thomas, 1991; 256
Reid, 1997] and dipole-dipole interactions [Saunders and Plane, 2006] could impact the 257
choice of the sticking coefficient and the coagulation kernels. Further study is needed to 258
reduce the uncertainty in the effect of the coagulation process upon the distribution of 259
meteoric dust particles. 260
261
3.4 Smaller Initial Particle Size 262
Lacking observations, there is uncertainty as to what the particle size will be after the 263
ablation and initial recombination in the meteor trail. Hunten et al. [1980] used sizes 264
ranging from 0.2 nm to 10 nm, with the smallest size being the approximate radius of an 265
iron oxide molecule [Rosinski and Snow, 1961]. Figure 16 shows the results of simulation 266
4, which assumes an initial particle radius of 0.2 nm and a concentration production rate 267
that scales with radius to conserve the total mass being produced at ~16 kt yr-1. The 268
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 14 of 54
increase in the small particle production rate causes an increase in particle concentration 269
near the source region. Because the transport of dust away from the summer pole by the 270
meridional circulation is relatively rapid, coagulation cannot quickly create 1 nm particles 271
so there is a reduction in the number of particles 1 nm or larger in the summer mesopause 272
region relative to the 1 nm case. The reduction in the effective radius causes a slower 273
descent of mass into the stratosphere leaving increased mass densities in the mesosphere. 274
The size distribution and number concentrations are sensitive to the initial size of the 275
particles as they emerge from the meteor trail. Comparisons of vertical profiles of dust 276
concentration and surface area density and size distributions to Hunten et al. [1980] are 277
shown in Figure 17 and Figure 18, with differences similar to those seen in the control 278
run. Attempts to measure dust particles in the mesosphere suggest that initial sizes less 279
than 1 nm and closer to 0.2 nm are most likely (John Plane, personal communication, 280
2007); however, additional measurements would help constrain the model. 281
282
3.5 Gravity Waves 283
Gravity waves generated in the troposphere are filtered by the mesosphere depositing 284
momentum, which affects the circulation in the mesopause region and alters the 285
mesopause height and temperature [Fritts and Alexander, 2003; Lindzen, 1981]. 286
WACCM does not resolve these gravity waves, so a gravity wave parameterization is 287
used that is tuned to provide a reasonable climatology in both the stratosphere and 288
mesosphere [Beres, et al., 2005; Garcia, et al., 2007; Sassi, et al., 2002]. The summer 289
mesopause created in WACCM by the default tuning is lower and warmer than 290
observations [Lubken, 1999; Lubken, et al., 2004]. In simulations 5 and 6, we used a 291
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 15 of 54
smaller shear stress for the background gravity waves τb*, which results in the waves 292
breaking at a higher altitude. When the waves break at a higher altitude, the mesopause 293
height increases, the mesopause temperature decreases and the winds intensify (Figure 294
14). Intensified winds result in a decrease in dust concentration near the summer 295
mesopause, but cause an increase in dust concentration in the mesosphere inside the polar 296
vortex (Figure 19). With increased shear stress, the pattern of vertical winds in the winter 297
pole is shifted to a higher altitude, resulting in an increased descent of dust above 60 km 298
and a decreased descent of dust below 60 km at the winter pole relative to the normal 299
gravity wave tuning. The impact of changes to the gravity wave parameterization upon 300
the overall climatology of the model has not been studied; however, the distribution of 301
meteoric dust is sensitive to the gravity wave parameterization. 302
303
3.6 Dust Near The Summer Mesopause 304
Meteoric dust is one of the leading candidates to be the source of condensation nuclei 305
(CN) needed for the formation of PMCs [Rapp and Thomas, 2006; Turco, et al., 1982]. 306
Turco et al. [1982] indicate that concentrations of several hundred particles cm-3 with a 307
radius of at least 1 to 1.5 nm are necessary to form PMCs under typical summer 308
mesopause conditions. However, this requirement rests on numerous assumptions and is 309
not based upon observations. Large temperature variations driven by gravity waves could 310
permit ice nucleation on the more numerous smaller dust particles. A sufficient 311
concentration of large particles are present in the control run to nucleate under the 312
traditional assumptions, but reductions in the number concentration of large particles are 313
seen when starting with a smaller initial radius and when the gravity wave 314
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 16 of 54
parameterization is tuned to make a more realistic mesopause. In simulation 7, we 315
investigate the availability of meteoric dust CN by using a higher vertical resolution that 316
provides ~0.5 km spacing near the mesopause, an initial dust particle radius of 0.2 nm, 317
and a background shear stress of 0.002 N m-2. Our results from simulation 7 show that the 318
dust concentrations of particles 1 nm and larger are 15 to 30 cm-3 in the summer 319
mesopause region with maximum values above the temperature minimum and very few 320
particles below the mesopause (Figure 20). Figure 21 shows that these patterns are fairly 321
symmetric and centered near the pole during the summer in the northern hemisphere. The 322
dust particle size distribution at several levels near the summer polar mesopause from 323
simulation 7 is compared to Hunten et al. [1980] in Figure 22. There is a significant 324
depletion of particles at high latitude in the summer relative to Hunten et al. [1980]. 325
However, the ascending winds do appear to loft a small amount of larger particles. The 326
