<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE article PUBLIC "-//NLM//DTD Journal Publishing with OASIS Tables v3.0 20080202//EN" "journalpub-oasis3.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:oasis="http://docs.oasis-open.org/ns/oasis-exchange/table" dtd-version="3.0">
  <front>
    <journal-meta>
<journal-id journal-id-type="publisher">ACP</journal-id>
<journal-title-group>
<journal-title>Atmospheric Chemistry and Physics</journal-title>
<abbrev-journal-title abbrev-type="publisher">ACP</abbrev-journal-title>
<abbrev-journal-title abbrev-type="nlm-ta">Atmos. Chem. Phys.</abbrev-journal-title>
</journal-title-group>
<issn pub-type="epub">1680-7324</issn>
<publisher><publisher-name>Copernicus Publications</publisher-name>
<publisher-loc>Göttingen, Germany</publisher-loc>
</publisher>
</journal-meta>

    <article-meta>
      <article-id pub-id-type="doi">10.5194/acp-16-6681-2016</article-id><title-group><article-title>Representation of the tropical stratospheric zonal wind in global
atmospheric reanalyses</article-title>
      </title-group><?xmltex \runningtitle{Representation of the tropical stratospheric zonal wind}?><?xmltex \runningauthor{Y. Kawatani et al.}?>
      <contrib-group>
        <contrib contrib-type="author" corresp="yes" rid="aff1">
          <name><surname>Kawatani</surname><given-names>Yoshio</given-names></name>
          <email>yoskawatani@jamstec.go.jp</email>
        <ext-link>https://orcid.org/0000-0003-3260-8830</ext-link></contrib>
        <contrib contrib-type="author" corresp="no" rid="aff2">
          <name><surname>Hamilton</surname><given-names>Kevin</given-names></name>
          
        <ext-link>https://orcid.org/0000-0002-3560-8097</ext-link></contrib>
        <contrib contrib-type="author" corresp="no" rid="aff1">
          <name><surname>Miyazaki</surname><given-names>Kazuyuki</given-names></name>
          
        <ext-link>https://orcid.org/0000-0002-1466-4655</ext-link></contrib>
        <contrib contrib-type="author" corresp="no" rid="aff3">
          <name><surname>Fujiwara</surname><given-names>Masatomo</given-names></name>
          
        <ext-link>https://orcid.org/0000-0001-5567-4692</ext-link></contrib>
        <contrib contrib-type="author" corresp="no" rid="aff4">
          <name><surname>Anstey</surname><given-names>James A.</given-names></name>
          
        </contrib>
        <aff id="aff1"><label>1</label><institution>Japan Agency for Marine-Earth Science and Technology, Yokohama,
236-0001, Japan</institution>
        </aff>
        <aff id="aff2"><label>2</label><institution>International Pacific Research Center, University of Hawaii, Honolulu,
96822, USA</institution>
        </aff>
        <aff id="aff3"><label>3</label><institution>Faculty of Environmental Earth Science, Hokkaido University, Sapporo,
060-0810, Japan</institution>
        </aff>
        <aff id="aff4"><label>4</label><institution>Canadian Centre for Climate Modelling and Analysis, University of
Victoria, Victoria, V8W 2Y2, Canada</institution>
        </aff>
      </contrib-group>
      <author-notes><corresp id="corr1">Yoshio Kawatani (yoskawatani@jamstec.go.jp)</corresp></author-notes><pub-date><day>2</day><month>June</month><year>2016</year></pub-date>
      
      <volume>16</volume>
      <issue>11</issue>
      <fpage>6681</fpage><lpage>6699</lpage>
      <history>
        <date date-type="received"><day>27</day><month>January</month><year>2016</year></date>
           <date date-type="rev-request"><day>1</day><month>February</month><year>2016</year></date>
           <date date-type="rev-recd"><day>10</day><month>May</month><year>2016</year></date>
           <date date-type="accepted"><day>19</day><month>May</month><year>2016</year></date>
      </history>
      <permissions>
<license license-type="open-access">
<license-p>This work is licensed under a Creative Commons Attribution 3.0 Unported License. To view a copy of this license, visit <ext-link ext-link-type="uri" xlink:href="http://creativecommons.org/licenses/by/3.0/">http://creativecommons.org/licenses/by/3.0/</ext-link></license-p>
</license>
</permissions><self-uri xlink:href="https://acp.copernicus.org/articles/.html">This article is available from https://acp.copernicus.org/articles/.html</self-uri>
<self-uri xlink:href="https://acp.copernicus.org/articles/.pdf">The full text article is available as a PDF file from https://acp.copernicus.org/articles/.pdf</self-uri>


      <abstract>
    <p>This paper reports on a
project to compare the representation of the monthly-mean zonal wind in the
equatorial stratosphere among major global atmospheric reanalysis data sets.
The degree of disagreement among the reanalyses is characterized by the
standard deviation (SD) of the monthly-mean zonal wind and this depends on
latitude, longitude, height, and the phase of the quasi-biennial oscillation
(QBO). At each height the SD displays a prominent equatorial maximum,
indicating the particularly challenging nature of the reanalysis problem in
the low-latitude stratosphere. At 50–70 hPa the geographical distributions
of SD are closely related to the density of radiosonde observations. The
largest SD values are over the central Pacific, where few in situ
observations are available. At 10–20 hPa the spread among the reanalyses
and differences with in situ observations both depend significantly on the
QBO phase. Notably the easterly-to-westerly phase transitions in all the
reanalyses except MERRA are delayed relative to those directly observed in
Singapore. In addition, the timing of the easterly-to-westerly phase
transitions displays considerable variability among the different reanalyses
and this spread is much larger than for the timing of the
westerly-to-easterly phase changes. The eddy component in the monthly-mean
zonal wind near the Equator is dominated by zonal wavenumber 1 and 2
quasi-stationary planetary waves propagating from midlatitudes in the
westerly phase of the QBO. There generally is considerable disagreement among
the reanalyses in the details of the quasi-stationary waves near the Equator.
At each level, there is a tendency for the agreement to be best near the
longitude of Singapore, suggesting that the Singapore observations act as a
strong constraint on all the reanalyses. Our measures of the quality of the
reanalysis clearly show systematic improvement over the period considered
(1979–2012). The SD among the reanalysis declines significantly over the
record, although the geographical pattern of SD remains nearly constant.</p>
  </abstract>
    </article-meta>
  </front>
<body>
      

