Mount Everest is often referred to as the earth's ‘third’ pole. As such it is relatively inaccessible and little is known about its meteorology. In 1998, a portable weather station was operated at the mountain's South Col, elevation 7,986 m. We believe that this represents the highest elevation at which continuous weather data has ever been collected and thus represents a valuable dataset with which to investigate the meteorology of the high Himalaya. In this paper, we compare the observations with reanalyses from the National Centers for Environmental Prediction and the European Center for Medium-Range Weather Forecasts. We find that both reanalyses capture much of the synoptic-scale variability in the temperature and pressure at the South Col site, especially in the pre-monsoon season. Furthermore, we show that an observed weather event was the result of convection associated with a jet streak and an intrusion of stratospheric air into the upper-troposphere.
 For almost a century, Mount Everest has been the subject of exploration for both scientific and recreational purposes [Venables, 2003]. Much of the focus of scientific interest in Mount Everest has been on the impact that the low barometric pressure that exists near its summit has on human physiology and performance [West, 1999; Huey and Eguskitza, 2001]. However the weather and climate of the region exhibit a number of phenomena that are also of considerable interest. Blanford  and Walker  suggested that spring snowfall in the Himalaya was inversely correlated with the intensity of the following Indian Summer Monsoon. In the intervening years, the correlation has weakened and recently appears to have undergone a reversal in its sign [Bamzai and Shukla, 1999; Robock et al., 2003]. As a result of its extreme height, the summit of Mount Everest is often buffeted by high winds associated with the subtropical jet stream [Krishnamurti, 1961]. In recent years, there has been an increase in the number of people who access the region in hopes of reaching its summit, often with tragic results in which weather has been a contributing factor [Krakauer, 1999].
 Despite this interest, the remoteness of the Mount Everest region and its harsh environment have limited data collection activities that could contribute to the study of its weather and climate. A notable exception is the pioneering work by Reiter and Heuberger  during the Austrian Expedition to Cho-Oyu, a 8,201 m high mountain just west of Mount Everest, in 1954. In recent years, the ‘Pyramid’ meteorological observatory has been established near Mount Everest at an elevation of 5,050 m [Bertolani and Bollasina, 2000]. Although these efforts have contributed to our knowledge of the meteorology of the region, there is still a pressing need for data from the summit region. In an attempt to satisfy this demand, a portable weather station was installed at an elevation of 7,986 m on the South Col of Mount Everest in early May 1988 [Lau, 1998].
 In this paper, we will compare the observations of pressure and temperature collected at the South Col site to the reanalyses from the National Centers for Environmental Prediction (NCEP) and the European Center for Medium-Range Weather Forecasts (ERA-40) [Kalnay et al., 1996; Simmons and Gibson, 2000]. As we shall show, both reanalyses are able to capture much of the synoptic-scale variability in the temperature and pressure at the South Col site. In addition, we show that a period of high impact weather recorded by the weather station and experienced by people on the mountain was the result of convection associated with a passing jet streak and tropopause fold. The high elevation of the South Col site, which allows for the sampling of the upper-troposphere, makes it a unique location with which to observe these events.
2. Weather at the South Col
 For this paper, we will make use of observations of atmospheric temperature and pressure made by a portable weather station [Lau, 1998] that was deployed at an elevation of 7,986 m on the South Col of Mount Everest. Up to 30 measurements a day were recorded and transmitted via the ARGOS satellite telecommunication network to a base station at MIT. Raw instrument counts received from the station were converted into measurements of temperature and pressure using calibration curves made prior to the instrument's deployment [Lau, 1998]. The South Col weather station operated from May 5 to August 26 1998. For this research, the asynchronous measurements were scanned for outliers and then converted into 6-hourly mean values for the period of the station's operation. The diurnal cycle in the temperature from the station indicated that solar heating of the instrument was most likely leading to anomalously warm temperatures during the day [Linacre, 1992]. As a result, we chose to look at minimum daily temperatures derived from the 6-hourly mean values.