meteoric dust concentration, mass density and size distribution are compared to Megner 327
et al. [2006] in Figure 23 and Figure 24. The summer depletion and winter enhancement 328
in concentration and mass density are not as large as those seen in Megner et al. [2006], 329
particularly for particles that are 1 nm and larger in the summer. This simulation shows 330
that size distributions like Hunten et al. [1980] often used in PMC models, significantly 331
overstate the number of 1 nm and larger particles available and that particle 332
concentrations are extremely depleted below the mesopause. However, summer 333
conditions are not as depleted as indicated by Megner et al. [2006]. If meteoric dust is a 334
CN for PMC particles, nucleation is more likely to occur at or above the mesopause and 335
nucleation may require higher supersaturations than is typically assumed. It is also 336
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 17 of 54
possible that the dust may interact with other gases and aerosols to create larger particles 337
[Mills, et al., 2005]. 338
339
4. Summary and Future Work 340
All simulations showed a strong seasonal pattern with meteoric dust particles being 341
transported horizontally from the summer pole to the winter pole where they are quickly 342
transported down into the stratosphere within the polar vortex. The increased 343
concentration of dust particles in the vortex creates larger particles in the stratosphere 344
than previously suggested by one-dimensional simulations and may make meteoric dust 345
important to the polar CN layer and the sulfate aerosol layer. Any impact on the 346
characteristics of sulfate aerosols in the winter stratosphere could also affect PSC 347
nucleation. 348
349
The calculated depletion of particles in the summer mesopause region may reduce the 350
concentration of large particles near the mesopause below the number traditionally 351
thought to be needed for PMC nucleation; although, the amount and location of the 352
depletion depends on the initial particle radius and the gravity wave parameterization. 353
Uncertainties in the coagulation efficiency could also have an impact on the number of 354
larger particles available. Additional dust observations and laboratory studies would be 355
useful to constrain the model. For studies of the summer mesopause, additional tuning 356
and validation of the WACCM gravity wave parameterization is necessary. We have 357
constructed a PMC model using WACCM/CARMA and the dust source described here, 358
which we plan to use to evaluate the potential for meteoric dust particles to be a 359
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 18 of 54
significant source of condensation nuclei for PMCs. We intend to compare the model 360
results to observations from the recently launched Aeronomy of Ice in the Mesosphere 361
satellite mission, as well as other PMC climatologies. 362
363
Based upon radar observations, Janches et al. [2006] have suggested that there are 364
temporal and spatial variations in the meteoric influx. Some of this variation may get 365
averaged out as the dust is accumulated and transported; however, it may be important in 366
areas where dust is depleted such as the summer mesopause. We plan to use a dust source 367
function derived from Janches et al. [2006] and Fentzke and Janches (manuscript in 368
preparation, 2007) to investigate the sensitivity of the dust distribution to temporal and 369
spatial variations in the source function. 370
371
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 19 of 54
Acknowledgements 371
The authors thank Guy Brasseur, Rolando Garcia, Doug Kinnison, Francis Vitt and Stacy 372
Walters of the National Center for Atmospheric Research for providing us with and 373
supporting our use of the WACCM model. 374
375
Support for Charles Bardeen was partially provided by a NASA Earth System Science 376
Fellowship, grant No. NNG05GQ75J. Additional support for this project came from 377
Aura grant No. NNG06GE80G and NSF grant No. ATM0435713. 378
379
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 20 of 54
References 379 380
Barth, M. C., P. J. Rasch, J. T. Kiehl, C. M. Benkovitz, and S. E. Schwartz (2000), Sulfur 381 chemistry in the National Center for Atmospheric Research Community Climate Model: 382 Description, evaluation, features, and sensitivity to aqueous chemistry, J. Geophys. Res., 383 105, 1387-1415. 384
Beres, J. H., R. R. Garcia, B. A. Boville, and F. Sassi (2005), Implementation of a gravity 385 wave source spectrum parameterization dependent on the properties of convection in the 386 Whole Atmosphere Community Climate Model (WACCM), J. Geophys. Res., DOI: 387 10.1029/2004JD005504. 388
Colella, P., and P. R. Woodward (1984), The Piecewise Parabolic Method (PPM) for 389 Gas-Dynamical Simulations, J. Comput. Phys., 54, 174-201. 390
Curtius, J., et al. (2005), Observations of meteoric material and implications for aerosol 391 nucleation in the winter Arctic lower stratosphere derived from in situ particle 392 measurements, Atmos. Chem. Phys., 5, 3053-3069. 393 Cziczo, D. J., et al. (2001), Meteoritic and organic material in the atmosphere, Abstr. 394 Pap. Am. Chem. Soc., 221, U543-U543. 395 Fischer, M., and A. O'Neill (1993), Rapid descent of mesospheric air into the 396 stratospheric polar vortex, Geophys. Res. Lett., 20, 1267-1270. 397 Fritts, D. C., and M. J. Alexander (2003), Gravity wave dynamics and effects in the 398 middle atmosphere, Rev. Geophys., DOI: 10.1029/2001RG000106. 399 Fuchs, N. A. (1964), The Mechanics of Aerosols, Pergamon, New York. 400
Gabrielli, P., et al. (2004), Meteoric smoke fallout over the Holocene epoch revealed by 401 iridium and platinum in Greenland ice, Nature, 432, 1011-1014. 402
