Sudden stratospheric warming (SSW) is a large-scale meteorological process in the winter hemisphere lasting several days or weeks. The Incoherent Scatter World Day campaign conducted on January 17–February 1, 2008 was arranged during a minor SSW event and focuses on studies of thermospheric and ionospheric response to stratospheric changes. We analyze ion temperature observations at 100–300 km height obtained by the Millstone Hill incoherent scatter radar (42.6°N, 288.5°E). Alternating regions of warming in the lower thermosphere and cooling above 150km altitude were observed by the radar. We use National Center for Environmental Prediction (NCEP) temperature data at 10hPa (∼30km) level and the F10.7 and Ap indices to identify any cause-effect relationship between observed variations in the temperature and stratospheric warming event. We conclude that the seasonal trend, solar flux and geomagnetic activity cannot account for the observed warming and cooling temperature variation and suggest that this variation is associated with stratospheric warming. This study demonstrates a link between the lower atmosphere and the ionosphere which has not been considered before and indicates that ionospheric variability as part of space weather should be considered in conjunction with stratospheric changes.
 Studies of ionospheric variability over many years have produced sufficient evidence that a significant portion of day-to-day variations in ionospheric parameters cannot be explained by such major and relatively well understood drivers as solar ionizing flux and geomagnetic activity. Portion of ionospheric variability not accounted by these drivers typically amounts to ∼20% of F-region maximum electron density [i.e., Forbes et al., 2000; Mendillo et al., 2002] and is thought to be associated with lower atmospheric processes. Sudden stratospheric warming (SSW) is one of the most spectacular manifestations of strong vertical coupling between different atmospheric regions. At high-latitude stratospheric heights, the zonal mean temperatures and zonal wind flow are disrupted, leading to rise in temperatures up to several tens of degrees and decrease or even reversal of zonal mean flow from eastward to westward [Naujokat and Labitzke, 1993]. The commonly accepted primary mechanism leading to SSW is interaction of planetary waves with zonal mean flow [Matsuno, 1971].
 Large variability in the mesosphere in association with SSW is evident in numerous observational studies. Stratospheric warmings lead to a cooling at mesospheric altitudes, as was reported by Labitzke [1972, 1981]. Recent progress in experimental techniques resulted in a number of studies reporting effects of SSW at the altitudes of upper mesosphere and lower thermosphere, with mesospheric cooling at 70–90 km altitude found in OH airglow temperature [Walterscheid et al., 2000, Azeem et al., 2005], lidar, meteor radar [Hoffmann et al., 2002, 2007] and SABER [Siskind et al., 2005] data, to name a few. Cooling in the mesosphere reportedly precedes warming in the stratosphere and is also associated with large dynamical changes in the mesosphere, which include zonal wind reversal, increase in planetary wave activity and changes in the gravity wave activity [i.e., Hoffmann et al., 2007].
 SSW manifestations in the MLT region remain an active area of research, as they are not well characterized and understood. In contrast, any effects at higher altitudes (> 100 km) are practically unknown. TIMEGCM simulations [Liu and Roble, 2002] indicate secondary lower thermospheric warming (∼120–130 km) associated with SSW, but this prediction was not tested up to now for the lack of data in this altitude region. Some indication of such warming was found in SABER data, though at lower altitudes [Siskind et al., 2005]. In the F-region, decrease in the electron density with 1-day delay relative to the peak warming at 30hPa was reported by Kazimirovsky et al. . Although not directly related to SSW, a large body of research indicates relationship between stratospheric conditions and ionospheric electron density [i.e., Danilov and Vanina, 2003]. These studies showed significance in the statistical sense, but little is known about the mechanism of stratosphere-ionosphere coupling. During SSW, when major variations in the state of stratosphere and mesosphere occur, this coupling undoubtedly changes, thus making SSW a convenient, nature-provided event for studying connections between the lower atmosphere and ionosphere.
 In January 2008, Incoherent Scatter World Day campaign was organized with a goal of extending studies of SSW effects to heights above 100 km. The main objectives of the experiment were understanding and quantifying if and how large stratospheric changes are related to variability of ionospheric parameters. All existing incoherent scatter radars (ISR) participated in the campaign, collecting data on temporal and altitudinal variations in such parameters as electron density (Ne), electron (Te) and ion (Ti) temperatures, F-region and E-region neutral winds. The present report is focused on ion temperature results obtained by Millstone Hill ISR (42.6°N, 288.5°E) during this campaign. We use ion temperature data in this study as it provides direct evidence of energy coupling between different layers of the upper atmosphere. In addition, ion temperature is a good measure of neutral temperature for lower heights, and close to exospheric temperature at ∼250 km. Variations in other parameters and latitudinal and longitudinal relationship between ionospheric changes and location of the SSW will be investigated in separate papers.