 In Figure 1 we present the daily mean pressure and the daily minimum temperature at the South Col site as observed by the weather station. Both time series include a trend towards higher values as the summer progresses. This is the result of the warming and concomitant thickening of the troposphere. Around July 1st, the weather station was covered in snow terminating the station's measurement of atmospheric temperature [Lau, 1998]. There is evidence of variability on the timescale of several days to one week in the South Col data. The most notable event occurred around May 10 when the pressure was observed to fall to 379 mb – the lowest pressure observed during the station's operation. This event resulted in a major impact to operations on Mount Everest with high winds and heavy snowfall reported (W. Berg, personal communication, 1998). Another event occurred around May 30 and soon after it, there appeared to be a qualitative change in the nature of the variability in pressure with more frequent but lower amplitude fluctuations being the norm. We propose that this transition was associated with the establishment of the summer monsoon over the region. Both the May 10 and May 30 events were associated with a minimum in temperature and a subsequent warming. The warming after the May 30 event was particularly large and may have been associated with the oncoming monsoon.
Figure 1 also includes the extractions of daily mean pressure and daily minimum temperature at the South Col site from the NCEP and ERA-40 reanalyses. The data from both reanalyses was interpolated in the horizontal to the location of Mount Everest and in the vertical to the height of the South Col. In doing this interpolation, it must be emphasized that the topography in the vicinity of Mount Everest is not resolved in the reanalyses. As a result, there will be variability in the observed pressure and temperature fields that is not captured in these datasets. In comparison to the observed pressure, both reanalyses had RMS errors of approximately 1.6 mb and a correlation coefficient of approximately 0.9. For the minimum daily temperature, the corresponding values are approximately 4.5°C and 0.75. This difference reflects the fact that the pressure is not as influenced by the presence/absence of the mountain as is the temperature. Both the May 10 and May 30 events are clearly captured by the reanalyses, although the magnitude of the pressure drop during the events was underestimated by approximately 40% with respect to the observations. In the latter part of the time period, there were some noticeable events in the observed pressure that were not resolved in the reanalyses. Most notable is the period of higher pressure around July 1. This change in the nature of the correlation may be related to the onset of the monsoon.
 The total column ozone as measured by the space-based TOMS instrument provides a succinct dataset with which to identify intrusions of stratospheric air into the upper-troposphere [Hudson et al., 2003]. In Figure 2, we present the time series of the TOMS total column ozone over Mount Everest for the period May to August 1998. There is a trend towards lower column ozone values as the summer progresses. This is associated with the warming and thickening of the troposphere. Superimposed on this background are numerous events where high total column ozone values were observed over Mount Everest. The largest of these events occurred around May 10 during which values in excess of 300 Dobson Units were observed. Based on a climatology of TOMS-derived total column ozone over Mount Everest (not shown), values of this magnitude occur less then 5% of the time during mid-May. Another peak was observed around May 30. As described above, significant weather events were observed to occur at the South Col on these dates. There are a number of other high column ozone events during the remainder of the period of interest. Inspection of Figure 1 indicates no clear evidence of these events in the observations. We hypothesize that the lack of expression of these events in the weather station data is the result of the thickening of the atmosphere that raised the height of the tropopause thereby reducing the magnitude of the signal of the stratospheric intrusions at the South Col site.
3. The May 10 1998 High Impact Weather Event
 We now turn our attention to a determination of the large-scale atmospheric circulation associated with the May 10th event. Motion in the upper-troposphere, where the summit of Mount Everest is located, is quasi-adiabatic and occurs, to first order, along surfaces of constant potential temperature [Hoskins et al., 1985]. Therefore to visualize the large-scale circulation during this event, we have chosen to view the flow on the 340 K potential temperature surface. During May, the height of this surface in the Mount Everest region is approximately 9–10 km. Figure 3 shows the horizontal wind and potential vorticity fields on the 340 K potential temperature surface from the ERA-40 reanalysis at 0 GMT on May 10 1998. Values of potential vorticity in excess of 1 or 2 PVU (1 PVU = 10−6 K s−2 kg−1) typically indicate the presence of stratospheric air [Davies and Schuepbach, 1994; Morgan and Nielsen-Gammon, 1998] and the 340 K surface intersects the tropopause in the vicinity of Mount Everest. On this day, an elongated region of high wind speed known as a jet streak [Keyser and Shapiro, 1986; Doswell and Bosart, 2001] is present along the tropopause to the south and west of Mount Everest.