Garcia, R. R., D. R. Marsh, D. E. Kinnison, B. A. Boville, and F. Sassi (2007), 403 Simulation of secular trends in the middle atmosphere, 1950-2003, J. Geophys. Res., 404 DOI: 10.1029/2006JD007485. 405 Gelinas, L. J., et al. (2005), Mesospheric charged dust layer: Implications for neutral 406 chemistry, J. Geophys. Res., DOI: 10.1029/2004JA010503. 407 Gelinas, L. J., et al. (1998), First observation of meteoritic charged dust in the tropical 408 mesosphere, Geophys. Res. Lett., 25, 4047-4050. 409 Harvey, V. L., R. B. Pierce, T. D. Fairlie, and M. H. Hitchman (2002), A climatology of 410 stratospheric polar vortices and anticyclones, J. Geophys. Res., DOI: 411 10.1029/2001JD001471. 412
Havnes, O., T. Aslaksen, and A. Brattli (2001), Charged dust in the Earth's middle 413 atmosphere, Phys. Scr., T89, 133-137. 414
Hofmann, D. J., J. M. Rosen, and J. W. Harder (1988), Aerosol Measurements in the 415 Winter/Spring Antarctic Stratosphere: 1. Correlative Measurements with Ozone, J. 416 Geophys. Res., 93, 665-676. 417
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 21 of 54
Hughes, D. W. (1978), Meteors, in Cosmic Dust, edited by J. A. M. McDonnel, pp. 123-418 185, Wiley, London. 419
Hunten, D. M., R. P. Turco, and O. B. Toon (1980), Smoke and Dust Particles of 420 Meteoric Origin in the Mesosphere and Stratosphere, J. Atmos. Sci., 37, 1342-1357. 421
Jacobson, M. Z., R. P. Turco, E. J. Jensen, and O. B. Toon (1994), Modeling Coagulation 422 among Particles of Different Composition and Size, Atmos. Environ., 28, 1327-1338. 423
Janches, D., C. J. Heinselman, J. L. Chau, A. Chandran, and R. Woodman (2006), 424 Modeling the global micrometeor input function in the upper atmosphere observed by 425 high power and large aperture radars, J. Geophys. Res., DOI: 10.1029/2006JA011628. 426 Jensen, E. J., and G. E. Thomas (1991), Charging of Mesospheric Particles: Implications 427 for Electron Density and Particle Coagulation, J. Geophys. Res., 96, 18,603-618,615. 428 Kalashnikova, O., M. Horanyi, G. E. Thomas, and O. B. Toon (2000), Meteoric smoke 429 production in the atmosphere, Geophys. Res. Lett., 27, 3293-3296. 430 Kane, T. J., and C. S. Gardner (1993), Lidar Observations of the Meteoric Deposition of 431 Mesospheric Metals, Science, 259, 1297-1300. 432 Lindzen, R. S. (1981), Turbulence and Stress Owing to Gravity-Wave and Tidal 433 Breakdown, J. Geophys. Res., 86, 9707-9714. 434 Love, S. G., and D. E. Brownlee (1993), A Direct Measurement of the Terrestrial Mass 435 Accretion Rate of Cosmic Dust, Science, 262, 550-553. 436 Lubken, F. J. (1999), Thermal structure of the Arctic summer mesosphere, J. Geophys. 437 Res., 104, 9135-9149. 438 Lubken, F. J., A. Mullemann, and M. J. Jarvis (2004), Temperatures and horizontal winds 439 in the Antarctic summer mesosphere, J. Geophys. Res., DOI: 10.1029/2004JD005133. 440 Lynch, K. A., et al. (2005), Multiple sounding rocket observations of charged dust in the 441 polar winter mesosphere, J. Geophys. Res., DOI: 10.1029/2004JA010502. 442 Megner, L., M. Rapp, and J. Gumbel (2006), Distribution of meteoric smoke - sensitivity 443 to microphysical properties and atmospheric conditions, Atmos. Chem. Phys., 6, 4415-444 4426. 445
Mills, M. J., O. B. Toon, and G. E. Thomas (2005), Mesospheric sulfate aerosol layer, J. 446 Geophys. Res., DOI: 10.1029/2005JD006242. 447
Murphy, D. M., D. S. Thomson, and T. M. J. Mahoney (1998), In situ measurements of 448 organics, meteoritic material, mercury, and other elements in aerosols at 5 to 19 449 kilometers, Science, 282, 1664-1669. 450 Plane, J. M. C. (2003), Atmospheric chemistry of meteoric metals, Chem. Rev., 103, 451 4963-4984. 452 Plumb, R. A., et al. (2002), Global tracer modeling during SOLVE: High-latitude descent 453 and mixing, J. Geophys. Res., DOI: 10.1029/2001JD001023. 454 Prather, M. J., and J. M. Rodriguez (1988), Antarctic Ozone - Meteoric Control of Hno3, 455 Geophys. Res. Lett., 15, 1-4. 456
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 22 of 54
Rapp, M., et al. (2005), Observations of positively charged nanoparticles in the nighttime 457 polar mesosphere, Geophys. Res. Lett., DOI: 10.1029/2005GL024676. 458
Rapp, M., and F. J. Lubken (2001), Modeling of particle charging in the polar summer 459 mesosphere: Part 1 - General results, J. Atmos. Sol. Terr. Phys., 63, 759-770. 460
Rapp, M., and G. E. Thomas (2006), Modeling the microphysics of mesospheric ice 461 particles: Assessment of current capabilities and basic sensitivities, J. Atmos. Sol. Terr. 462 Phys., DOI: 10.1016/j.jastp.2005.10.015. 463 Reid, G. C. (1997), On the influence of electrostatic charging on coagulation of dust and 464 ice particles in the upper mesosphere, Geophys. Res. Lett., 24, 1095-1098. 465 Rosinski, J., and R. H. Snow (1961), Secondary Particulate Matter from Meteor Vapors, 466 J. Met., 18, 736-745. 467 Sassi, F., R. R. Garcia, B. A. Boville, and H. Liu (2002), On temperature inversions and 468 the mesospheric surf zone, J. Geophys. Res., DOI: 10.1029/2001JD001525. 469 Saunders, R. W., and J. M. C. Plane (2006), A laboratory study of meteor smoke 470 analogues: Composition, optical properties and growth kinetics, J. Atmos. Sol. Terr. 471 Phys., 68, 2182-2202. 472
Toon, O. B., R. P. Turco, D. Westphal, R. Malone, and M. S. Liu (1988), A 473 Multidimensional Model for Aerosols - Description of Computational Analogs, J. Atmos. 474 Sci., 45, 2123-2143. 475 Turco, R. P., P. Hamill, O. B. Toon, R. C. Whitten, and C. S. Kiang (1979), One-476 Dimensional Model Describing Aerosol Formation and Evolution in the Stratosphere .1. 477 Physical Processes and Mathematical Analogs, J. Atmos. Sci., 36, 699-717. 478
Turco, R. P., O. B. Toon, P. Hamill, and R. C. Whitten (1981), Effects of Meteoric 479 Debris on Stratospheric Aerosols and Gases, J. Geophys. Res., 86, 1113-1128. 480
Turco, R. P., O. B. Toon, R. C. Whitten, R. G. Keesee, and D. Hollenbach (1982), 481 Noctilucent Clouds - Simulation Studies of Their Genesis, Properties and Global 482 Influences, Planet. Space Sci., 30, 1147-1181. 483 484
485
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 23 of 54
Figure Captions 485 Figure 1. CARMA sectional microphysics is implemented as an interactive “physics 486
