Corresponding author: D. E. Siskind, Space Science Division, Naval Research Laboratory, Washington, DC, USA. (firstname.lastname@example.org)
 Observations of polar mesospheric clouds by the Aeronomy of Ice in the Mesosphere Explorer show that for the Northern summers of 2007–2010, the cloud ice water content (IWC) and occurrence frequency varied with the meteorological forcing from the Southern winter stratosphere. With the increase in solar flux in the last two years, expectations were that the clouds would decrease due to reduced water vapor (H2O) and/or higher temperatures. Surprisingly, we observe more clouds in 2011 and 40% greater IWC in 2011 and 2012. The increase is particularly pronounced in the clouds with highest IWC. These high IWC clouds are associated with significant enhancements in total H2O (vapor and ice). We suggest this implies an additional source of H2O and that this is provided by space traffic exhaust. A preliminary estimate of the H2O released from summertime space traffic over the last six years is qualitatively consistent with this suggestion.
 Our general understanding of polar mesospheric clouds (PMCs) is that their occurrence and brightness should be anticorrelated with variations in the solar UV flux. This is because increases in the flux of solar Lyman α radiation should cause reduced formation of PMCs due either to increased photodissociation of H2O [Garcia, 1989] or increased heating [Siskind et al., 2005]. Satellite observations from the Halogen Occultation Experiment [Hervig and Siskind, 2006] and the Solar Backscatter Ultraviolet instruments [Shettle et al., 2009] have shown this anticorrelation of PMCs with Lyman α. Recently, a variation associated with the 27 day solar rotation has also been identified [Robert et al., 2010]. Ground-based results, however, have been less conclusive, with a solar cycle variation identified in only a subset of the available data [Fiedler et al., 2011].
 One goal of NASA's Aeronomy of Ice in the Mesosphere (AIM) mission has been to identify and quantify the response of PMCs to solar activity concomitant with detailed observations of temperature and water vapor. However, fulfilling this objective has been hindered by the fact that the March 2007 launch of AIM coincided with the onset of a prolonged solar minimum. The first four Northern summers of AIM were characterized by very low solar activity, with average values of the F10.7 index (a proxy for both Lyman α and the far-ultraviolet solar flux with units of sfu) at or below 80 sfu. Thus, instead of solar-induced variability, analysis of AIM/PMC data to date has placed greater emphasis on the response of PMCs to meteorological forcing from the stratosphere [e.g., Becker and Fritts, 2006; Karlsson et al., 2009; Siskind et al., 2011].
 In the last two years, however, solar activity has increased. Figure 1 shows the mean value of the F10.7 index for the six Julys of the AIM mission (http://www.swpc.noaa.gov/ftpmenu/indices/old_indices.html). The value of 136 for July 2012 is halfway up to a strong solar maximum. However, as is also seen in Figure 1 and as we discuss in more detail below, the two indicators of PMC occurrence from the two experiments on AIM (SOFIE: the Solar Occulation for Ice Experiment and CIPS: the Cloud Imaging and Particle Size experiment) do not anticorrelate with the recent solar flux increase. The purpose of this paper is to quantify and highlight a possible reason for this disagreement with our expectation.
2 Review of Data Sets
 Aeronomy of Ice in the Mesosphere measures PMCs using two approaches [Russell et al., 2009]. SOFIE uses solar occultation to produce limb profiles of temperature, pressure, key trace constituents, and PMC extinction at 11 wavelengths from 292 to 5316 nm [Hervig et al., 2009]. The CIPS experiment uses UV nadir imaging to observe scattered sunlight from PMCs, from which cloud morphology and properties are derived [Bailey et al., 2009]. Because these techniques are so different, we use different diagnostics to characterize cloud variability. For SOFIE, ice mass density is a fundamental measurement because it is proportional to the observed infrared extinction. The vertical integral of mass density yields the column ice water content (IWC) of PMCs. For CIPS, we use frequency of ice occurrence (for clouds with albedo greater than 2 × 10−6 sr−1) because the brightness enhancements against a Rayleigh scattered background are readily detected [Lumpe et al., 2013]. We use CIPS data from the ascending node because the measurements from that portion of the AIM orbit are nearly simultaneous with the SOFIE measurements [Russell et al., 2009].
 Figure 1 presents a time series of July averaged Northern Hemisphere (NH) PMC IWC as recorded by SOFIE, and occurrence frequency as recorded by CIPS. In contrast to the solar activity, which shows a monotonic increase over the last 4 years, both of the PMC time series show a double peaked variation with maxima in 2008 and 2011. While the 2008 maximum appears to occur at solar minimum, the 2011 maximum is completely unexpected. Furthermore, IWC and occurrence frequency remain high in 2012 despite the sharp increase in solar activity.