2. January 2008 Sudden Stratospheric Warming
Figure 1 summarizes stratospheric and geophysical conditions during the campaign period, January 17, 2008–February 1, 2008. After staying at historically low levels in December 2007 and first part of January 2008, stratospheric temperatures began increasing on January 21–22 and reached a peak on January 24, 2008, indicating sudden stratospheric warming. Figure 1a shows NCEP stratospheric temperatures at 10hPa (∼30km) for 90°N (triangles) and zonally averaged temperatures for 55–75°N (circles) in January 2008 (solid lines) in comparison with ∼30-year median temperatures (dashed lines). At 90°N, the warming exceeded 70K and the peak temperature of 267K broke all-time record. The temperature anomaly shows a clear downward progression, with peak warming at 30hPa (∼23 km) occurring 2–3 days later (not shown). The stratospheric circulation, characterized in Figure 1b by a zonal mean zonal wind at 60°N and 10hPa, shows decrease in the eastward wind. This SSW occurred during very low and slowly changing solar activity (F10.7 = 71–74) and low geomagnetic activity (Kp < 3+, Ap3 = 0–22, average Ap3 = 7), thus reducing influence of these major drivers of ionospheric variability.
3. Results and Discussion
 Measurements of ionospheric parameters (Ne, Te, Ti, wind) were obtained by the Millstone Hill ISR from January 17, 2008 to February 1, 2008. We limit this study to daytime data only to avoid influences from the midlatitude trough, which was observed on several nights. To minimize temperatures variations due to solar ionizing flux, geomagnetic activity, and season [Zhang and Holt, 2007], we use as a baseline case data from January 20–23, 2007, with F10.7 = 79 and Kp < 3+ (Ap3 = 3–8, average Ap3 = 5). Figure 2 presents difference field of daytime ion temperature at altitudes of 100–300 km between mean January 2008 data (i.e., Jan 17–Feb 1 period) and mean January 2007 data (i.e., Jan 20–23 period). A 20–50K decrease in mean January 2008 temperature is observed above ∼140 km, with maximum temperature differences recorded in the morning hours (7–11LT) and afternoon hours (15–19LT). The lower thermospheric warming in the altitude range of ∼120–140 km exceeds 30–50K in the afternoon. The observed variation in ion temperature is consistent for all three antenna pointing directions and for both alternating code (i.e., ∼5km altitude resolution) and single pulse (i.e., ∼18km altitude resolution) modes.
Figure 3 (left) shows baseline (i.e., January 2007) ion temperatures at 130 km and 230 km (F-region peak), with error bars representing standard deviation for 1-hour bins. Figure 3 (right) shows the observed difference between January 2008 data and baseline data for 130km and 230km altitudes (dark symbols) as well as the difference expected from the empirical model (light symbols). As the reference case of Jan 20–23, 2007 had slightly different solar flux and magnetic activity conditions, we have estimated the resulting Ti changes based on a local empirical ionospheric model [Zhang and Holt, 2007]. As could be expected, at the lower altitude small change in F10.7 and Ap leads to very small difference in Ti (∼1K). This indicates that solar flux and magnetic activity could not be responsible for the ∼20–100K temperature increase observed in January 2008 in the lower thermosphere. In addition, in the baseline case the diurnal behavior of Ti is strongly controlled by the solar zenith angle with minor influence of tides, which usually dissipate or become evanescent by 130 km [Goncharenko and Salah, 1998]. In contrast, in January 2008 daily behavior in Ti seems to be strongly affected by tidal influences.
 At higher altitudes, combined effects of solar flux and magnetic activity variations are more pronounced, particularly around local post-noon, where they can reach as much as 10–12K and account for the observed temperature differences. However, the largest temperature differences at 230 km reach 30–35K and are observed in the morning and afternoon, when solar flux influences are expected to account for 6–10K change. Overall, we conclude that such usual drivers of ionospheric variability as solar flux and magnetic activity could not be responsible for the observed variation in Ti for the following reasons:
 1. The magnitude of the observed Ti variation is much higher than could be driven by minor variation in F10.7 and Ap.
 2. Observed variation in Ti is qualitatively different from changes driven by F10.7 and Ap, i.e., we report warming at lower altitudes (120–140km) and cooling above 150 km, while decrease in F10.7 is expected to generate cooler temperatures at all altitudes.