package” in WACCM. This means that CARMA processes the state for one WACCM 487
column at a time, providing tendencies on dust particles back to the parent model. The 488
standard run defines 21 dust size bins from 1 to 100 nm, which are all added as 489
constituent in WACCM. WACCM is responsible for the advection, diffusion and wet 490
deposition of these particles, while CARMA is responsible for the source function, 491
sedimentation and coagulation. 492
Figure 2. The production rates of meteoric dust particles from Kalashnikova et al. [2000] 493
(red) and Hunten et al. [1980] (green) as a function of pressure. Both of these curves 494
assume all the particles have a 1.3 nm radius. The black lines indicate that the source 495
algorithm reproduces the Kalashnikova et al. [2000] production rate (dotted) and indicate 496
how the rate is scaled to preserve mass for the 1 nm radius (dashed) and 0.2 nm radius 497
(solid) cases. 498
Figure 3. Evolution of the vertical distribution of the mass of meteoric dust in the 499
atmosphere during the 10 years of the control simulation (left) and larger view of the last 500
2 years (right). At high altitudes, dust descending in the winter pole creates a semiannual 501
pattern in dust distribution, while at low altitudes there is an annual pattern with 502
enhanced loss during the southern hemisphere winter and spring. Steady state takes the 503
longest time to attain in the lower stratosphere. 504
Figure 4. The atmosphere does not quite reach steady state during the 10 years of the 505
control simulation as indicated by the total mass of meteoric dust (solid line). The 506
atmosphere above 30 km (dotted line) reaches steady state within 3 years, while after 10 507
years the model is only just reaching steady state below 30 km (dashed line). There is a 508
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 24 of 54
seasonal cycle in the total mass, which is likely caused by enhanced loss below 30 km in 509
the polar vortex during the southern hemisphere winter. 510
Figure 5. Zonal average plots for January (top) and July (bottom) of the meteoric dust 511
particle concentration, mass density, and effective radius for the control simulation. The 512
data is an average of the last 7 years of the 10-year simulation. A strong seasonal 513
variation is evident with particles being advected from the summer to winter pole by the 514
mesospheric circulation. There is also rapid descent through the mesosphere at the winter 515
pole. The peak in effective radius at low latitudes and altitudes is because the circulation 516
ascends from the summer pole to the low latitudes in the winter hemisphere and thus only 517
large particles with higher sedimentation velocities will be able to descend in the tropics. 518
Figure 6. Polar plots of the mass mixing ratio of meteoric dust in the southern polar 519
region showing the months of peak concentrations at the selected altitudes as the dust 520
descends in the polar vortex. The white arrows are the monthly averaged horizontal wind 521
vectors. The monthly averaged dust mass mixing ratio contours have been scaled 522
differently for each month to emphasize the structure. 523
Figure 7. The annual variation in the vertical profile of the meteoric dust mass mixing 524
ratio inside (left) and outside (right) the polar vortex for the northern (top) and southern 525
(bottom) hemispheres. The dust descends in the polar vortex rapidly through the 526
mesosphere and more gradually in the stratosphere. The peaks near the mesopause in the 527
winter to spring time frame are because of the mesospheric meridional circulation and are 528
not confined to the vortex. 529
Figure 8. The ratio of the average mass mixing ratio of meteoric dust inside the polar 530
vortex to that outside the vortex for the Northern Hemisphere (left) and Southern 531
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 25 of 54
Hemisphere (right). The descent of dust within the vortex is clearly visible. The vortex is 532
particularly isolated in the Southern Hemisphere winter from 700 to 2000K. 533
Figure 9. Comparison of the average vertical profile of concentration and surface area 534
density of meteoric dust between Hunten et al. [1980] (thick green line) and the control 535
simulation (thin lines). The color indicates the latitude range and the line style indicates 536
the time period being averaged. The left panels are the annual averages and the right 537
panels are the JJA (dotted) and DJF (dashed) seasonal averages. Hunten et al. [1980] is a 538
mid-latitude model; however, the mid-latitude annual averages from the control run (red 539
and blue lines) are significantly reduced below ~75 km in both concentration and surface 540
area density relative to Hunten et al. [1980]. The summer season is significantly depleted 541
and winter season is enhanced in concentration and surface area density at mid to high 542
latitudes. 543
Figure 10. Comparison of the average size distribution of meteoric dust at selected 544
altitudes between Hunten et al. [1980] (thick green line) and the control simulation (thin 545
lines). The color indicates the latitude range and the line style indicates the time period 546
being averaged. The top panels are the annual averages and the bottom panels are the JJA 547
(dotted) and DJF (dashed) seasonal averages. The small sizes are depleted during the mid 548
and high latitude summer at 60 km and 30 km. The large sizes are enhanced at 30 km 549
because of increased coagulation in the polar vortex. 550
Figure 11. Comparison of particle concentrations between the control simulation and 551
rocket soundings by Lynch et al. [2005] from Poker Flats, Alaska. Their observation of 552
charged particles has been converted into dust particles by assuming that 5% of the dust 553
particles are charged and that the particle radius is ~1 nm [Gelinas, et al., 2005; Lynch, et 554
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 26 of 54