<sec id="Ch1.S1" sec-type="intro">
  <title>Introduction</title>
      <p>The dynamics governing the circulation in the tropical stratosphere have
attracted much interest over the years (e.g., Wallace, 1973; Baldwin et al.,
2001). As in other regions of the middle atmosphere, the day-to-day and
higher-frequency variations of the flow in the tropical stratosphere are
believed to be dominated by a spectrum of vertically propagating waves
excited in the troposphere. What makes the low-latitude stratosphere so
remarkable is that the forcing of the zonal-mean flow by these waves leads to
the very large-amplitude, low-frequency quasi-periodic cycle known as the
quasi-biennial oscillation (QBO). In the tropical stratosphere, the QBO
clearly dominates other aspects of interannual variability and even swamps
the annual and semiannual variations in the zonal-mean circulation, at least
up to <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 3 hPa. Although rooted in the tropics, the QBO has global
impacts. The QBO strongly influences interannual variations in circulation
and composition throughout the stratosphere. The QBO also affects the
circulation at the Earth's surface and is an important consideration in
extended-range weather forecasts (e.g., Baldwin et al., 2001). Recently, Yoo
and Son (2016) indicate the QBO exerts greater influence on the Madden–Julian
Oscillation than does the El Niño–Southern Oscillation.</p>
      <p>The state of the QBO up to the middle stratosphere can be characterized by
the time series of monthly-mean, near-equatorial zonal winds at levels
between 10 and 70 hPa maintained by the Free University of Berlin (FUB; e.g.,
Naujokat, 1986) since 1953. The monthly values in the FUB series are based on
operational balloon soundings, and the FUB record has been stitched together
from such observations at Canton Island (2.8<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S, 172<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> W
from January 1953 to August 1967), Gan (0.7<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S, 73<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> E from
September 1967 to December 1975), and Singapore (1.4<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N,
104<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> E since 1976). The high quality of these balloon data, and the
close proximity of the stations to the Equator, has led FUB series to be
widely used, despite being based on only a single station each month (and
despite modest inhomogeneities that the changes of station location may
introduce into the record; see Sect. 3 below).</p>
      <p>Global atmospheric analyses that assimilate all available satellite remote
sensing and in situ observations are another potential source of information
about the QBO and other aspects of the circulation in the tropical
stratosphere. Data assimilation is the technique for combining different
observational data sets with a model, by considering the characteristics of
each measurement and taking into account errors in both the measurements and
the model (e.g., Kalnay, 2003). Advanced data assimilation schemes like the
4D-Var technique use the information provided by various measurements, such
as radiosonde and satellite-derived measurements, and propagates it, in time
and space, from a limited number of observable variables to a wide range of
meteorological variables to provide global fields that are dynamically
consistent and in agreement with the observations. Meteorological reanalyses
have been conducted at operational centers using various approaches, which
ingest a variety of observations over the period of each reanalysis product.
Differences in the forecast model, assimilated measurements, and data
assimilation technique used for producing reanalysis data sets can lead to
differences in their representation of the mean state, variability, and
long-term trend of atmospheric fields.</p>
      <p>A number of factors combine to make the global meteorological analysis
process particularly challenging in the tropical middle atmosphere.
<list list-type="custom"><list-item><label>i.</label><p>One challenge is the relative paucity of in situ data, even when attention
is restricted to levels at and below 10 hPa (i.e., the usual upper bound for
most operational balloon soundings). There are large ocean areas in the
tropics with no balloon observations. Even over land areas, the observations
in the stratosphere at many stations are sparse. At stratospheric levels, the
zonal wind measurements at many stations near the Equator tend to have short
overall records or records with many months that have no observations (or
not enough to compute a stable monthly mean). This leads to gaps in time
series of monthly-mean winds (e.g., Hamilton, 1984; Kawatani and Hamilton,
2013).</p></list-item><list-item><label>ii.</label><p>Near the Equator the Coriolis parameter is small, so observations of the
temperature from satellite remote sensing do not constrain the wind field as
strongly as at higher latitudes. Even when we assume the near-equatorial flow
really is close to thermal wind balance (Reed, 1962; Randel et al., 1999), the
computed geostrophic wind shears are extremely sensitive to small errors in
the observed temperatures.</p></list-item><list-item><label>iii.</label><p>The flow in the tropical stratosphere exhibits variations on very small
vertical scales. This limits the usefulness of the relatively
coarse-resolution satellite remote-sensing temperature retrievals. Even the
monthly-mean zonal wind in this region displays thin layers where the wind
can change by <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 30 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> over <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 3 km. Satellite radiances
used in global assimilations have an effective vertical resolution of several
kilometers (e.g., Huesmann and Hitchman, 2003). Huesmann and Hitchman (2003)
note that in such shear regions the assimilation scheme will have to
reconcile the strong wind shears measured directly by balloons with the
somewhat weaker thermal winds consistent with the satellite-derived
temperature gradients (artificially damped due to the coarse vertical
resolution).</p></list-item><list-item><label>iv.</label><p>Modern data assimilation approaches use general circulation model (GCM)
simulations to obtain background (i.e., first guess) information and to
determine data assimilation analysis increments. The tropical stratosphere is
perhaps the only region of the atmosphere where most free-running GCMs have
simulations with zero-order errors in the zonal-mean circulation (i.e.,
nothing even resembling a realistic QBO). This is because most GCMs display
fairly steady, weak prevailing zonal winds due to failure to reproduce a
spontaneous QBO. Such models typically relax any QBO-like zonal winds in the
initial condition towards the model climatology (e.g., Hamilton and Yuan,
1992; Saha et al., 2006; Boer and Hamilton, 2008) and so the persistent model
bias could act against the data assimilation analysis increments and damp the
QBO signals as introduced by data assimilation in the reanalysis fields. It
should be noted that the situation is now slowly changing and several models
have simulated a fairly realistic spontaneous the QBO in the equatorial
stratosphere. Such models typically either employ fine vertical resolution as
well as representation of momentum transports from small-scale non-stationary
gravity waves either through high horizontal resolution (Kawatani et al.,
2010, and reference therein) or by parameterization (Orr et al., 2010;
Kawatani and Hamilton, 2013; Anstey et al., 2016).</p></list-item></list>
These special challenges in data assimilation for the tropical stratosphere
have led to problems in actual global assimilation products, although there
have been clear improvements over the last 2 decades in the ability of
analysis systems to represent this region of the atmosphere. Trenberth (1992)
documents a total misrepresentation of the QBO in the tropical stratosphere
of ECMWF operational global analyses in the early 1980s, but finds that the
situation improved considerably after changes in the assimilation scheme were
introduced in May 1986.</p>
      <p>In the 1990s some major meteorological centers started producing
retrospective reanalysis products that employed multivariate statistical data
assimilation methods such as optimal interpolation  and a three-dimensional
variational (3D-VAR) scheme to combine all available data over some extended
period. Such reanalyses have obvious advantages for research applications
requiring the most homogeneous possible data set throughout an extended
period. Two of the early major products were the ECMWF “ERA-15” reanalysis
(covering 1979 to 1994) and the first NCAR/NCEP reanalysis (covering
1948–near present). The QBO in the tropical stratosphere in these products
was examined by Pawson and Fiorino (1998, 1999) and Huesmann and
Hitchman (2001, 2003), who found that the reanalyses displayed equatorial QBO
variations that were clearly smaller in amplitude than in the real
atmosphere. Randel et al. (2004) indicated that QBO variations in temperature
and zonal wind were underestimated in most reanalyses available at that time,
as compared to Singapore radiosonde data. They found that only ERA-40 had
realistic zonal wind amplitudes above 30 hPa.</p>
      <p>Some centers have now produced multiple reanalyses covering the same (or
overlapping) periods; these more recent products are derived with updated
data assimilation systems. Notable among these are the ECMWF ERA-40
(September 1958 to August 2002) and ERA-Interim (1979 to present). These
products have improved representation of the QBO amplitude over that in
ERA-15 (e.g., Pascoe et al., 2005; Baldwin and Gray, 2005). Baldwin and
Gray (2005) compared the FUB zonal wind with ERA-40 data and showed that the
zonal mean equatorial ERA-40 wind is quite close to that indicated in the FUB
data.</p>
      <p>Additional major reanalysis data sets have become available recently. The
present paper reports on one of the studies contributing to the SPARC
Reanalysis Intercomparison Project (S-RIP; Fujiwara and Jackson, 2013; Errera
et al., 2015), which is focused on evaluating reanalysis output for the
stratosphere. We compare the representation of the circulation in the
tropical stratosphere among several contemporary reanalysis products and
validate the reanalyses against in situ observations. We have mostly limited
our attention to just a single aspect of the circulation, namely the
monthly-mean zonal wind. We limit our consideration to low latitudes
(20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N–20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S) and a height range (10–70 hPa) in which
the QBO is strong and where substantial balloon data are available for
comparison. We have also restricted our attention to the period starting in
1979 when NOAA operational satellite radiance observations became available
and were incorporated as an important data source in all the reanalyses. We
appreciate that other aspects of the reanalyses in this region are also of
interest, notably the ability of the reanalyses to represent accurately the
large-scale transport circulation in the stratosphere (e.g., Coy and
Swinbank, 1997; Abalos et al., 2015; Miyazaki et al., 2016), but these
aspects will be left for future research.</p>
      <p>Each reanalysis uses a different forecast model and assimilation scheme, and
the types and numbers of assimilated observational data are also different
among reanalyses. Furthermore, it is not feasible to determine exactly what
observational data were actually assimilated at each data assimilation
analysis step (e.g., what data quality control and bias correction
procedures were actually applied). These complications make it somewhat
difficult to attribute conclusively all the differences among the reanalyses
products. However, we believe it is interesting to investigate
representations of key phenomena in the reanalyses and we hope that such
investigation will contribute to basic understanding and to improving future
reanalysis products. We will show that, even with our somewhat narrow focus,
some interesting conclusions concerning the representation of the tropical
stratospheric circulation will emerge.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F1" specific-use="star"><caption><p>Time-height cross section of the FUB zonal wind and each reanalysis
zonal wind over Singapore (104<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> E, 1.4<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N) from 1979 to
2012. The color intervals are 5 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>.</p></caption>
        <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f01.pdf"/>

      </fig>

      <p>The outline of this paper is as follows. Section 2 briefly describes the
reanalysis products that we evaluated and the station balloon data we
employed. Section 3 investigates differences of the tropical zonal wind among
reanalyses. Section 4 discusses evolution of the differences among reanalyses
with time as different data sources become available. Section 5 summarizes
the study and provides concluding remarks.</p>
</sec>
<sec id="Ch1.S2">
  <title>Reanalysis and radiosonde observation data</title>
      <p>We analyzed monthly-mean zonal wind and temperature in nine sets of global
reanalyses data (NCEP-1, Kalnay et al., 1996; NCEP-2, Kanamitsu et al., 2002;
NCEP-CFSR, Saha et al., 2010; ERA-40, Uppala et al., 2005; ERA-I, Dee et al.,
2011; JRA-25, Onogi et al., 2007; JRA-55, Kobayashi et al., 2015; MERRA,
Rienecker et al., 2011, and MERRA-2, Molod et al., 2015). Monthly-mean data
for 20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N and 10–70 hPa after 1979 are mainly
analyzed except for MERRA-2, which has data only after January 1980. Data
before December 2012 are investigated, except for ERA-40 and NCEP-CFSR, which
are available until August 2002 and December 2010, respectively. With the
exception of MERRA-2, none of the global dynamical models used in the
assimilations would simulate the QBO when in free running mode. The dynamical
model used in producing the MERRA-2 reanalyses is able to simulate a
spontaneous QBO in the tropical stratosphere because it includes quite strong
parameterized momentum fluxes from non-orographic gravity waves (Fig. 3 of
Molod et al., 2015).</p>
      <p>One key observational data set we employed for comparisons is the
near-equatorial monthly-mean values of operational balloon-borne radiosonde
observations compiled by FUB (Freie Universität Berlin, 2016; Naujokat,
1986). The FUB data during 1979–2012 (i.e., period analyzed in this study)
are based on Singapore observations.</p>
      <p>We have also used monthly-mean values of the zonal wind at many other
radiosonde stations provided in the Integrated Global Radiosonde Archive
(NOAA, 2016; Durre et al., 2006). For 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N IGRA
includes balloon data from <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 220 stations. When twice daily
observations are available, the IGRA provides monthly means of observations
at 00:00 and at 12:00 UTC separately. As in Kawatani and Hamilton (2013),
where possible, the monthly values for 00:00 and 12:00 UTC are averaged, but
the single values are used when only 00:00 or only 12:00 UTC data are
available. The diurnal cycle of the wind in the lower stratosphere is
expected to be small, and indeed we have found that generally the
monthly-mean zonal wind data measured at 00:00 UTC are nearly identical to
that at 12:00 UTC.</p>
      <p>In order to compare the data among reanalyses which have different spatial
resolutions, all reanalysis variables are interpolated linearly to the ERA-I
grid (i.e., 1.5<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> resolution in longitude and latitude, 37 vertical
levels from 1000 to 1 hPa). When reanalysis data are compared with FUB/IGRA
observational data, the reanalysis values are interpolated to the station
locations.</p>
</sec>
<sec id="Ch1.S3">
  <title>Differences of the tropical zonal wind among reanalyses</title>
<sec id="Ch1.S3.SS1">
  <title>Dependence of differences on longitude and height</title>
      <p>Figure 1 shows the FUB zonal wind and the zonal wind in each reanalysis
included in this study in Singapore (104<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> E, 1.4<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N) from
1979 to 2012. All the reanalyses have a reasonably close resemblance to the
FUB observations. Notably each reanalysis clearly captures the basic features
of the QBO seen in the FUB data, including the cycle-to-cycle variation in
period and amplitude.</p>
      <p>Figure 2 displays time variations of the zonal wind in each reanalysis at 10,
20, 30, 50, and 70 hPa over Singapore compared with the FUB observations.
Also shown are the root mean square (RMS) differences between FUB and each
reanalysis zonal wind averaged from 10 to 70 hPa. Reanalysis zonal winds are
generally close to the FUB zonal wind. The RMS differences display somewhat
noisy variations but an overall trend to smaller values over time is
apparent. The zonal winds in NCEP-1 and NCEP-2 are generally underestimated,
especially in the 1980s over 20–50 hPa. The most obvious anomaly in the
reanalysis winds at 10 hPa is found in MERRA-2, which exhibits spurious
semiannual variations in the 1980s and in late 1993 particularly during
easterly phase of the QBO (Figs. 1g and 2a). The downward propagation of
westerly semiannual oscillation (SAO) phases is enhanced during  these
periods, which could be caused by overly strong gravity wave forcing (Fig. 3
of Molod et al., 2015). Coy et al. (2016) note that MERRA-2 appears to
overemphasize the annual cycle before 1995.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F2"><caption><p>Time variations of the zonal wind over Singapore at <bold>(a)</bold> 10,
<bold>(b)</bold> 20, <bold>(c)</bold> 30, <bold>(d)</bold> 50, and <bold>(e)</bold> 70 hPa
for FUB observation (black), ERA-40 (light blue), ERA-I (blue), JRA-25 (light
purple), JRA-55 (purple), MERRA (red), MERRA-2 (pink), NCEP-1 (orange),
NCEP-2 (yellow), and NCEP-CFSR (green); <bold>(f)</bold> Root mean square
differences between the FUB and each reanalysis zonal wind, averaged at
70–10 hPa.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f02.pdf"/>