 To properly understand and quantify the physical significance of these distinctions, we must consider the cloud occurrence in the context of the meteorological forcing due to teleconnections from the winter hemisphere [e.g., Karlsson et al., 2009; Siskind et al., 2011]. Figure 2 shows the same PMC data from Figure 1, now plotted against a teleconnection index (TI) similar to that presented by Karlsson et al. . In the present work the TI was obtained from the monthly averaged stratospheric temperature deviation from the mean value over 2002–2012 (obtained from http://acdb-ext.gsfc.nasa.gov/Data_services/met/nmc_climatology.html) for 50°S, averaged over pressures from100 to 10 mb. For the July average PMC data shown here, we use an averaging interval of 25 June to 25 July. This allows for a 5 day lag between winter temperatures and summer clouds as discussed by Karlsson et al. . Positive (negative) values of the TI indicate coupling from a relatively warmer (colder) winter stratosphere. Thus in 2007, the SH stratosphere was warmer and this generated a pattern of warmth that extended into the NH mesosphere, causing lower IWC and reduced PMC frequency in the NH [Siskind et al., 2011]. For 2008, the opposite was true and NH PMC IWC and occurrence were likewise high. Figure 2 shows that during the extended solar minimum (2007–2010), PMC variations from CIPS and SOFIE are almost completely determined by teleconnection from the winter hemisphere. The correlation coefficient between the TI and the two PMC data parameters are 0.99 for IWC and 0.96 for frequency. This excellent correlation, even if only for 4 years, is in agreement with the in-depth look at the global temperature structure presented by Siskind et al.  for 2007–2009.
 Thus, the 2008 peak in Figure 1 is explained by stratospheric meteorology, but the 2011 peak requires further examination. The magnitude of the 2011 anomaly can be quantified by comparing the actual IWC and occurrence frequency to that predicted by the TI and illustrated by the solid lines in Figure 2. For IWC, the TI index predicts 53.5 g km−2 for 2011; the actual value of 78.8 exceeds this by a factor of 1.47. For frequency, the TI predicts 27.5%, the actual value of 34.5% exceeds this by a factor of 1.25. SOFIE IWC also shows an enhancement in 2012, which is not seen in CIPS occurrence frequency. Nonetheless, with the increase in solar activity in 2011 and, especially, in 2012 we would have expected those points to fall below the TI lines in Figure 2. Based upon previously reported solar cycle variations from the Halogen Occultation Experiment [Hervig and Siskind, 2006] and Solar Backscatter Ultraviolet [Shettle et al., 2009]; a half solar cycle should be associated with a decrease of at least 20%. For 2012, this implies an averaged SOFIE IWC of about 40–45 g km−2 and CIPS frequency of about 20%, both of which are well below what we see. SOFIE and CIPS are therefore both consistent in failing to show any suggestion of a solar cycle decrease in PMCs despite the significant enhancement in solar flux for both July 2011 and July 2012. Given that SOFIE and CIPS view the atmosphere with profoundly different techniques, we are satisfied by the overall agreement in the interannual variability displayed by the two instruments.
 To further understand what might be underlying these cloud anomalies, Figure 3 presents histograms of the number of occurrences of IWC measured by SOFIE, binned in intervals of 10 g km−2, for July observations during the 6 year record. In a given month there are approximately 400–450 individual observations of clouds by SOFIE (out of 31 days with ~15 observations per day). Each panel in Figure 3 shows the distribution of these observations in IWC. The main point is that the most dramatic year-to-year changes are for the high IWC (> 100 g km−2) values. Thus, all six Julys consistently display about 25–35 occurrences of IWC values in the bins below 50 g km−2. However, for example, in 2007, the year with the fewest clouds, very few clouds greater than 100 g km−2 were recorded, while in 2011, many clouds were observed with IWC extending well above 200 g km−2. High IWC values (> 150 g km−2) were also seen in 2008 and 2012. Based upon the analysis associated with Figure 2, we link the occurrence of the high IWC clouds in 2008 to the high value of the TI.
 To focus more on the high IWC clouds seen in 2011 and 2012, we examined the water budget for the PMCs observed by SOFIE for the period 2008–2010 compared with 2011–2012. Water can be in two forms, ice or vapor; however, the total amount of water vapor in the mesopause region should follow the constraints established by the influx of water vapor and methane (CH4) up through the tropical tropopause region [Wrotny et al., 2010]. SOFIE gas phase water vapor data has been validated by Rong et al. , indicating agreement to within 20% of independent observations. Figures 4a and 4b show that for the overall cloud population in July, the amount of water tied up in ice is a small fraction of the overall water budget and that the average total water (ice + vapor) maximizes near 7–7.5 ppmv. This is in good general agreement with our understanding of the middle atmosphere total hydrogen budget, which maximizes around 7.25 ppmv [Wrotny et al., 2010]. For the subset of very high IWC clouds, a large fraction of the H2O is in the ice phase and the total water exceeds the stratospheric values. Figure 4 shows that for 2008–2010 the peak water averages about 8.5–9.0 ppmv for the high IWC clouds, whereas for the 2011–2012 average, this total is greater than 10 ppmv. Thus, it suggests that for the same population of high mass density PMCs (IWC > 200 g km−2), the water budget is greater in 2011–2012 than for the 2008–2010 period.