al., 2005]. The black line is the average of the last 7 years of the control run taken from 555
the 4˚x5˚model grid point containing Poker Flats using a daily average for the date 556
indicated assuming 1 nm particles. The blue line is sampled in a similar way from 557
simulation 7, and assumes the particle radius is 0.8 nm to 1.3 nm. The vertical resolution 558
in the control run near the mesopause is ~3.5 km, while simulation 7 has an increased 559
resolution of ~0.5 km. The points on the solid lines indicate the midpoint altitudes of the 560
model levels. The dashed lines are the observations. Both simulations do a good job of 561
matching the observations; however, simulation 7 with a higher vertical resolution that 562
supports sharper gradients is a better match at lower and higher altitudes. 563
Figure 12. Zonal average plots for July of the concentration and mass density for 564
simulation 2, the no sedimentation case. The difference between these results and the 565
control simulation are shown in the right panels. Sedimentation is important in 566
establishing the boundary near 90 km of high gradients in dust concentrations. It plays a 567
lesser role in transporting dust into the stratosphere. 568
Figure 13. Meteoric dust fall velocities from CARMA as a function of altitude for each 569
particle bin from 0.2 nm (left) to 100 nm (right). Velocities for mean radii of 1 nm and 10 570
nm are highlighted (bold lines). 571
Figure 14. Zonal average plots for July of the fourth year of the simulation for the 572
temperature and the vertical wind in the control simulation and simulations 5 and 6, 573
which use different values for the gravity wave parameterization. The black lines are 574
temperature contours from 130 to 160 K. A smaller background shear stress (τb*) causes 575
the waves to break at higher altitudes. Waves breaking at higher altitudes raise and cool 576
the summer mesopause and increase the downdrafts in the winter polar vortex. 577
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 27 of 54
Figure 15. Zonal average plots for July of the mass density for simulation 3, the no 578
coagulation case. The difference between this result and the control simulation are shown 579
in the right panel. Below 80 km, this simulation is similar to the no sedimentation case 580
shown in Figure 12, since there are no large particles created and thus sedimentation 581
velocities are low. 582
Figure 16. Zonal average plots for July of the concentration of particles with 1 nm or 583
larger radius, mass density, and effective radius for simulation 4, the 0.2 nm initial radius 584
case. The difference between these results and the control simulation are shown in the 585
right panels. Starting with a smaller initial radius reduces the number of larger particles 586
formed by coagulation and therefore slightly reduces the sedimentation into the 587
stratosphere. The effective radius in the summer near 80 km is also significantly reduced 588
relative to the control run. 589
Figure 17. Comparison of the average vertical profile of concentration and surface area 590
density of meteoric dust between Hunten et al. [1980] (thick green line) and simulation 4, 591
the 0.2 nm initial radius case (thin lines). The color indicates the latitude range and the 592
line style indicates the time period being averaged. The left panels are the annual 593
averages and the right panels are the JJA (dotted) and DJF (dashed) seasonal averages. 594
Hunten et al. [1980] is a mid-latitude model; however, the mid-latitude annual averages 595
from the control run (red and blue lines) below ~75 km are significantly reduced in both 596
concentration and surface area density relative to Hunten et al. [1980]. The summer 597
season is significantly depleted and winter season is enhanced in concentration and 598
surface area density at mid to high latitudes. 599
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 28 of 54
Figure 18. Comparison of the average size distribution of meteoric dust at selected 600
altitudes between Hunten et al. [1980] (thick green line) and simulation 4, the 0.2 nm 601
initial radius case (thin lines). The color indicates the latitude range and the line style 602
indicates the time period being averaged. The top panels are the annual averages and the 603
bottom panels are the JJA (dotted) and DJF (dashed) seasonal averages. The small sizes 604
are depleted during the mid and high latitude summer at 30 km and 60 km because of the 605
mesospheric circulation. The large sizes are enhanced at 30 km because of increased 606
coagulation in the polar vortex. 607
Figure 19. Zonal average plots for July of the fourth year of the simulation for the 608
concentration and mass density of dust particles in the control simulation and simulations 609
5 and 6, which use different values for the gravity wave parameterization. The black lines 610
are temperature contours from 130 to 160 K. The increased meridional wind relative to 611
the control run causes a reduced concentration near the summer mesopause. Increased 612
vertical winds relative to the control run cause more rapid transport in the winter polar 613
vortex above 60 km. Below 60 km, decreased vertical winds relative to the control run 614
slow transport into the stratosphere. 615
Figure 20. Zonal average plots for January and July of the concentration of particles with 616
1 nm or larger radius for simulation 7, which has enhanced vertical resolution and a more 617
realistic mesopause thermal structure than the control run. The black lines are 618
temperature contours from 130 to 160 K. Particle concentrations near the summer 619
mesopause are reduced compared to other simulations. 620
Figure 21. Polar plots of the concentration of dust particles with a radius of 1 nm or 621
larger during the month of July in the northern hemisphere for simulation 7 at various 622
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 29 of 54