        </fig>

      <p>The RMS differences from FUB values are smallest in ERA-I, while those in
NCEP-1 and NCEP-2 are much larger than those in the other reanalyses. MERRA-2
represents large RMS differences in 1980s (Fig. 2f), mainly due to the
enhanced SAO at 10 hPa. The NCEP-CFSR actually uses the ERA-40 stratospheric
wind profiles as bogus observations in the tropics from 1 July 1981 to
31 December 1998 to obtain a reasonable QBO (Saha et al., 2010). We confirmed
the differences between NCEP-CFSR and ERA-40 during this period are nearly
0. For these reasons, our study focuses mainly on the five reanalyses:
ERA-40, ERA-I, JRA-25, JRA-55, and MERRA. But note that the MERRA-2 zonal
winds show improved representation of the QBO compared to MERRA at
30–50 hPa (Coy et al., 2016).</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F3"><caption><p>Latitude–height cross sections of zonal mean and time mean of the
standard deviation among <bold>(a)</bold> five reanalyses of ERA-40, ERA-I,
JRA-25, JRA-55, and MERRA from 1979 to 2001; <bold>(b)</bold> four reanalysis of
ERA-I, JRA-25, JRA-55, and MERRA from 1979 to 2012; and <bold>(c)</bold> three
reanalysis of ERA-I, JRA-55, and MERRA from 1979 to 2012. The color intervals
are 0.2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> and shaded with values larger than 0.6 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f03.png"/>

        </fig>

      <p>In order to quantify the spread among reanalyses, the standard deviations
(SDs) among the reanalyses are calculated as follows:

                <disp-formula id="Ch1.E1" content-type="numbered"><mml:math display="block"><mml:mrow><mml:mtext>SD</mml:mtext><mml:mo>=</mml:mo><mml:msqrt><mml:mrow><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mi>i</mml:mi><mml:mi>N</mml:mi></mml:munderover><mml:mo>(</mml:mo><mml:msub><mml:mi>u</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>-</mml:mo><mml:mfenced open="[" close="]"><mml:mi>u</mml:mi></mml:mfenced><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup><mml:mo>/</mml:mo><mml:mi>N</mml:mi></mml:mrow></mml:msqrt></mml:mrow></mml:math></disp-formula>

          where <inline-formula><mml:math display="inline"><mml:mi>i</mml:mi></mml:math></inline-formula> labels the individual data sets and there are <inline-formula><mml:math display="inline"><mml:mi>N</mml:mi></mml:math></inline-formula> data sets included.
The square brackets denote the mean over all <inline-formula><mml:math display="inline"><mml:mi>N</mml:mi></mml:math></inline-formula> reanalyses. The SD is
calculated for each month using monthly-mean zonal wind. Because the ERA-40
data are only provided until August 2002, attention will sometimes be
restricted to the SD during January 1979 to December 2001.</p>
      <p>The latitude–height distributions of the zonal mean and time mean of the SD
among five reanalyses (1979–2001) are displayed in Fig. 3a. Large SD is seen
in the upper troposphere and the stratosphere at low latitudes, while the SD
in the extratropical regions is much smaller. Figure 3b and c depict the SD
among four reanalyses (ERA-I, JRA-25, JRA-55, and MERRA) and among the three
latest reanalyses (ERA-I, JRA-55, and MERRA) in 1979–2012. These are
discussed in the next section. In any case, the distributions of the SD are
quite similar.</p>
      <p>In order to examine the SD in more detail later, the SD is divided into zonal
mean components (<inline-formula><mml:math display="inline"><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover></mml:math></inline-formula>; overbar denotes the zonal mean) and
components of deviation from zonal mean (<inline-formula><mml:math display="inline"><mml:mrow><mml:msup><mml:mi>u</mml:mi><mml:mo>′</mml:mo></mml:msup></mml:mrow></mml:math></inline-formula>: hereafter, referred to as eddy
components). Substituting <inline-formula><mml:math display="inline"><mml:mrow><mml:msub><mml:mi>u</mml:mi><mml:mi>i</mml:mi></mml:msub><mml:mo>=</mml:mo><mml:msub><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mi>i</mml:mi></mml:msub><mml:mo>+</mml:mo><mml:msubsup><mml:mi>u</mml:mi><mml:mi>i</mml:mi><mml:mo>′</mml:mo></mml:msubsup></mml:mrow></mml:math></inline-formula> and <inline-formula><mml:math display="inline"><mml:mrow><mml:mfenced open="[" close="]"><mml:mi>u</mml:mi></mml:mfenced><mml:mo>=</mml:mo><mml:mfenced close="]" open="["><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover></mml:mfenced><mml:mo>+</mml:mo><mml:mfenced close="]" open="["><mml:msup><mml:mi>u</mml:mi><mml:mo>′</mml:mo></mml:msup></mml:mfenced></mml:mrow></mml:math></inline-formula> in Eq. (1) and
taking the zonal mean yields:

                <disp-formula id="Ch1.E2" content-type="numbered"><mml:math display="block"><mml:mrow><mml:mover accent="true"><mml:mrow><mml:mo>(</mml:mo><mml:mtext>SD</mml:mtext><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mo>=</mml:mo><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mi>i</mml:mi><mml:mi>N</mml:mi></mml:munderover><mml:mover accent="true"><mml:mrow><mml:mo>(</mml:mo><mml:msub><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mi>i</mml:mi></mml:msub><mml:mo>-</mml:mo><mml:mo>[</mml:mo><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mo>]</mml:mo><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mo>/</mml:mo><mml:mi>N</mml:mi><mml:mo>+</mml:mo><mml:munderover><mml:mo movablelimits="false">∑</mml:mo><mml:mi>i</mml:mi><mml:mi>N</mml:mi></mml:munderover><mml:mover accent="true"><mml:mrow><mml:mo>(</mml:mo><mml:msubsup><mml:mi>u</mml:mi><mml:mi>i</mml:mi><mml:mo>′</mml:mo></mml:msubsup><mml:mo>-</mml:mo><mml:mo>[</mml:mo><mml:msup><mml:mi>u</mml:mi><mml:mo>′</mml:mo></mml:msup><mml:mo>]</mml:mo><mml:msup><mml:mo>)</mml:mo><mml:mn mathvariant="normal">2</mml:mn></mml:msup></mml:mrow><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:mo>/</mml:mo><mml:mi>N</mml:mi><mml:mo>.</mml:mo></mml:mrow></mml:math></disp-formula>

          Here, the first term on the right side represents the zonal mean variance
among reanalyses due to the zonal mean wind component and second term
exhibits the zonal mean variance attributable to eddy components.</p>
      <p>Figure 4 shows horizontal distributions of the SD at 10, 20, 30, 50, and
70 hPa. At 50–70 hPa, large SD is seen in the Indian Ocean, to the east of
the maritime continent, the Atlantic, and especially in the central Pacific.
At 10–30 hPa, however, the SD becomes more zonally uniform with
increasing height.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F4"><caption><p>Horizontal maps of the standard deviation among the five reanalyses
at <bold>(a)</bold> 10 hPa, <bold>(b)</bold> 20 hPa, <bold>(c)</bold> 30 hPa,
<bold>(d)</bold> 50 hPa, and <bold>(e)</bold> 70 hPa. The color intervals are
0.4 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for <bold>(a)</bold>, 0.3 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for <bold>(b)</bold>, and
0.2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for <bold>(c–e)</bold>. Color ranges are different among these
heights.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f04.png"/>

        </fig>

</sec>
<sec id="Ch1.S3.SS2">
  <title>Dependence in the lower stratosphere</title>
      <p>Satellite radiance observations are distributed much more uniformly and
homogeneously than in situ balloon observations, so it might be expected that
satellite data assimilation would not have a large effect on the spatial
distribution of the SD among reanalyses. We may expect the SD geographical
distributions to be related primarily to the availability of station
observations. Figure 5a shows the locations of all radiosonde stations in the
IGRA (magenta dots indicate the locations of the stations shown later in
Fig. 6). We calculated the fraction of months during 1979–2001 with valid
monthly-mean wind data at each level at each IGRA station. Figure 5b–f
display the fractional data coverage at each radiosonde station with at least
some useful data from 10 to 70 hPa (the 100 % coverage of Singapore in
the FUB data is also indicated). The contours in these panels reproduce the
values of the SD shown in Fig. 4. Over 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S, there
are about 220 radiosonde stations (Fig. 5a), but only a fairly small fraction
have significant data coverage at the stratospheric levels of interest here,
particularly at 10 hPa.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F5"><caption><p><bold>(a)</bold> Locations of all IGRA stations in the tropical region
(magenta dots indicate the locations of the stations shown in Fig. 6).
<bold>(b–f)</bold> IGRA stations with data coverage of (purple) 1–20 %,
(blue) 20–40 %, (green) 40–60 %, (yellow) 60–80 %, and (red)
80–100 % at <bold>(b)</bold> 10, <bold>(c)</bold> 20, <bold>(d)</bold> 30,
<bold>(e)</bold> 50, and <bold>(f)</bold> 70 hPa during 1979–2001. The contours
show the standard deviation among reanalyses as shown in Fig. 4. The contour
intervals at each height are the same as in Fig. 4.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f05.png"/>

        </fig>

      <p>At 50–70 hPa, distributions of the SD appear to be negatively correlated
with the density of radiosonde observations. The SD shows the local minima in
the zonal direction near the locations with high observational density, such
as the vicinity of Sumatra, Borneo and the Malay Peninsula, and over much of
South America. Large SD values are found over an extensive region in the
eastern and central Pacific where few radiosonde observations are available.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F6" specific-use="star"><caption><p>Upper panels show observed zonal wind (black) at each station and
reanalysis zonal wind (color) at 70 hPa interpolated to each observational
point from 1979 to 2001. Lower panels depict standard deviation among the
five reanalyses at 70 hPa. Blue lines show the standard deviation during
times when monthly-mean radiosonde data are available and red lines show the
standard deviation during times when they are not. In the upper right corner
of each bottom panel, the total data coverage (%) at each station and the
average value of the standard deviation during periods with (blue number) and
without (red number) radiosonde data are shown.</p></caption>
          <?xmltex \igopts{width=341.433071pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f06.pdf"/>

