Mesoscale temperature fluctuations in the Southern Hemisphere stratosphere

Isentrope surfaces in the Southern Hemisphere stratosphere reveal that air parcels undergo mesoscale temperature fluctuations that depend on latitude and season. The largest temperature fluctuations occur at high latitude winter, whereas the smallest fluctuations occur at high latitude summer. This is the same pattern found for the Northern Hemisphere stratosphere. However, the amplitude of the seasonal dependence in the Southern Hemisphere is only 37% of the Northern Hemisphere's seasonal amplitude.


Introduction
A companion article, "Mesoscale Temperature Fluctuations in the Stratosphere" (Gary, 2006), describes an analysis of stratospheric isentrope "wrinkles" in the North-10 ern Hemisphere using measurements by the Microwave Temperature Profiler (MTP) aboard NASA ER-2 and DC-8 aircraft. It was found that the amplitude of "mesoscale temperature fluctuations" depend on the following four independent variables: latitude, season, underlying topography and altitude. Figure 1, taken from that article, shows the latitude dependence of mesoscale fluctuation amplitude (MFA) for two seasons. 15 The present article is also based on measurements by the MTP. However, all flights in this analysis are in the Southern Hemisphere (SH). The purpose of the present analysis is to determine if MFA in the SH has the same dependencies on latitude and season that were found for the Northern Hemisphere (NH). The NH MFA analysis employed ER-2 and DC-8 measurements at altitude regimes centered on 19 and 11 km, 20 and therefore permitted an investigation to be made of the altitude dependence of MFA. All SH data were made using only an ER-2 so it is not possible in this study to evaluate the dependence of SH MFA on altitude. It was also not possible to evaluate the dependence of SH MFA on topography because all flights were over ocean.
The archive for this study consists of 22 ER-2 flights, most of which were based in and since MFA was suspected to depend on latitude, each flight was divided into segments with a latitude span of ∼10 degrees. This usually meant that each flight was divided into four segments of 1.5-h duration, producing a total of 81 flight segments; 75 were over ocean and 6 were over New Zealand. The median altitude for all flight segments is 19.2 km, with a range of 17.8 to 20.4 km.

2 Remote sensor
The Microwave Temperature Profiler aboard the ER-2 aircraft is a passive radiometer operating at frequencies 56.66 and 58.80 GHz. At these frequencies atmospheric opacity is dominated by oxygen molecules. At 19 km altitude the opacities for the two frequencies are ∼0.4 and 0.9 Nepers/km; the corresponding atmospheric source func-10 tions have exponential weighting functions with ranges of 2.5 and 1.1 km, respectively. A 45-degree reflector is rotated through a sequence of 10 elevation angles, including the horizon. At each elevation setting brightness temperature is measured at both frequencies. After measurements of sky brightness temperature are made, the reflector is rotated for viewing of an ambient target. An observing cycle requires ∼20 s to com-15 plete; this means that the MTP produces an altitude temperature profile at intervals of ∼4 km along the flight path. Statistical retrieval coefficients are calculated prior to deployment and these are used to convert the 20 observed brightness temperatures per observing cycle to a temperature profile. The derived profile of temperature versus altitude extends from ∼3 below flight level to ∼4 km above. Additional information about Introduction turn-around, corresponding to ∼26 degrees of latitude for a north/south flight. The outbound and inbound flight segments are at different altitudes, and the MTP data for each is used to create an "isentrope altitude cross-section" (IAC). It is possible to create a synoptic scale IAC for the same flight path, at the time of the flight, using a global data base of analyzed data from radiosondes and satellite temper-5 ature profiles. When this is done the two IACs are in approximate agreement except for two differences: 1) the MTP IAC shows much more spatial structure since it has a horizontal spatial resolution that is about 100 times better, ∼4 km versus ∼400 km, and 2) there are altitude offsets between the two IACs that persist for hundreds of km, typically, before they change sign. Because the ER-2 payload includes a high quality 10 outside air temperature (OAT) sensor it is possible to determine that the MTP IAC is invariably the correct one. Even the short timescale high spatial frequency IAC structure of the MTP near flight level is in agreement with the OAT sensor. For this reason it is preferable to smooth the MTP data instead of using the synoptic scale analyzed data for the purpose of determining a synotpic IAC. To convert a high spatial resolution 15 MTP IAC to a synoptic version the measured IAC is smoothed using a 400-km boxcar. A second 400-km boxcar smoothing step is performed, which produces structures that resemble a Fourier decomposition and reconstitution using only the 400 km and longer components. Figure 2 illustrates a typical relationship between MTP-measured isentropes and their 400 km double boxcar synoptic scale representation.

20
By subtracting the MTP-measured IAC from its counterpart synoptic scale version (the 400-km double boxcar smoother version), it is possible to produce a "mesoscale only" IAC. This IAC is free of the systematic errors that would exist if a synoptic scale IAC were used that had been produced from the satellite and RAOB-based analyzed data. It is a simple procedure to characterize the fluctuations of any flight segment of 25 "mesoscale only" isentrope altitude for a specific isentrope by converting the values to a histogram and measuring the histogram's full-width at half-maximum, FHWM. This was done using a spreadsheet, where fitting the histogram with a Gaussian function was done manually. In almost every case the Gaussian function is a good fit to the observed 9170 MFA value was entered into a spreadsheet with latitude and date. A season parameter called "Winterness" was calculated for each entry that varied in a sinusoidal manner from 0.0 on 15 July to 1.0 on 15 January (described in detail in the companion article). An MFA was categorized as belonging to "summer" if Winterness <0.4, and it was categorized as "winter" when Winterness >0.6. Winterness values between 0.4 and 0.6 were categorized as "spring." Figure 4 shows the individual MFA measurements plotted versus latitude, with different symbols for winter and summer seasons. This figure exhibits a much greater seasonal overlap of individual MFA measurements for all latitude regions compared with measurements for NH (Fig. 1). To aid in illustrating the presence of a seasonal 15 effect the MFA entries for a season were arranged by latitude in groups of ∼6 and a median combine of MFA was calculated. The median combine MFA values show a clear pattern of greater MFA in winter compared with summer. There is also weak evidence for an increasing seasonal difference with latitude. Both correlations are present in the NH MFA measurements (Fig. 1).

20
The straight line fits to the summer and winter MFA values in the SH data (Fig. 4)  difference between the SH and NH measurements is that the latitude dependence for the SH is 27±8% of the NH latitude dependence. A multiple regression model was created that incorporates latitude and the seasonal parameter as independent variables. The dependent variable was MFA corrected for altitude. Since the range of altitudes was small an altitude correction exponent could 5 not be solved for, so the correction equation found from NH data was used for correcting the SH MFA: where P is air pressure at flight altitude. The range of altitude corrections was 17% (maximum minus minimum correction).

10
The uncertainty of this correction is considered small compared with the correction, and it is especially small compared with the uncertainty of individual MFA measurements. The plot of measured versus model MFA in the SH is shown in Fig. 5