altitudes. There are widespread areas with large depletions relative to Hunten et al. 623
[1980] and to our other simulations, particularly below the mesopause. 624
Figure 22. Comparison of the average size distribution of meteoric dust during the 625
summer at high latitude between Hunten et al. [1980] (green line) at 90 km and 626
simulation 7 (black lines) at various altitudes near the mesopause. There is a significant 627
depletion in the concentration of smaller particles at all altitudes relative to Hunten et al. 628
[1980]. There is a small increase in the concentration of large particles due to particles 629
that are transported up from lower altitudes. 630
Figure 23. Comparison of the average vertical profile of concentration and mass density 631
of meteoric dust during the summer (solid) and winter (dashed) between Megner et al. 632
[2006] (green) and simulation 7 at high (purple) and mid (blue) latitude in the Northern 633
Hemisphere. Our model shows significantly less depletion in the summer and slightly 634
less enhancement in the winter relative to Megner et al. [2006], who used extreme 635
vertical velocities from CHEM2D for their winter and summer wind profiles. The 636
concentration of particles 1 nm and larger and the mass density in the summer in our 637
results are also influenced by the uplift of residual particles from the previous winter. 638
Figure 24. Comparison of the average size distribution of meteoric dust between 80 km 639
and 90 km during the summer (solid) and winter (dashed) between Megner et al. [2006] 640
(green) and simulation 7 at high (purple) and mid (blue) latitude in the Northern 641
Hemisphere. There are more large particles relative to Megner et al. [2006], particularly 642
in the summer. 643
644
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 30 of 54
Tables 644
Simulation Duration
(years)
Particle
Radius
(nm)
Levels τb*
(Pa)
Microphysical Processes
1 (Control) 10 1-100 66 0.006 all
2 4 1-100 66 0.006 coagulation, wet deposition
3 4 1-100 66 0.006 sedimentation, wet deposition
4 4 0.2-100 66 0.006 all
5 4 1-100 66 0.003 all
6 4 1-100 66 0.002 all
7 4 0.2-100 125 0.002 all
Table 1. Definitions of the control run and sensitivity tests. All of the simulations include 645
the Kalashnikova et al. [2000] source function applied to the smallest size bin and wet 646
deposition based on Barth et al. [2000]. Most simulations include the microphysical 647
processes of coagulation, sedimentation and wet deposition (all). The background gravity 648
wave shear stress is controlled by the τb* parameter in the WACCM gravity wave 649
parameterization [Garcia, et al., 2007; Sassi, et al., 2002]. The default for τb* is 0.006 Pa. 650
Smaller values cause the gravity waves to break at higher altitudes resulting in a higher 651
and colder mesopause. 652
653
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 31 of 54
Figures 653
654 Figure 2. CARMA sectional microphysics is implemented as an interactive “physics 655 package” in WACCM. This means that CARMA processes the state for one WACCM 656 column at a time, providing tendencies on dust particles back to the parent model. The 657 standard run defines 21 dust size bins from 1 to 100 nm, which are all added as 658 constituent in WACCM. WACCM is responsible for the advection, diffusion and wet 659 deposition of these particles, while CARMA is responsible for the source function, 660 sedimentation and coagulation. 661
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 32 of 54
662 Figure 2. The production rates of meteoric dust particles from Kalashnikova et al. [2000] 663 (red) and Hunten et al. [1980] (green) as a function of pressure. Both of these curves 664 assume all the particles have a 1.3 nm radius. The black lines indicate that the source 665 algorithm reproduces the Kalashnikova et al. [2000] production rate (dotted) and indicate 666 how the rate is scaled to preserve mass for the 1 nm radius (dashed) and 0.2 nm radius 667 (solid) cases. 668
669
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 33 of 54
669
670 Figure 3. Evolution of the vertical distribution of the mass of meteoric dust in the 671 atmosphere during the 10 years of the control simulation (left) and larger view of the last 672 2 years (right). At high altitudes, dust descending in the winter pole creates a semiannual 673 pattern in dust distribution, while at low altitudes there is an annual pattern with 674 enhanced loss during the southern hemisphere winter and spring. Steady state takes the 675 longest time to attain in the lower stratosphere. 676
677
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 34 of 54
677
678 Figure 4. The atmosphere does not quite reach steady state during the 10 years of the 679 control simulation as indicated by the total mass of meteoric dust (solid line). The 680 atmosphere above 30 km (dotted line) reaches steady state within 3 years, while after 10 681 years the model is only just reaching steady state below 30 km (dashed line). There is a 682 seasonal cycle in the total mass, which is likely caused by enhanced loss below 30 km in 683 the polar vortex during the southern hemisphere winter. 684
685
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 35 of 54
685
686 Figure 5. Zonal average plots for January (top) and July (bottom) of the meteoric dust 687 particle concentration, mass density, and effective radius for the control simulation. The 688 data is an average of the last 7 years of the 10-year simulation. A strong seasonal 689 variation is evident with particles being advected from the summer to winter pole by the 690 mesospheric circulation. There is also rapid descent through the mesosphere at the winter 691 pole. The peak in effective radius at low latitudes and altitudes is because the circulation 692 ascends from the summer pole to the low latitudes in the winter hemisphere and thus only 693 large particles with higher sedimentation velocities will be able to descend in the tropics. 694
695