        </fig>

      <p>Observational information is propagated with time and space through analysis
steps and forecast steps in data assimilation. In analysis steps, the
background error covariance matrix is used to determine the spatial
structure and the magnitude of analysis increment. Advanced data
assimilation techniques such as 4D-VAR (used in JRA-55 and ERA-I) are
expected to provide more efficient constraints even for remote points,
because of the use of flow-dependent analysis. The different data
assimilation methods used in each reanalysis would contribute to the SD even
if each reanalysis assimilated the same observational data. In addition, the
SD may also reflect differences among the climatologies of the GCMs used in
the analysis systems.</p>
      <p>Next, we consider the time variations of the spread among reanalyses at
individual places and its relation to the availability of local radiosonde
data. Each panel in Fig. 6 shows the time variations of monthly-mean 70 hPa
zonal wind measured at a particular radiosonde station together with the
values from each of the five reanalyses and the SD among the reanalyses. The
blue/red colors of the SD curves in each panel denote times with/without
radiosonde data at the station. In the upper right corner of each bottom
panel, the total data coverage (black number) and the average value of the SD
during periods with (blue number) and without (red number) radiosonde data at
each station are shown.</p>
      <p>At data-rich stations with coverage <inline-formula><mml:math display="inline"><mml:mo>≥</mml:mo></mml:math></inline-formula> 80 % such as
Thiruvananthapuram (Fig. 6c), Singapore (Fig. 6d), Chuuk (Fig. 6f), Majuro
(Fig. 6g), and Ascension (Fig. 6m), reanalyses zonal winds generally agree
well with the direct observations, and the SD among reanalyses is small. However, on Christmas Island (Fig. 6h), where no observational data are
available from these 23 years, the SD is large throughout the record. There
are no other stations with available data near Christmas Island (Fig. 5a),
which may also contribute to the extended period of large SD there.</p>
      <p>Menado (Fig. 6e) and San Cristobal (Fig. 6i) demonstrate that the reanalysis
SD at the locations of near-equatorial stations (within 1.5<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> of the
Equator; Fig. 5a) can be appreciably reduced during times when radiosonde
data from these stations are available. Bogota (Fig. 6j) and Manaus (Fig. 6k)
are good examples off the Equator also showing smaller SD during months with
balloon data than without these data.</p>
      <p>At Nairobi (Fig. 6a), an unrealistic spike in the westerlies is apparent in
the observed record for January 1981, but none of the reanalyses replicate
this spike. This might be an example of data quality control (e.g., gross
error check based on observation-minus-forecast departures statistics)
removing observational data during the assimilation processes. In the
evolution from late 1997 to early 1998, the direct observations at most
stations indicate double westerly peaks, but the second peak, in early 1998,
was not observed at Abidjan (Fig. 6n). ERA-I and ERA-40 data follow this
pattern and indicate a single peak but JRA-25, JRA-55, and MERRA display
double peaks. Consequently the SD has a local maximum during this period.
These results suggest that the error check procedures or assimilation schemes
employed in the different centers may respond differently to anomalous single
station observations.</p>
      <p>The correspondence between periods of availability of radiosonde data and SD
values among the reanalyses is sometimes unclear, such as at Seychelles
(Fig. 6b), for 1980–1982 at Belem (Fig. 6l), and for 1984–1998 at Abidjan (Fig. 6n).
However in general, the SD in Fig. 6 is smaller during the months when
radiosonde observations are available, and the SD tends to be larger when
they are not available. These results suggest that the observations at these
individual stations are having a significant influence constraining the
reanalyses of the zonal wind in the tropical lower stratosphere.</p>
      <p>To investigate what affects the longitudinal SD variation from the upper
troposphere to the lower stratosphere, Fig. 7 shows the longitude–height
cross section of zonal wind averaged over the five reanalyses and their SDs
(both averaged over the 1979–2001 period), as well as the temporal
correlation for the period 1979–2001 between the absolute value of the zonal
wind (i.e., the strength of the zonal wind, <inline-formula><mml:math display="inline"><mml:mrow><mml:mo>|</mml:mo><mml:mo>[</mml:mo><mml:mi>u</mml:mi><mml:mo>]</mml:mo><mml:mo>|</mml:mo></mml:mrow></mml:math></inline-formula>) and the SD. All panels
show the 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S average. Figure 7a displays the
familiar Walker circulation signal in prevailing zonal winds in the upper
troposphere. Figure 7a suggests the eddy components of the prevailing zonal
wind at 70 hPa (and to some extent even at 50 hPa) can be regarded as part
of the Walker circulation. Hamilton et al. (2004) showed in their general
circulation model experiments that, below about 60 hPa, the deviation of the
zonal wind from the zonal mean appears to be dominated by an extension of the
tropospheric Walker circulation.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F7"><caption><p>Longitude–height cross section of <bold>(a)</bold> averaged zonal wind
in the five reanalyses,  <bold>(b)</bold> the standard deviation among
reanalysis, and <bold>(c)</bold> the correlation between absolute value of zonal
wind and the standard deviation in 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S averaged
from 1979 to 2001. The color intervals are <bold>(a)</bold> 2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>,
<bold>(b)</bold> 0.2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for values less than 2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> and
0.5 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for values more than 2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>, and <bold>(c)</bold> 0.1, with
statistical significance of <inline-formula><mml:math display="inline"><mml:mo>≥</mml:mo></mml:math></inline-formula> 95 % colored.</p></caption>
          <?xmltex \igopts{width=184.942913pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f07.png"/>

        </fig>

      <p>A zonally elongated region of large SD exists around 100–150 hPa, the upper
part of the Walker circulation (Fig. 7b). This upper-tropospheric SD is
largest in the central Pacific, where few observational stations are located
(Fig. 5a). It is interesting that the region of large SD around the central
Pacific in the upper troposphere connects with that seen in the lower
stratosphere.</p>
      <p>The correlation between the absolute value of zonal wind and the SD among
reanalyses shows relatively high positive correlation in the upper part of
the Walker circulation in the eastern hemisphere, while the correlation is
relatively low in the central Pacific (Fig. 7c). In the mean state, the eddy
component in the tropical lower stratosphere might be associated with an
extension of the Walker circulation, but 70 hPa SD is
relatively small, and it does not seem correlated with the strength of the upper-tropospheric Walker circulation. The large SD from the upper troposphere to
the stratosphere in the central Pacific could be simply related to the fewer
in situ observations available there. In the middle stratosphere, the
correlation is negative, indicating that the reanalyses disagree the most
when the magnitude of the zonal wind is weakest. This corresponds to large SD
during the phase transition of the QBO, as shown in the next section.</p>
</sec>
<sec id="Ch1.S3.SS3">
  <title>Differences in the middle stratosphere</title>
      <p>In this subsection we discuss the SD at 10–30 hPa where generally fewer
radiosonde observations are available. At 20–30 hPa, the SD becomes more
zonally uniform compared with results at 50–70 hPa (Fig. 4). However,
relatively higher-density observations around the Indonesian maritime
continent, as well as over the North and South American and African
continents, do act to reduce the SD locally (Fig. 5c and d). At 10 hPa,
Singapore is the only station quite near the Equator with high-density
coverage, and several stations around 7–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N in the western
Pacific have data coverage of 40–80 % (Fig. 5b).</p>
      <p>Because the SD among reanalyses involves more zonally uniform structures in
the middle stratosphere (Fig. 4), the zonally averaged SD is investigated
first. Figure 8 shows zonal mean equatorial zonal wind in each reanalysis
plotted together with the zonally averaged SD among reanalysis at 10, 20, and
30 hPa. Large differences are seen when the QBO phase changes from easterly
to westerly. In addition, the SD is generally larger during the westerly
phase than during the easterly phase of the QBO. The dependence of SD on the
QBO phase is most pronounced at 10 hPa.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F8"><caption><p>Zonal mean equatorial zonal wind in each reanalysis (colors) and
zonally averaged equatorial standard deviation among reanalyses (black) at
<bold>(a)</bold> 10, <bold>(b)</bold> 20, and <bold>(c)</bold> 30 hPa.</p></caption>
          <?xmltex \igopts{width=184.942913pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f08.png"/>

        </fig>

      <p>To further investigate this dependence, a zonal wind composite based on the
phase of the QBO was computed for the FUB observations and for each
reanalysis over 1979–2001. For the easterly-to-westerly composite, the month
“0” is defined as the final month with a mean easterly value in the 10 hPa
FUB data and month “<inline-formula><mml:math display="inline"><mml:mo>+</mml:mo></mml:math></inline-formula>1” is the first month with a mean westerly value.
Composite values are then computed for <inline-formula><mml:math display="inline"><mml:mo>±</mml:mo></mml:math></inline-formula>6 months around these zero months
(month 0 in August 1979, January 1982, March 1984, etc.) for the
easterly-to-westerly phases; see Fig. 2a. A similar westerly-to-easterly
composite was also computed (with month 0 taken at June 1980, January 1983,
June 1985, etc.). We chose 10 hPa as presenting the biggest challenge for
the reanalyses as the numbers of radiosonde observations are fewest and the
SD among reanalyses is largest.</p>
      <p>The QBO composites for the zonal wind at 10 hPa are shown in Fig. 9a–d.
Results for the zonal wind in Singapore are shown in the left panels (a and
c) while results for the zonal mean equatorial zonal wind are shown in the
right panels (b and d). Through the easterly-to-westerly transition in
Singapore (Fig. 9a), the composite zonal winds in most of the reanalyses have
an easterly bias relative to the FUB observations particularly during months
<inline-formula><mml:math display="inline"><mml:mo>-</mml:mo></mml:math></inline-formula>1 to <inline-formula><mml:math display="inline"><mml:mo>+</mml:mo></mml:math></inline-formula>3. This results in a delay in the time of the zero-crossing in the
reanalyses relative to the FUB observations. The reanalyses differ
considerably in terms of this bias, however. The delay is
<inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 1.5–2 month in JRA-25 and JRA-55 and <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 0.5 month in ERA-40 and
ERA-I. The MERRA zonal winds show good agreement with FUB data during the
easterly-to-westerly transition. The zonal wind values in the reanalyses show
better agreement with the FUB observations and closer agreement among
themselves during the westerly-to-easterly phase transition (Fig. 9c).</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F9"><caption><p>Top: composite of the QBO in the zonal wind at 10 hPa during 1979
to 2001 where month 0 corresponds to <bold>(a, b)</bold> the easterly-to-westerly
transition of the FUB zonal wind at 10 hPa and to <bold>(c, d)</bold> the
westerly-to-easterly transition. Color lines show each reanalysis zonal wind,
and black dashed lines depict FUB zonal wind in Singapore. Results
for <bold>(a, c)</bold> the zonal wind in Singapore and for <bold>(b, d)</bold> the
zonal mean equatorial zonal wind. Bottom: composite of the zonal mean
equatorial standard deviation due to <bold>(e)</bold> zonal mean and
<bold>(f)</bold> eddy components during (red) easterly-to-westerly and (blue)
westerly-to-easterly transitions at 10 hPa. Note the ordinate axes are
different between <bold>(e, f)</bold>.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f09.pdf"/>