 3. Observed variation in Ti maximizes in the morning (7-11LT) and afternoon (15–19LT), while solar flux effects are most pronounced around local noon.
Figure 4 illustrates development of Ti cooling at 230 km in comparison with the baseline data. To decrease influence of daily variations in F10.7 and Kp and to compare with the 3-day mean values from January 2007, a 3-day running means and 1-hour binning are used for January 2008. Dashed lines correspond to zero variation from the background for a particular day and are separated by 100K. During the first three days of experiment, Jan. 17–19, the temperature differences are small, indicating data consistency from one winter to another. A major decrease in Ti begins developing on January 20, i.e., four days prior to peak in stratospheric warming at 10hPa. A particularly large cooling reaching 75K is observed in morning hours on January 25–27, with subsequent slow recovery of Ti and return to January 2007 level by February 1, 2008. As decrease in Ti occurs at the time of warmer stratospheric temperatures and cannot be (at least fully) driven by variations in solar and geomagnetic activity, we suggest it is associated with sudden stratospheric warming.
 Our observations of Ti decrease prior to the peak of the stratospheric warming are consistent with earlier reports of mesospheric cooling and wind reversal preceding SSW events [i.e., Walterscheid et al., 2000; Azeem et al., 2005; Hoffmann et al., 2007]. This could be related to the fact that significant changes in atmospheric parameters are easier to be produced at higher altitudes due to lower neutral density. The weakening or even reversal of the stratospheric jet during SSW allows more eastward penetrating gravity waves to propagate into the MLT region, causing stronger equatorward and upward meridional circulation, which leads to adiabatic cooling in the mesosphere as predicted by Liu and Roble . The lower thermospheric warming predicted by the same simulation is generated by a secondary downward circulation induced by the equatorward mesospheric circulation. Our observations of this warming at 120–140 km agrees well with TIMEGCM predictions (see Figure 3 in Liu and Roble ), albeit it is observed at the middle latitude site. In contrast, decrease in Ti at higher altitudes is an unexpected result not seen in the simulation. Its interpretation from single site observations is complicated, but the temporal and altitudinal variation in the temperature anomaly suggests it might be related to semidiurnal modulation. Large diurnal and semidiurnal variations could be created through combined forcing of wave 1 components (quasi-stationary wave 1, migrating diurnal and semidiurnal wave 1) and/or through combined forcing of wave 2 components (quasi-stationary and migrating semidiurnal tide). It remains an open question whether planetary waves can propagate from the stratosphere up to thermosphere above 100 km and affect ionospheric variability or whether planetary waves forced in situ by filtered gravity waves play a role at these altitudes. Further studies including analysis of other ionospheric and thermospheric parameters and other locations in conjunction with simulations are required to understand the relative importance of different processes.
 Observations of ion temperatures at the middle latitude site (42.6°N, 288.5°E) during stratospheric warming of January 2008 reveal alternating regions of cooling in a large altitude range (150–300km) and warming in a narrow altitude band (120–140km). We rule out seasonal trend, solar flux and geomagnetic activity as significant causes of such variation and suggest that it is associated with conditions leading to a stratospheric warming. The cooling reaches as much as 75K in the F1–F2 regions (11–16% of the background temperature) and is pronounced mostly in the morning and afternoon hours. The maximum warming exceeds 80K and is largest in the morning and afternoon hours as well. The cooling begins to develop 4 days prior to peak stratospheric warming and subsides 8 days after the peak stratospheric warming.
 Alternating areas of warm and cold vertical zones in the atmosphere are well established phenomena, with mesospheric cooling accompanying stratospheric warming. Our observations show for the first time that areas of warming and cooling extend to altitudes of upper thermosphere (∼300 km) and suggest that conditions leading to sudden stratospheric warmings may impact the ionosphere. This finding indicates strong links between the state of the lower atmosphere and ionosphere. Further research of these links will provide better understanding of processes governing variability in ionospheric parameters and lead to better predictions of ionospheric parameters as part of space weather.
 This work was supported by NSF Cooperative Agreement ATM 0417666. The Millstone Hill incoherent scatter radar is supported by the US National Science Foundation (NSF) as part of the Upper Atmosphere Facility Program. The authors gratefully acknowledge the access to the NCEP data.