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 36 of 54
695
696 Figure 6. Polar plots of the mass mixing ratio of meteoric dust in the southern polar 697 region showing the months of peak concentrations at the selected altitudes as the dust 698 descends in the polar vortex. The white arrows are the monthly averaged horizontal wind 699 vectors. The monthly averaged dust mass mixing ratio contours have been scaled 700 differently for each month to emphasize the structure. 701
702
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 37 of 54
702
703 Figure 7. The annual variation in the vertical profile of the meteoric dust mass mixing 704 ratio inside (left) and outside (right) the polar vortex for the northern (top) and southern 705 (bottom) hemispheres. The dust descends in the polar vortex rapidly through the 706 mesosphere and more gradually in the stratosphere. The peaks near the mesopause in the 707 winter to spring time frame are because of the mesospheric meridional circulation and are 708 not confined to the vortex. 709
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 38 of 54
710 Figure 8. The ratio of the average mass mixing ratio of meteoric dust inside the polar 711 vortex to that outside the vortex for the Northern Hemisphere (left) and Southern 712 Hemisphere (right). The descent of dust within the vortex is clearly visible. The vortex is 713 particularly isolated in the Southern Hemisphere winter from 700 to 2000K. 714 715
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 39 of 54
716 Figure 9. Comparison of the average vertical profile of concentration and surface area 717 density of meteoric dust between Hunten et al. [1980] (thick green line) and the control 718 simulation (thin lines). The color indicates the latitude range and the line style indicates 719 the time period being averaged. The left panels are the annual averages and the right 720 panels are the JJA (dotted) and DJF (dashed) seasonal averages. Hunten et al. [1980] is a 721 mid-latitude model; however, the mid-latitude annual averages from the control run (red 722 and blue lines) are significantly reduced below ~75 km in both concentration and surface 723 area density relative to Hunten et al. [1980]. The summer season is significantly depleted 724 and winter season is enhanced in concentration and surface area density at mid to high 725 latitudes. 726
727
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 40 of 54
727
728 Figure 10. Comparison of the average size distribution of meteoric dust at selected 729 altitudes between Hunten et al. [1980] (thick green line) and the control simulation (thin 730 lines). The color indicates the latitude range and the line style indicates the time period 731 being averaged. The top panels are the annual averages and the bottom panels are the JJA 732 (dotted) and DJF (dashed) seasonal averages. The small sizes are depleted during the mid 733 and high latitude summer at 60 km and 30 km. The large sizes are enhanced at 30 km 734 because of increased coagulation in the polar vortex. 735
736
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 41 of 54
736
737 Figure 11. Comparison of particle concentrations between the control simulation and 738 rocket soundings by Lynch et al. [2005] from Poker Flats, Alaska. Their observation of 739 charged particles has been converted into dust particles by assuming that 5% of the dust 740 particles are charged and that the particle radius is ~1 nm [Gelinas, et al., 2005; Lynch, et 741 al., 2005]. The black line is the average of the last 7 years of the control run taken from 742 the 4˚x5˚model grid point containing Poker Flats using a daily average for the date 743 indicated assuming 1 nm particles. The blue line is sampled in a similar way from 744 simulation 7, and assumes the particle radius is 0.8 nm to 1.3 nm. The vertical resolution 745 of the control run near the mesopause is ~3.5 km, while simulation 7 has an increased 746 resolution of ~0.5 km. The points on the solid lines indicate the midpoint altitudes of the 747 model levels. The dashed lines are the observations. Both simulations do a good job of 748 matching the observations; however, simulation 7 with a higher vertical resolution that 749 supports sharper gradients is a better match at lower and higher altitudes. 750
751
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 42 of 54
751
752 Figure 12. Zonal average plots for July of the concentration and mass density for 753 simulation 2, the no sedimentation case. The difference between these results and the 754 control simulation are shown in the right panels. Sedimentation is important in 755 establishing the boundary near 90 km of high gradients in dust concentrations. It plays a 756 lesser role in transporting dust into the stratosphere. 757
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 43 of 54
758 Figure 13. Meteoric dust fall velocities from CARMA as a function of altitude for each 759 particle bin from 0.2 nm (left) to 100 nm (right). Velocities for mean radii of 1 nm and 10 760 nm are highlighted (bold lines). 761
762
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 44 of 54
762
763 Figure 14. Zonal average plots for July of the fourth year of the simulation for the 764 temperature and the vertical wind in the control simulation and simulations 5 and 6, 765 which use different values for the gravity wave parameterization. The black lines are 766 temperature contours from 130 to 160 K. A smaller background shear stress (τb
*) causes 767 the waves to break at higher altitudes. Waves breaking at higher altitudes raise and cool 768 the summer mesopause and increase the downdrafts in the winter polar vortex. 769
770
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 45 of 54
770
771 Figure 15. Zonal average plots for July of the mass density for simulation 3, the no 772 coagulation case. The difference between this result and the control simulation are shown 773 in the right panel. Below 80 km, this simulation is similar to the no sedimentation case 774 shown in Figure 12, since there are no large particles created and thus sedimentation 775 velocities are low. 776