        </fig>

      <p>In composites of the zonal mean zonal wind (Fig. 9b, d), during the 6
months following the transition date the maximum westerly in each reanalysis
is slightly smaller than in the zonal wind over Singapore
(<inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 3–4 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for ERA-40 and ERA-I, <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 2–3 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>
for JRA-25 and JRA-55, nearly the same for MERRA), while the maximum easterly
is nearly same. In both phase transitions, the spread of reanalysis zonal
mean zonal winds is larger than that over Singapore. It should be noted that
the timing of both phase transitions in the zonal mean is generally the same
as those seen just using results in Singapore (although there is a slight
delay of <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 0.5 months in the easterly-to-westerly transition of the
zonal mean in the JRA-25 and JRA-55).</p>
      <p><?xmltex \hack{\newpage}?>Figure 9e and f show a QBO composite of the zonal mean equatorial SD among
five reanalyses due to zonal mean and eddy components (square roots of the
first and second terms on the right in Eq. (2), respectively) at 10 hPa. It
is clear that the SD due to zonal mean components is large during the phase
transition from easterly to westerly, while the SD does not change much
during the 12 composite months during the westerly-to-easterly (Fig. 9e). The
SD due to eddy components has larger values during the westerly than the
easterly phase (Fig. 9f).</p>
      <p>It is well known that the transition of the QBO from easterly-to-westerly is
considerably faster than that from the westerly-to-easterly (e.g., Baldwin et
al., 2001). In other words, the zonal wind tendency <inline-formula><mml:math display="inline"><mml:mrow><mml:mfenced close="|" open="|"><mml:mo>∂</mml:mo><mml:mi>u</mml:mi><mml:mo>/</mml:mo><mml:mo>∂</mml:mo><mml:mi>t</mml:mi></mml:mfenced></mml:mrow></mml:math></inline-formula> is larger during the easterly-to-westerly transition.
As the QBO could not be simulated without data assimilation from each
reanalysis model (except MERRA-2), one possible reason for the large SD in
the easterly-to-westerly is weak forcing by resolved waves in the reanalysis
model, leading to a slow change of the zonal wind and a resulting delay of the
phase transition. The results of Fig. 9 suggest that the total observational
constraints are insufficient to completely compensate for model bias,
resulting in slow change of the zonal wind and delay of the phase transition.
The another possibility is that during periods of weak zonal wind there will
be delays in the zonal advective propagation of information introduced into
the analysis system from observations at individual stations (notably
Singapore). We are not able to definitely assess these two possible
mechanisms, but the fact that the lag in the reanalysis winds is more
pronounced in the more rapid easterly-to-westerly wind transition may favor
the importance of the model bias over the slow advective propagation
mechanism.</p>
      <p>Each reanalysis model may not have sufficiently fine vertical resolution to
represent completely the interaction between vertically propagating waves and
the mean flow, which is thought to be crucial to the QBO dynamics (e.g.,
Baldwin et al., 2001), because the vertical wavelengths of waves become
smaller as they approach critical levels. Podglajen et al. (2014) show
vertical profiles of meridional wind by radiosonde observation in Singapore
and in ECMWF operational analysis and indicate that ECMWF does not adequately
represent wave disturbances due to insufficient vertical resolution (Fig. 9
of their study). MERRA has the finest vertical resolution among five
reanalyses in the 20–30 km layer, which may allow a better representation
of wave forcing in the strong vertical shears of the QBO. Kim and Chun (2015)
investigate the momentum forcing of the QBO by resolved waves in reanalysis
models and show that MERRA has the larger net-resolved wave forcing than
ERA-I, MERRA, and JRA-55 (Fig. 2 in their study). They also show that mean
residual vertical velocity in the tropics, which generally acts in the
opposite sense to wave forcing in driving the zonal wind acceleration (e.g.,
Kawatani et al., 2011), is much smaller in MERRA than in other reanalyses
(Fig. 5a in their study). These are possible reasons why the MERRA shows
faster transitions both in easterly-to-westerly and westerly-to-easterly
transitions compared with those in the other reanalyses.</p>
      <p>Next we will discuss the representation of eddy components of the wind in the
reanalysis data sets. Figure 10a–e show <inline-formula><mml:math display="inline"><mml:mrow><mml:msup><mml:mi>u</mml:mi><mml:mo>′</mml:mo></mml:msup></mml:mrow></mml:math></inline-formula>, the deviation of the
monthly-mean zonal wind values from the zonal mean, at 10 hPa for each
reanalysis in January 1999, during the westerly phase of the QBO. In all the
reanalyses the eddy component near the Equator appears to be dominated by
zonal wavenumber 1 and 2 quasi-stationary planetary waves propagating from
midlatitudes, but the patterns differ somewhat among the reanalyses. For
example, ERA-40 shows a large, local, positive anomaly to the east of the
Indonesian maritime continent, while JRA-55 does not. Each reanalysis shows a
different shape for this positive anomaly. In addition, the negative
anomalies around the Bay of Guinea and central Pacific are also different.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F10"><caption><p>Deviation of the zonal wind values from the zonal mean (<inline-formula><mml:math display="inline"><mml:mrow><mml:msup><mml:mi>u</mml:mi><mml:mo>′</mml:mo></mml:msup><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> at
10 hPa of <bold>(a–e)</bold> five reanalyses in January 1999 during the westerly
phase and <bold>(f)</bold> of the ERA-I in January 1996 during the easterly phase of
the QBO. The color interval is <inline-formula><mml:math display="inline"><mml:mo>±</mml:mo></mml:math></inline-formula>3, 6, 9, 12, 15, 20, or 25 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>.
<bold>(g)</bold> Longitudinal variations of the zonal wind of each reanalysis at
10 hPa over the Equator in January 1999 and January 1996.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f10.png"/>

        </fig>

      <p>Figure 10f shows one example of the eddy components of the zonal wind in
January 1996, during the easterly phase of the QBO in ERA-I. The eddy
components are very small over the Equator, which is expected as stationary
planetary waves cannot propagate into the equatorial mean easterlies.
Figure 10g depicts the longitudinal variations of each reanalysis zonal wind
at 10 hPa over the Equator in January 1996 and 1999. It is clear that the
spread of the zonal wind among reanalyses is larger in the westerly phase,
and longitudinal variation of the zonal wind is smaller in the easterly
phase. We have checked other periods and found the monthly-mean eddy
structures in the middle stratosphere associated with quasi-stationary
planetary waves are qualitatively similar among years (not shown). The large
contribution to SD from eddy components during the westerly phase of the QBO
(Fig. 9f) mainly results from different representation of midlatitude
planetary waves propagating into the Equator among reanalyses.</p>
</sec>
<sec id="Ch1.S3.SS4">
  <title>Difference depending on the QBO phase</title>
      <p>In this subsection, we investigate the dependence of the SD among the
reanalyses on the QBO phase. We have computed the SD separately for the
westerly and easterly phases of the QBO defined by the sign of the five
reanalyses averaged zonal mean zonal wind at the Equator (i.e., defined as
<inline-formula><mml:math display="inline"><mml:mrow><mml:mo>[</mml:mo><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:msub><mml:mo>]</mml:mo><mml:mtext>eq</mml:mtext></mml:msub><mml:mo>&gt;</mml:mo><mml:mn mathvariant="normal">0</mml:mn></mml:mrow></mml:math></inline-formula> for the westerly and <inline-formula><mml:math display="inline"><mml:mrow><mml:mo>[</mml:mo><mml:mover accent="true"><mml:mi>u</mml:mi><mml:mo mathvariant="normal">‾</mml:mo></mml:mover><mml:msub><mml:mo>]</mml:mo><mml:mtext>eq</mml:mtext></mml:msub><mml:mo>&lt;</mml:mo><mml:mn mathvariant="normal">0</mml:mn></mml:mrow></mml:math></inline-formula> for the easterly phase; subscript <italic>eq</italic> indicates the Equator).
Figure 11a and b are maps of the SD among reanalyses at 10 hPa during the
westerly and easterly phases of the QBO for 1979–2001. Figure 11c–g plot
the zonal variation of the SD averaged over 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N at
each height in the easterly and westerly QBO phases. At 10 hPa, large SD
values in the westerly phase are seen in the Pacific and from the Atlantic to
Africa, where representations of quasi-stationary planetary waves differ
significantly among the reanalyses (see Fig. 10). However, in the easterly
phase, the SD is more zonally uniform. It is interesting that, both in the
westerly and easterly phases, the SD declines significantly around the
longitudes near Singapore. At 10 hPa the Singapore station provides the only
continuous record of in situ data near the Equator (Fig. 5b), and the results
in Fig. 11c suggest that observations from this one station act as a strong
constraint on the reanalyses.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F11"><caption><p>Horizontal distributions of the standard deviation among reanalyses
at 10 hPa during the <bold>(a)</bold> westerly and <bold>(b)</bold> easterly phase of
the QBO averaged in the period 1979–2001. The color interval is
0.3 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>. Longitudinal variations of the standard deviation at
10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N in (red) westerly and (blue) easterly phase at
<bold>(c)</bold> 10, <bold>(d)</bold> 20, <bold>(e)</bold> 30, <bold>(f)</bold> 50, and
<bold>(g)</bold> 70 hPa.</p></caption>
          <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f11.pdf"/>