 On the northern flank of the jet streak, there exists a banded feature characterized by air of stratospheric origin that can be seen extending southwards from mid-latitudes towards Mount Everest. This latter feature is known a stratospheric intrusion or tropopause fold [Keyser and Shapiro, 1986; Appenzeller and Davies, 1992]. As a result of the propagation of the jet streak, the ERA-40 wind speeds on the 340 K surface in the vicinity of Mount Everest increased from 10 m s−1 on the 7th to over 40 m s−1 on the 10th. Actual wind speeds near the summit of Mount Everest may have been higher as a result of a systematic bias of approximately 10% in jet stream wind speeds that exists in global analyses as well as unresolved local orographic effects [Barry, 1992; Tenenbaum, 1996]. Over the next 3 days, the jet streak and the associated intrusion of stratospheric air propagated in an eastward direction past the Mount Everest region. Peak values of potential vorticity near the summit of Mount Everest were on the order of 1.5 PVU.
 The north east quadrant of a jet streak is a region where convective activity, which is the result of a circulation that develops in the plane perpendicular to the jet streak, preferentially occurs. [Uccellini and Johnson, 1979; Keyser and Shapiro, 1986]. From Figure 3, we can see that Mount Everest was located in this quadrant on May 10. The presence of convective activity in the vicinity of Mount Everest during this event is confirmed in Figure 4, which shows an infra-red image from the NOAA family of polar orbiting satellites at 8 GMT on May 10th. This image shows an outbreak of convection over the Tibetan plateau with the coldest and therefore highest clouds, which are usually associated with the most severe weather, in the region around Mount Everest.
 In this paper, we have analysed the barometric pressure and temperature data from a weather station at the South Col of Mount Everest that was operated from May to August of 1998. We have shown that both the NCEP and the ERA-40 reanalyses are able to capture much of the synoptic-scale variability in the pressure and temperature fields at the South Col site. We have also argued that the high impact weather event observed by the weather station around May 10th was associated with the passage of a jet streak and an intrusion of ozone-rich stratospheric air into the upper-troposphere. Based on values of potential vorticity, it is likely that air of stratospheric origin was present near the summit of Mount Everest during this event.
 We have shown that there was an outbreak of organized convection in the vicinity of Mount Everest during this event. We propose that the orientation of the jet streak was such that the synoptic-scale rising motion associated with it was responsible for the triggering of the convection. We are therefore drawn to the conclusion that the presence of a jet streak in the vicinity of Mount Everest may have been the trigger that initiated the convective component of the severe weather observed this event [Doswell and Bosart, 2001]. In all likelihood this rising motion was enhanced by more local ascent of orographic origin as well as that associated with the diurnal cycle of solar heating over the Tibetan plateau [Yanai and Li, 1994]. This convective motion is not be well represented by the reanalyses [Newman et al., 2000] and its absence may be a source of the discrepancy in barometric pressure between these datasets and the South Col observations (Figure 1).
 Our results suggest that it is possible to use the reanalyses and the models upon which they are based to study and forecast high impact weather systems in the Mount Everest region especially in the pre-monsoon period. This is of great practical benefit as this is the period during which much of the activity on the mountain takes place [Huey and Salisbury, 2003].
 The authors thank Michael Hawley from the MIT Media Laboratory for access to the South Col observations. The NCEP Reanalysis data provided by the NOAA-CIRES Climate Diagnostics Center, Boulder, Colorado, USA. The ERA-40 Reanalysis data was provided by the European Center for Medium Range Weather Forecasting. The authors would like to thank Ian Renfrew, Mark Stastna and 2 anonymous reviewers for their helpful comments.