777
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 46 of 54
778 Figure 16. Zonal average plots for July of the concentration of particles with 1 nm or 779 larger radius, mass density, and effective radius for simulation 4, the 0.2 nm initial radius 780 case. The difference between these results and the control simulation are shown in the 781 right panels. Starting with a smaller initial radius reduces the number of larger particles 782 formed by coagulation and therefore slightly reduces the sedimentation into the 783 stratosphere. The effective radius in the summer near 80 km is also significantly reduced 784 relative to the control run. 785
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 47 of 54
786 Figure 17. Comparison of the average vertical profile of concentration and surface area 787 density of meteoric dust between Hunten et al. [1980] (thick green line) and simulation 4, 788 the 0.2 nm initial radius case (thin lines). The color indicates the latitude range and the 789 line style indicates the time period being averaged. The left panels are the annual 790 averages and the right panels are the JJA (dotted) and DJF (dashed) seasonal averages. 791 Hunten et al. [1980] is a mid-latitude model; however, the mid-latitude annual averages 792 from the control run (red and blue lines) below ~75 km are significantly reduced in both 793 concentration and surface area density relative to Hunten et al. [1980]. The summer 794 season is significantly depleted and winter season is enhanced in concentration and 795 surface area density at mid to high latitudes. 796
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 48 of 54
797 Figure 18. Comparison of the average size distribution of meteoric dust at selected 798 altitudes between Hunten et al. [1980] (thick green line) and simulation 4, the 0.2 nm 799 initial radius case (thin lines). The color indicates the latitude range and the line style 800 indicates the time period being averaged. The top panels are the annual averages and the 801 bottom panels are the JJA (dotted) and DJF (dashed) seasonal averages. The small sizes 802 are depleted during the mid and high latitude summer at 30 km and 60 km because of the 803 mesospheric circulation. The large sizes are enhanced at 30 km because of increased 804 coagulation in the polar vortex. 805
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 49 of 54
806 Figure 19. Zonal average plots for July of the fourth year of the simulation for the 807 concentration and mass density of dust particles in the control simulation and simulations 808 5 and 6, which use different values for the gravity wave parameterization. The black lines 809 are temperature contours from 130 to 160 K. The increased meridional wind relative to 810 the control run causes a reduced concentration near the summer mesopause. Increased 811 vertical winds relative to the control run cause more rapid transport in the winter polar 812 vortex above 60 km. Below 60 km, decreased vertical winds relative to the control run 813 slow transport into the stratosphere. 814
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 50 of 54
815 Figure 20. Zonal average plots for January and July of the concentration of particles with 816 1 nm or larger radius for simulation 7, which has enhanced vertical resolution and a more 817 realistic mesopause thermal structure than the control run. The black lines are 818 temperature contours from 130 to 160 K. Particle concentrations near the summer 819 mesopause are reduced compared to other simulations. 820
821
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 51 of 54
821
822 Figure 21. Polar plots of the concentration of dust particles with a radius of 1 nm or 823 larger during the month of July in the northern hemisphere for simulation 7 at various 824 altitudes. There are widespread areas with large depletions relative to Hunten et al. 825 [1980] and to our other simulations, particularly below the mesopause. 826
827
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 52 of 54
827 Figure 22. Comparison of the average size distribution of meteoric dust during the 828 summer at high latitude between Hunten et al. [1980] (green line) at 90 km and 829 simulation 7 (black lines) at various altitudes near the mesopause. There is a significant 830 depletion in the concentration of smaller particles at all altitudes relative to Hunten et al. 831 [1980]. There is a small increase in the concentration of large particles due to particles 832 that are transported up from lower altitudes. 833
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 53 of 54
834 Figure 23. Comparison of the average vertical profile of concentration and mass density 835 of meteoric dust during the summer (solid) and winter (dashed) between Megner et al. 836 [2006] (green) and simulation 7 at high (purple) and mid (blue) latitude in the Northern 837 Hemisphere. Our model shows significantly less depletion in the summer and slightly 838 less enhancement in the winter relative to Megner et al. [2006], who used extreme 839 vertical velocities from CHEM2D for their winter and summer wind profiles. The 840 concentration of particles 1 nm and larger and the mass density in the summer in our 841 results are also influenced by the uplift of residual particles from the previous winter. 842
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ddd, Addddd, doi:10.1029/2007JAdddddd, 2007.
DRAFT 29 May 2007 54 of 54
843 Figure 24. Comparison of the average size distribution of meteoric dust between 80 km 844 and 90 km during the summer (solid) and winter (dashed) between Megner et al. [2006] 845 (green) and simulation 7 at high (purple) and mid (blue) latitude in the Northern 846 Hemisphere. There are more large particles relative to Megner et al. [2006], particularly 847 in the summer. 848