        </fig>

      <p>The overall larger SD in the westerly QBO phases, as compared to the easterly
QBO phases, especially over the Pacific, is still clear at 20 hPa
(Fig. 11d). However, this feature is not apparent at 30 hPa (Fig. 11e),
suggesting that the effects of quasi-stationary planetary waves become much
smaller at 30 hPa. At 50 hPa, the SD in the westerly and easterly phases is
nearly identical (Fig. 11f). At 70 hPa the longitudinal variation of the SD
in the westerly phase is similar to that in the easterly phase (Fig. 11g).
This indicates that the QBO phase is not very important for SD zonal
variations in the lower stratosphere.</p>
</sec>
</sec>
<sec id="Ch1.S4">
  <?xmltex \opttitle{Evolution of the differences among reanalyses\hack{\break} with time}?><title>Evolution of the differences among reanalyses<?xmltex \hack{\break}?> with time</title>
      <p>In this section, we discuss how the SD among reanalyses changes with time as
the in situ and remote data sources available for the reanalyses evolve. We
will show the results from the four reanalyses data sets (ERA-I, JRA-25,
JRA-55, and MERRA) that exist over the entire 34 year period 1979 to 2012.
Figure 12a–c are maps of the SD averaged over 50–70 hPa for three 11-year
intervals: 1979–1989, 1990–2000, and 2001–2011. The SD among reanalyses
reduces significantly with time, but it is interesting that geographical
distributions of the SD remain quite similar over these three periods.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F12"><caption><p>Standard deviation among four reanalyses (ERA-I, JRA-25, JRA-55, and
MERRA) at 50–70 hPa for each 11-year mean: <bold>(a)</bold> 1979–1989,
<bold>(b)</bold> 1990–2000, and <bold>(c)</bold> 2001–2011. The color intervals are
0.2 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula>.</p></caption>
        <?xmltex \igopts{width=213.395669pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f12.png"/>

      </fig>

      <?xmltex \floatpos{t}?><fig id="Ch1.F13" specific-use="star"><caption><p>Time variations of zonal mean standard deviation of (black) all,
(red) zonal mean, and (blue) eddy components among (left and middle) four
reanalyses (ERA-I, JRA-25, JRA-55, and MERRA), and among (right) three
reanalyses (ERA-I, JRA-55, and MERRA) from 1979 to 2012 at 10, 20, 30, 50,
and 70 hPa in 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N.</p></caption>
        <?xmltex \igopts{width=369.885827pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f13.png"/>

      </fig>

      <p>The left panels in Fig. 13 show time variations of zonal mean SD among the
four reanalyses at each height averaged over 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N.
At 50 and 70 hPa (Fig. 13j, m), the overall level of SD reduces gradually in
the 1980s and 1990s, and a clear drop is found around 1998 at 50 hPa. At
10–30 hPa (Fig. 13a, d, g), the SD also reduces with time, but a sudden
drop is not apparent.</p>
      <p>The middle panels in Fig. 13 plot the same quantity as the left panels, but
for the SD due to zonal mean and eddy components separately. At 70 hPa
(Fig. 13n), eddy components are comparable to zonal mean components for all
periods, and both components reduce gradually with time until <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 2000.
The typical level of SD in both components remains fairly constant after
2000. At 50 hPa, the drop of SD in the zonal mean components around 1998 is
pronounced; the eddy components also appear to become somewhat smaller after
<inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 1998.</p>
      <p>At 10–30 hPa, zonal mean contribution to SD always dominates, but gradually
reduces with time (Fig. 13b, e, h). However, throughout the record this
quantity spikes generally near a particular phase of the QBO (i.e., during the
easterly-to-westerly phase transition). The eddy contribution to SD also
reduces slightly with time but not dramatically.</p>
      <p><?xmltex \hack{\newpage}?>One possible reason for reduction of the zonal wind SD over time would be the
upgrading of satellite radiance observations. From 1979 to 2006, the TOVS
(TIROS (Television InfRrared Operational Satellite) Operational Vertical
Sounder) Stratospheric Sounding Unit (SSU) and Microwave Sounding Unit
(MSU) were available. After May 1998, data from the Advanced Microwave
Sounding Unit (AMSU) became available. The advent of these new satellite
measurements is a possible reason for the decline in the SD around 1998.</p>
      <p>Figure 14a shows the time–height section of zonal mean SD of temperature
among the four reanalyses. A large SD is found around 30–50 hPa, and this
SD reduces significantly after <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 1998. The JRA team has reported on
stratospheric temperature issues with JRA-25 and discussed the cause of the
biases (Onogi et al., 2007). Fujiwara et al. (2015) explained that the
radiative scheme used in the JRA-25 forecast model has a known cold bias in
the stratosphere, and the TOVS SSU/MSU measurements do not have a sufficient
number of channels to correct the model's cold bias; after introducing the
ATOVS AMSU-A measurements in 1998, such a cold bias disappeared in the JRA-25
data product. However, with JRA-55 using a new radiation scheme in
the forecast model, the stratospheric temperature during the TOVS period has
been much improved.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F14"><caption><p>Time–height cross sections of zonal mean standard deviation of
temperature among <bold>(a)</bold> four reanalyses (ERA-I, JRA-25, JRA-55, and
MERRA) and <bold>(b)</bold> three reanalysis (ERA-I, JRA-55, and MERRA) at
10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N from 1979 to 2012. The color intervals are
0.1 K for values less than 0.8 and 0.2 K for values more than 0.8 K.</p></caption>
        <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f14.pdf"/>

      </fig>

      <p>In light of the possible issues with the JRA-25 temperatures we show in
Fig. 14b the temperature SD excluding JRA-25 (i.e., including only ERA-I,
JRA-55, and MERRA). When JRA-25 is excluded, the very large SD values over
20–50 hPa before 1998 seen in Fig. 14a are greatly diminished. Even so the
temperature SD among the three reanalyses (Fig. 14b) does reduce with time,
with an especially large reduction over the full record at 100 and 10 hPa.
The right panels of Fig. 13 show the same quantity as the left panels but
for zonal wind SD among just the ERA-I, JRA-55, and MERRA reanalyses. The SD
among the three reanalyses is generally smaller than that among four
reanalyses and the dramatic change in zonal wind SD at 50 hPa in 1998 become
much weaker with the removal of JRA-25 (Fig. 13l), but the time evolution
looks very similar.</p>
      <p>Figure 15 shows the evolution of the number of available monthly-mean
radiosonde observations over 10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N and 10–70 hPa.
The number of radiosonde observations available generally increased with time
at all levels. For example at 70 hPa (50 hPa), the mean number of
station months of wind observations was <inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 33.7 (30.1) in 1979–1989,
<inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 37.6 (35.1) in 1990–2000, and 44.5 (43.0) in 2001–2011 (i.e.,
periods corresponding to those in Fig. 12). So, relative to 1979–1989, the
number of radiosonde observations at 70 hPa (50 hPa) increased by
11.6 % (16.6 %) in 1990–2000 and 32.0 % (42.9 %) in
2001–2011.</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F15"><caption><p>Time variation in number of radiosonde observations at
10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N at <bold>(a)</bold> 10, <bold>(b)</bold> 20,
<bold>(c)</bold> 30, <bold>(d)</bold> 50, and <bold>(e)</bold> 70 hPa from 1979 to
2012.</p></caption>
        <?xmltex \igopts{width=156.490157pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f15.png"/>

      </fig>

      <p>In contrast, the zonal wind SD among the three reanalyses at
10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S–10<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N (right panels in Fig. 13) and 70 hPa
(50 hPa) reduces 20.8 % (20.4 %) in 1990–2000 and 33.7 %
(36.3 %) in 2001–2011, compared with those in 1979–1989. The abrupt
drop around 1998 is not clearly reflected in the number of radiosonde
observations at 50 hPa, implying that the AMSU satellite observations could
have significant impact on reanalysis of the zonal wind in the tropical
stratosphere (see Fig. 13j, l).</p>
      <p>At 10–30 hPa, the number of radiosonde observations increased significantly
from 1979 to 2012, which should contribute to a reduction of the zonal wind
SD with time at these altitudes, in addition to having more satellite data
available. The same 10 hPa QBO composites as shown in Fig. 9a–d but during
1998–2012 are illustrated in Fig. 16. Here the zonal wind in ERA-40 is
excluded (ERA-40 data exist only up to August 2002) but MERRA-2 is included
(the unrealistic features noted earlier in MERRA-2 are not present after
1998). The general features are same as those in 1979–2001 such as an
apparent delay in easterly-to-westerly transitions in JRA-25 and JRA-55
compared with the FUB observation and closer agreement among reanalyses in
westerly-to-easterly transitions. It should be noted here that in these
reanalyses the 10hPa QBO in 1998–2012 are closer to the FUB observations
both during easterly-to-westerly and westerly-to-easterly transitions than
during 1979–2001 (Fig. 9). In addition, the spread of reanalysis zonal winds
become much smaller in the later period, especially for the zonal mean wind.
Figure 16 also shows that the MERRA-2 reanalyses display an
easterly-to-westerly phase transition in Singapore that is even more rapid
than in the direct balloon observations. These results may indicate that the
gravity wave sources in MERRA-2 are now excessive (Coy et al., 2016).</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F16"><caption><p>Same as Fig. 9a–d but during 1998 to 2012, excluding ERA-40 and
including MERRA-2.</p></caption>
        <?xmltex \igopts{width=236.157874pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f16.png"/>

