QBO data and assimilation

Intro

This document and the attached data sets are compiled for activities in SCOUT-O3 and CCMVal [--> CCMVal]. Within both projects it is planned to compare climate chemistry simulations for the past few decades, which should include "slow" modes of variability following the observations. Slow refers to time scales longer than a year, as typically generated by El Nino/La Nina, QBO, solar sunspot cycle, volcanic eruptions, trends in emissions etc. Most of these forcings are external to an atmosphere-only model, with the exception of the QBO. The QBO is generated by internal processes of the atmosphere, but is not simulated in most current climate chemistry models (CCMs). Consequently QBO effects on circulation and chemistry are often neglected. However, assimilation of the zonal wind in the QBO domain can add the QBO to the system, thus providing for example its effects on transport and chemistry. For this reason it is planned to include the QBO in climate chemistry experiments for SCOUT-O3 and CCMVal by means of assimilation, so that the QBO follows a prescribed external QBO. The following sections describe the construction of a dataset that has been used for this purpose. This is followed by a short discussion of the pros and cons of the assimilation of the QBO.

Observations of the QBO

The QBO in zonal wind (see Baldwin et al. (2001) for a recent review) is directly observed in operational wind measurements by rawinsondes at equatorial meteorological observatories. Ideally these measurements are made within 2 degree latitude from the equator. Barabara Naujokat of the stratospheric research group at the Free University Berlin has collected and processed radio sonde measurements from 1953 onward from Canton Island, Gan(Maledives) and Singapore as compiled in Table 1 (Naujokat, 1986; Labitzke et al., 2002). These profiles are commonly used to describe the QBO.

Table 1. Stations, their geographic coordinates and active time periods until December 2004.

Zonal wind data of these stations have been analyzed and gridded on 7 standard levels from 70 to 10 hPa until December 1986. Later profiles have been analyzed for characteristic points and interpolated to a higher resolved 14 levels grid from 90 to 10 hPa (Table 2). Observations are usually available twice daily at 00 and 12. Only earlier years have less frequent observations, sometimes less than 10 per month. Additional values for 100 hPa, as obtained from CLIMAT TEMP data, are added in 1997 and later. Data files, as distributed by Barbara Naujokat, are compiled in Table 3. Flags indicate if data are missing or if the number of observations in a month is less than 10.

Table 2. Pressure levels used for the low and high resolution data sets with 7 or 14 levels, respectively, and the extended dataset with 19 levels. Standard level are in bold, levels above 10 hPa in italics.

Table 3. Data files for Canton Island, Gan/Maledives and Singapore from January 1953 to December 2004. Until 1986 the data are gridded on 7 standard levels.

Merged datasets

The QBO shows a high degree of zonal symmetry (see for example in Naujokat, 1986). This allows to merge the equatorial zonal wind profiles of the individual stations into one dataset covering a longer time period. Here, the time series of Canton Island, Gan/Maledives and Singapore are merged at either the low resolution of 7 levels (Figure 1), thus subsampling the profiles after 1986, or at the higher resolution of 14 levels (Figure 2), for which the low resolution profiles are interpolated using 3rd order polynomials (Table 4). 

Table 4. Data files for merged datasets 1953-2004 on 7 leves or 14 levels.

Figure 1. QBO in observed monthly mean zonal wind in m/s on 7 levels from 1953 to 2004.

Figure 2. QBO in observed monthly mean zonal wind in m/s on 14 levels from 1953 to 2004.

Vertical extensions

Table 5. Data file for merged datasets 1953-2004 on 7 leves or 14 levels.

Figure 3. Observed QBO in zonal wind in m/s wirthin red box extended vertically by the propagation method upward to 3 hPa and downward to 90 hPa, for 1953 to 2004.

Extensions in time

For some purposes it is useful to have arbitrarily long QBO time series based on observed QBO wind profiles, which are available for about 50 years. This calls for a method to expand the observed time series in a reasonable way. Here you find a description of a pragmatic method that consists in the repetition of previous periods of the observed QBO. The following recipe has been used for the purpose of extending the observed QBO time series after January 2005:

1. Use the observed time series up to January 2005 as observed
2. Search for zonal wind profiles which deviate as little as possible in the vertically averaged mean square difference from the observed profile in June 2005.  This search results in a list of 11 years:
   1960, 1962, 1965, 1970, 1972, 1974, 1979, 1989, 1991, 1996, 1998
3. Make a linear transition from January 2005 to July 2005 so that for the 1960 case:
• U_{ext(2005->1960)}(Jan.2005)=U(Jan.2005)
• U_{ext(2005->1960)}(July2005)=U(July1960)
4. Then append the remaining timeseries of Aug.1960 to December 2004 to U_{ext(2005->1960)}.

References

Baldwin M. P. et al. , The quasi-biennial oscillation. Rev. Geophys., 39, 179-229, 2001.

Giorgetta, M. A., and L. Bengtsson, Potential role of the quasi-biennial oscillation in the stratosphere-troposphere exchange as found in water vapor in general circulation model experiments, J. Geophys. Res., 104, 6003-6019, 1999.

Giorgetta, M. A., E. Manzini, and E. Roeckner, Forcing of the quasi-biennial oscillation from a broad spectrum of atmospheric waves, Geophys. Res. Lett., 29, 10.1029/2002GL014756, 2002.

Labitzke et al. 2002: The Berlin stratospheric data series. Meteorological Institute, Free University of Berlin, CD-ROM.

Naujokat, B., 1986: An update of the observed quasi-biennial oscillation of the stratospheric winds over the tropics. J. Atmos. Sci., 43, 1873-1877.

Impressum / Imprint - Datenschutzerklärung / Privacy Policy