 To emphasize the difference between the 2011–2012 period and the 2008–2010 period, Figure 5 presents a histogram of the number of occurrences of peak H2O for the highest IWC clouds (> 200 g km−2). There are only 11 such events for the 2008–2010 period and 27 for 2011–2012. Figure 5 shows that the peak H2O occurrence is 9.0 ppmv; however, for 2011–2012 there is a tail of very high values of peak H2O (> 12 ppmv) not seen in earlier years. Although there are only a limited amount of clouds with IWC > 200 g km−2, there is a suggestion that there are some very high values of peak H2O in 2011–2012, well in excess of the stratospheric budget.
 One way to exceed the stratospheric H2O budget is with an additional source. We suggest that this source is from space traffic exhaust. Table 1 presents an estimate of the amount of H2O delivered to the summer mesopause by liquid-fueled launch vehicles worldwide during the six Julys under discussion (Stevens, M. et al., in preparation, 2013). These values were derived using tabulated trajectory information to estimate the total exhaust for each vehicle in the atmospheric layer from 90 to 140 km weighted by the fraction of water vapor contained in the exhaust. For the space shuttle, for example, this fraction is near 100%. Except as noted below, we make no other assumptions regarding the latitude of the launch site or time of day of the launch. The table shows that space traffic released more H2O in 2011 than in any year since the beginning of the AIM mission. The effects of STS 135, launched on 8 July 2011 at 11:40 EDT, have already been documented, and it was shown that some of the brightest PMCs ever observed by CIPS and from the ground were formed from the main engine exhaust of this shuttle [Stevens et al., 2012]. This is consistent with not only the July 2011 enhancement we see here but also the occurrence of very high IWC in the 2011 SOFIE data (Figure 3). However, we should note that for SOFIE the highest IWC clouds are recorded in the last week of the month, almost 3 weeks after the shuttle launch. This may have some interesting implications for the residence time scale of launch exhaust, a potential topic for further study.
Table 1. Lower Thermospheric H2O Injected by Space Traffic in July: 2007–2012
aFrom space traffic worldwide between 90 and 140 km altitude.
412 tons from shuttle, July 15 (see text)
Shuttle, July 9
 For years prior to 2011, the estimated amount of H2O from space traffic is much lower, with the exception of 2009. The large 2009 value is driven by the launch of STS 127 on 15 July at 1700 local solar time. However, it is likely that the lower thermospheric dynamics were not favorable for transport to the summer pole. Siskind et al.  suggested that there is a local time dependence such that plumes from morning and midday launches travel to the north and plumes from evening launches travel to the south. Indeed Pumphrey et al.  observed that this STS-127 plume had moved south rather than north 20 h after launch. The amount of H2O emitted in 2012 was less than half as much as in 2011, but still twice as high as in 2008, the next highest year. It is interesting that 2012 shows enhanced H2O vapor from space traffic even without any Shuttle launches (the tabulated H2O is due, in part, to launches of the Russian Soyuz). This may explain the relatively greater number of IWC > 200 g km−2 occurrences recorded by SOFIE in 2012.
 We have shown that for the first four NH summers of the AIM mission, the variations in average July PMC IWC and occurrence frequency are almost completely accounted for by temperature anomalies, which teleconnect from the winter stratosphere to the summer mesosphere. These first four years corresponded to very low solar activity and thus serve as a useful baseline for conditions of relatively “pure” meteorological forcing of PMCs. In the most recent two years, despite the increase in solar activity, we do not see any decrease in PMCs from that predicted by teleconnections. Instead we see enhancements that were most pronounced in those clouds with extremely high IWC (> 200 g km−2).
 Our suggestion that the explanation lies in an H2O source from space traffic steps beyond previous discussions of this effect in three ways. First, previous reports have concentrated on relatively isolated events of PMC production by shuttle exhaust [e.g., Stevens et al., 2005, 2012]. Here we are suggesting that average PMC mass density and frequency, and its year-to-year variability from 2007 to 2012, are modulated by space traffic. Furthermore, the July 2012 changes observed by SOFIE raise the provocative suggestion that measurable effects of rocket exhaust on PMCs might outlive the termination of the shuttle program. This question probably cannot be resolved with the data in hand, but can be as more data is taken in the next several years. Finally, our results suggest that both meteorological variability due to teleconnections and the effects of rocket exhaust overwhelm solar cycle forcing in midsummer PMC occurrence for the latitude band 66–71°N.
 This work was funded by the NASA Aeronomy of Ice in the Mesosphere (AIM) project, under NASA's Small Explorers Program.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.