      </fig>

      <p>Satellite radiance data will presumably affect the assimilated temperatures
in the stratosphere but will also have some influence on the wind (cf. Iida
et al., 2014), mainly through the use of multivariate background error
covariance matrix at analysis step even though the coupling of temperature
and wind may be somewhat weak as the Coriolis parameter is small near the
Equator. Thermal wind shears will be very sensitive to small errors in
observed temperatures. After 2000 the amount of available satellite data
increased greatly (e.g., Kobayashi et al., 2015), and the contributions of
satellite radiance data to better representation of the tropical winds
presumably also increased. However, it is hard to quantify the relative roles
of global satellites and in situ radiosonde observations in reducing the
zonal wind SD among reanalyses.</p>
      <p>Finally, Fig. 17a–d are the horizontal maps at 50–70 hPa SD among the
three most modern reanalyses (ERA-I, JRA-55, and MERRA) illustrated for
2001–2011, when the SD is smallest (Fig. 12). Here, we also calculated SD
between ERA-I and JRA-55, ERA-I and MERRA, and JRA-55 and MERRA separately.
The SD becomes larger when JRA-55 data are included, and the SD between ERA-I
and MERRA is smallest. However, in any case, the geographical patterns of the
SD are always similar, independent of the reanalysis or period selected (see
Figs. 4, 12, and 17). These results indicate the importance of zonal wind
observations by local radiosondes for better representation of the tropical
zonal wind in the lower stratosphere in reanalysis data sets. Just for
reference, the same figure, but between MERRA and MERRA-2, is shown in
Fig. 17e. The geographical pattern is also similar, even though the MERRA-2
assimilation was notable in using a dynamical model that simulated a
spontaneous QBO.</p>
</sec>
<sec id="Ch1.S5" sec-type="conclusions">
  <title>Summary and concluding remarks</title>
      <p>This paper reports on a project to compare the representation of the
monthly-mean zonal wind in the equatorial stratosphere among major global
atmospheric reanalyses data sets. We mainly confined our analysis to
20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> N–20<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mo>∘</mml:mo></mml:msup></mml:math></inline-formula> S and 10–70 hPa and to data starting in 1979
(i.e., after the availability of operational satellite temperature soundings).
We compared the zonal wind in the reanalyses with the FUB data set for equatorial monthly-mean winds, which, for this
period, were based on balloon observations in Singapore. We also compared
reanalysis values with balloon observations at many other low-latitude
stations included in the IGRA.</p>
      <p>All nine reanalyses data sets we examined display monthly-mean zonal wind
values in Singapore that are reasonably close to the directly observed values
at all levels between 70 and 10 hPa (Fig. 2). However, the NCEP-1 and NCEP-2
analyses have significantly larger differences with the Singapore
observations than  the other reanalyses. Notably, NCEP-1 and NCEP-2 display
considerably weaker (as much as 10 m s<inline-formula><mml:math display="inline"><mml:mrow><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> easterly QBO extremes than
is apparent in the balloon observations, particularly during the 1980s.
MERRA-2 also stands out as an outlier in displaying large differences from
FUB at 10 hPa in the 1980s and in November–December 1993. The NCEP-CFSR
uses the ERA-40 winds in the tropical stratosphere from July 1981 to
December 1998, and the differences between NCEP-CFSR and ERA-40 during this
period are nearly 0. Further analysis and discussion here was largely
restricted to the other five reanalyses (ERA-40, ERA-I, JRA-25, JRA-55, and
MERRA).</p>
      <p>We characterized the degree of disagreement among the reanalyses by the
SD of the monthly-mean zonal wind values in the five
data sets. This measure is a function of height, latitude, longitude, and
time. At each height the SD displays a prominent equatorial maximum,
indicating the particularly challenging nature of the reanalysis problem in
the low-latitude stratosphere where in situ observations are relatively
sparse and the Coriolis parameter is small. The SD in the tropical
stratosphere also depends significantly on both longitude and height
(Fig. 4). At 50–70 hPa, a large SD is seen for the Indian Ocean, to the
east of the maritime continent, the central Pacific, and the Atlantic,
showing clear zonally non-uniform structures. At 10–30 hPa, however, the SD becomes more zonally uniform.</p>
      <p>At 50–70 hPa, the distributions of the SD are closely related with those of
in situ radiosonde observations (Fig. 5e, f). The region of largest SD
extends over wide zonal range in the central Pacific, where few observations
are available. In the vicinity of individual radiosonde stations the SD at
70 hPa generally reduces in periods when observations are available relative
to periods with no observations (Fig. 6).</p>
      <p>At each level there is a tendency for agreement to be best near the longitude
of Singapore (a result that holds in all QBO phases), suggesting that the high-quality Singapore balloon observations act as a strong constraint on all the
reanalyses. At 10 and 20 hPa, the zonal mean SD has a clear dependence on
the phase of the QBO (Fig. 8). Specifically the SD generally jumps to
5–10 m s<inline-formula><mml:math display="inline"><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup></mml:math></inline-formula> for a few months around the transition from easterly to
westerly. At 30 hPa, the difference is smaller (almost always less than
5 m s<inline-formula><mml:math display="inline"><mml:mrow><mml:msup><mml:mi/><mml:mrow><mml:mo>-</mml:mo><mml:mn mathvariant="normal">1</mml:mn></mml:mrow></mml:msup><mml:mo>)</mml:mo></mml:mrow></mml:math></inline-formula> but it still shows the same QBO dependence. The main
discrepancy with FUB observations at these heights is a tendency for the
zonal wind accelerations to be delayed in the all reanalyses, except MERRA,
around the easterly-to-westerly transition (Fig. 9).</p>
      <p>At 10 hPa, all the reanalyses show that the eddy component (i.e., the
deviation of the zonal wind values from the zonal mean) in the monthly-mean
zonal wind near the Equator is dominated by zonal wavenumber 1 and 2
quasi-stationary planetary waves propagating from midlatitudes (Fig. 10).
Consistent with this, the eddy component of the zonal wind near the Equator
is small in months with zonal mean easterlies and generally much larger in
months with zonal mean westerlies. While the different reanalyses agree in
broad terms about the structure of the eddy components in the middle
stratosphere, there generally remains considerable disagreement among the
reanalyses in the details of the quasi-stationary waves near the Equator. At
50–70 hPa the SD distributions are not strongly affected by the phase of
the QBO. The eddy features at these heights, rather than being planetary
waves from midlatitudes, might be related to stratospheric penetration of
the Walker circulation (Fig. 7).</p>

      <?xmltex \floatpos{t}?><fig id="Ch1.F17"><caption><p>Same as Fig. 12c but for the standard deviation among
<bold>(a)</bold> three reanalysis (ERA-I, JRA-55, and MERRA),
<bold>(b)</bold> between ERA-I and JRA-55, <bold>(c)</bold> between ERA-I and MERRA,
<bold>(d)</bold> between JRA-55 and MERRA, and <bold>(e)</bold> between MERRA and
MERRA-2.</p></caption>
        <?xmltex \igopts{width=184.942913pt}?><graphic xlink:href="https://acp.copernicus.org/articles/16/6681/2016/acp-16-6681-2016-f17.png"/>

      </fig>

      <p>Our measures reflecting the reliability of the reanalyses show systematic
improvement over the 1979–2012 period considered. Specifically, the
agreement of the reanalysis results with the Singapore data improves
significantly with time, with average deviations from the observation reduced
by almost half from that in the early 1980s to the 2010s (Fig. 2f). At
50–70 hPa, the magnitude of the SD among the reanalyses declines
significantly over the period of record (Fig. 13), although the geographical
pattern of differences (as measured by the SD of the five reanalysis values)
is nearly constant (Fig. 12).</p>
      <p>For any individual reanalysis, the same atmospheric dynamical model and the
same assimilation procedures are used throughout the whole period; thus any
improvement seen in the quality with time must be related to changes in the
observational data used. The number of available in situ radiosonde
observations increase with time at all levels (Fig. 15). In addition,
satellite data streams vary throughout this period. The end of 1978 saw the
introduction of the SSU, which was flown on NOAA
operational satellites for the next 2 decades (NOAA8 though NOAA14). We
expect an important upgrade may have occurred in May 1998 when the
temperature sounders on NOAA operational satellites were upgraded from the
SSU to the AMSU. The satellite streams
continued to show changes with the introduction of more AMSU instruments
flying on other satellites. Additionally, new sources of radiance data also
appeared, such as the Atmospheric Infrared Sounder (AIRS) flown on the NASA
Aqua satellite in 2002.</p>
      <p><?xmltex \hack{\newpage}?>Doing assimilation experiments removing certain data sets from the analysis
and exploring the impact of particular data sets is one possible way to
investigate the data included. The effects of satellite data sets might be
investigated by comparison between JRA-55 and JRA-55c (which excludes all
satellite data as input). Kobayashi et al. (2014) compare the JRA-55 and
JRA-55C reanalyses and indicate that the tropical zonal wind difference
between the JRA-55 and JRA-55C is large in the upper stratosphere compared to
that in the lower stratosphere, and the amplitude in JRA-55C is smaller than
that of the JRA-55. More detail comparisons along these lines are reserved
for future work.</p>
      <p>The magnitude of the SD in the horizontal winds throughout the tropics would
be worth highlighting for trajectory studies given that there is some
inclination to assume that the key the inter-reanalysis differences for
chemical transport are in the vertical velocities. Abalos et al. (2015) have
recently compared the stratospheric Brewer–Dobson circulation among MERRA,
ERA-I, and JRA55, while Miyazaki et al. (2016) did a somewhat similar study
with six different reanalyses. Abalos et al. (2015) find quite large
(<inline-formula><mml:math display="inline"><mml:mo>∼</mml:mo></mml:math></inline-formula> 40 %) differences among the reanalyses in the overall strength
of the BD circulation and a large spread among the estimates
of the fraction of variance in mean vertical motion explained by the QBO. Coy
et al. (2016) found that
MERRA-2, which internally generates the QBO, has reduced the zonal wind
analysis increments compared to MERRA, so that the QBO mean meridional
circulation can be expected to be more physically forced and more physically
consistent.</p>
      <p>The difference among reanalyses using twice daily data should be much larger
than our SD based on monthly-mean data (cf. Baker et al., 2014). Podglajen et
al. (2014) compare reanalyses winds with independent in situ observations
performed along long-duration balloon flights. They report that ERA-I and
MERRA represent similar disturbances associated with equatorial waves, but
reanalyses depart from the balloon observations. As the present study focus
on monthly-mean field, our analyses ignore variability with shorter timescales.</p>
      <p>Our study confirms that even with relatively few balloon stations, the high-accuracy and high-resolution wind measurements by in situ radiosondes have
provided important constraints to reanalyses of circulation in the tropical
stratosphere. Our analysis also shows that changes in both satellite and
balloon data availability over the years have significantly affected the
reanalysis of zonal wind data for the equatorial stratosphere. This
conclusion has a negative aspect as it suggests that using reanalysis
products (even restricting the data to after 1978) does not enable a
researcher to avoid all significant artificial trends. On the positive side,
it seems that the changes in available data have led to continually improving
reanalysis representation of the tropical stratosphere.</p>
<sec id="Ch1.S5.SSx1" specific-use="unnumbered">
  <title>Data availability</title>
      <p>The monthly mean zonal wind values of operational balloon-borne radiosonde
observations compiled by the Free University of Berlin are available at
<uri>http://www.geo.fu-berlin.de/en/met/ag/strat/produkte/qbo/</uri> (Freie
Universität Berlin, 2016).</p>
      <p>The monthly-mean values of the zonal wind at many radiosonde stations
provided in the Integrated Global Radiosonde Archive are available at
<uri>https://www.ncdc.noaa.gov/data-access/weather-balloon/integrated-global-radiosonde-archive</uri>
(NOAA, 2016).</p>
</sec>
</sec>

      
      </body>
    <back><ack><title>Acknowledgements</title><p>We thank M. A. Geller, A. Hertzog, and three anonymous reviewers for
constructive comments on the original manuscript. We also express our
gratitude to S-RIP members for providing useful information about the
reanalyses data. This research was supported by Grant-in-Aid for Scientific
Research B (26287117) and Joint international Research (15KK0178) from the
Japan Society for the Promotion of Science, and by the Environment Research
and Technology Development Fund (2-1503) of the Ministry of the Environment,
Japan. This research was also supported by the Japan Agency for Marine-Earth
Science and Technology (JAMSTEC) through its sponsorship of research at the
International Pacific Research Center and by NOAA through grant
no. NA11NMF4320128. The GFD-DENNOU Library and GrADS were used to draw the
figures.<?xmltex \hack{\newline}?><?xmltex \hack{\newline}?> Edited by: P. Haynes</p></ack><ref-list>
    <title>References</title>

      <ref id="bib1.bib1"><label>1</label><mixed-citation>Abalos, M., Legras, B., Ploeger, F. and Randel, W. J.: Evaluating the
advective Brewer-Dobson circulation in three reanalyses for the period
1979–2012, J. Geophys. Res.-Atmos., 120, 7534–7554,
<ext-link xlink:href="http://dx.doi.org/10.1002/2015JD023182" ext-link-type="DOI">10.1002/2015JD023182</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bib2"><label>2</label><mixed-citation>
Anstey, J. A., Scinocca, J. F., and Keller, M.: Simulating the QBO in an
atmospheric general circulation model: sensitivity to resolved and
parameterized forcing, J. Atmos. Sci., 73, 1649–1665, 2016.</mixed-citation></ref>
      <ref id="bib1.bib3"><label>3</label><mixed-citation>Baldwin, M. P. and Gray, L. J.: Tropical stratospheric zonal winds in ECMWF
ERA-40 reanalysis, rocketsonde data, and rawinsonde data, Geophys. Res.
Lett., 32, L09806, <ext-link xlink:href="http://dx.doi.org/10.1029/2004GL022328" ext-link-type="DOI">10.1029/2004GL022328</ext-link>, 2005.</mixed-citation></ref>
      <ref id="bib1.bib4"><label>4</label><mixed-citation>
Baldwin, M. P., Gray, L. J., Dunkerton, T. J., Hamilton, K., Haynes, P. H.,
Randel, W. J., Holton, J. R., Alexander, M. J., Hirota, I., Horinouchi, T.,
Jones, D. B. A., Kinnersley, J. S., Marquardt, C., Sato, K., and Takahashi,
M.: The quasi-biennial oscillation, Rev. Geophys., 39, 179–229, 2001.</mixed-citation></ref>
      <ref id="bib1.bib5"><label>5</label><mixed-citation>Baker, W. E., Atlas, R., Cardinali, C., Clement, A., Emmitt, G., Gentry, B.
M., Hardesty, R. M., Källén, E., Kavaya, M. J., Langland, R., Ma,
Z.,Masutani, M., McCarty, W., Pierce, R. B., Pu, Z., Riishojgaard, L. P.,
Ryan, J., Tucker, S., Weissmann, M., and Yoe, J. G.: LIDAR-MEASURED WIND
PROFILES The Missing Link in the Global Observing System, B. Am. Meteorol.
Soc., 95, 543–564, <ext-link xlink:href="http://dx.doi.org/10.1175/BAMS-D-12-00164.1" ext-link-type="DOI">10.1175/BAMS-D-12-00164.1</ext-link>, 2014.</mixed-citation></ref>
      <ref id="bib1.bib6"><label>6</label><mixed-citation>Boer, G. and Hamilton, K.: QBO influence on extratropical predictive skill,
Clim. Dynam., 31, 987–1000, <ext-link xlink:href="http://dx.doi.org/10.1007/s00382-008-0379-5" ext-link-type="DOI">10.1007/s00382-008-0379-5</ext-link>, 2008.</mixed-citation></ref>
      <ref id="bib1.bib7"><label>7</label><mixed-citation>
Coy, L. and Swinbank, R.: Characteristics of stratospheric winds and
temperatures produced by data assimilation, J. Geophys. Res., 102,
25763–25781, 1997.</mixed-citation></ref>
      <ref id="bib1.bib8"><label>8</label><mixed-citation>Coy, L., Wargan, K., Molod, A., McCarty, W., and Pawson, S: Structure and
Dynamics of the Quasi-Biennial Oscillation in MERRA-2, J. Climate,
<ext-link xlink:href="http://dx.doi.org/10.1175/JCLI-D-15-0809.1" ext-link-type="DOI">10.1175/JCLI-D-15-0809.1</ext-link>, online first, 2016.</mixed-citation></ref>
      <ref id="bib1.bib9"><label>9</label><mixed-citation>
Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P.,
Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N.,
Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S.
B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler,
M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J.-J.,
Park, B.-K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.-N., and
Vitart, F.: The ERA-Interim reanalysis: configuration and performance of the
data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553–597, 2011.</mixed-citation></ref>
      <ref id="bib1.bib10"><label>10</label><mixed-citation>
Durre, I., Russell S. V., and David B. W.: Overview of the Integrated Global
Radiosonde Archive, J. Climate, 19, 53–68, 2006.</mixed-citation></ref>
      <ref id="bib1.bib11"><label>11</label><mixed-citation>
Errera, Q., Fujiwara, M., Long, C., and Jackson, D.: Report on the 10th SPARC
data assimilation workshop and the 2014 SPARC Reanalysis Intercomparison
Project (S-RIP) workshop, SPARC Newsletter, 44, 31–37, 2015.</mixed-citation></ref>
      <ref id="bib1.bib12"><label>12</label><mixed-citation>Freie Universität Berlin: The Quasi-Biennial-Oscillation (QBO) Data
Serie, available at:
<uri>http://www.geo.fu-berlin.de/en/met/ag/strat/produkte/qbo/</uri>, last access:
1 June 2016.</mixed-citation></ref>
      <ref id="bib1.bib13"><label>13</label><mixed-citation>
Fujiwara, M. and Jackson, D.: SPARC Reanalysis Intercomparison Project
(S-RIP) Planning Meeting, 29 April–1 May 2013, Exeter, UK: SPARC Newsletter,
41, 52–55, 2013.</mixed-citation></ref>
      <ref id="bib1.bib14"><label>14</label><mixed-citation>Fujiwara, M., Hibino, T., Mehta, S. K., Gray, L., Mitchell, D., and Anstey,
J.: Global temperature response to the major volcanic eruptions in multiple
reanalysis data sets, Atmos. Chem. Phys., 15, 13507–13518,
<ext-link xlink:href="http://dx.doi.org/10.5194/acp-15-13507-2015" ext-link-type="DOI">10.5194/acp-15-13507-2015</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bib15"><label>15</label><mixed-citation>
Hamilton, K.: Mean wind evolution through the quasi-biennial cycle of the
tropical lower stratosphere, J. Atmos. Sci., 41, 2113–2125, 1984.</mixed-citation></ref>
      <ref id="bib1.bib16"><label>16</label><mixed-citation>
Hamilton, K. and Yuan, L.: Experiments on tropical stratospheric mean wind
variations in a spectral general circulation model, J. Atmos. Sci., 49,
2464–2483, 1992.</mixed-citation></ref>
      <ref id="bib1.bib17"><label>17</label><mixed-citation>
Hamilton, K., Hertzog, A., and Vial, F.: and Stenchikov, G., Longitudinal
variation of the stratospheric quasi-biennial oscillation, J. Atmos. Sci.,
61, 383–402, 2004.</mixed-citation></ref>
      <ref id="bib1.bib18"><label>18</label><mixed-citation>
Huesmann, A. S. and Hitchman, M. H.: The stratospheric quasi-biennial
oscillation in the NCEP reanalyses: Climatological structures, J. Geophys.
Res., 106, 11859–11874, 2001.</mixed-citation></ref>
      <ref id="bib1.bib19"><label>19</label><mixed-citation>Huesmann, A. S. and Hitchman, M. H.: The 1978 shift in the NCEP reanalysis
stratospheric quasi-biennial oscillation, Geophys. Res. Lett., 30, 1048,
<ext-link xlink:href="http://dx.doi.org/10.1029/2002GL016323" ext-link-type="DOI">10.1029/2002GL016323</ext-link>, 2003.</mixed-citation></ref>
      <ref id="bib1.bib20"><label>20</label><mixed-citation>
Iida, C., Hirooka, T., and Eguchi, N.: Circulation changes in the
stratosphere and mesosphere during the stratospheric sudden warming event in
January 2009, J. Geophys. Res.-Atmos., 119, 7104–7115, 2014.</mixed-citation></ref>
      <ref id="bib1.bib21"><label>21</label><mixed-citation>
Kalnay, E.: Atmospheric Modeling, Data Assimilation and Predictability,
Cambridge University Press, 341 pp., 2003.</mixed-citation></ref>
      <ref id="bib1.bib22"><label>22</label><mixed-citation>
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L.,
Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Chelliah, M.,
Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski, C., Wang,
J., Leetmaa, A., Reynolds, R., Jenne, R., and Joseph, D.: The NCEP/NCAR
40-year reanalyses project, B. Am. Meteorol. Soc., 77, 437–471, 1996.</mixed-citation></ref>
      <ref id="bib1.bib23"><label>23</label><mixed-citation>
Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S.-K., Hnilo, J. J.,
Fiorino, M., and Potter, G. L.: NCEP–DOE AMIP-II Reanalysis (R-2), B. Am.
Meteorol. Soc., 83, 1631–1643, 2002.</mixed-citation></ref>
      <ref id="bib1.bib24"><label>24</label><mixed-citation>
Kawatani, Y. and Hamilton, K.: Weakened stratospheric quasibiennial
oscillation driven by increased tropical mean upwelling, Nature, 497,
478–481, 2013.</mixed-citation></ref>
      <ref id="bib1.bib25"><label>25</label><mixed-citation>
Kawatani, Y., Sato, K., Dunkerton, T. J., Watanabe, S., Miyahara, S., and
Takahashi, M.: The roles of equatorial trapped waves and internal
inertia-gravity waves in driving the quasi-biennial oscillation. Part I:
Zonal mean wave forcing, J. Atmos. Sci., 67, 963–980, 2010.</mixed-citation></ref>
      <ref id="bib1.bib26"><label>26</label><mixed-citation>Kawatani, Y., Hamilton, K., and Watanabe, S.: The quasi-biennial oscillation
in a double CO<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">2</mml:mn></mml:msub></mml:math></inline-formula> climate, J. Atmos. Sci., 68, 265–283, 2011.</mixed-citation></ref>
      <ref id="bib1.bib27"><label>27</label><mixed-citation>Kim, Y.-H. and Chun, H.-Y.: Momentum forcing of the quasi-biennial
oscillation by equatorial waves in recent reanalyses, Atmos. Chem. Phys., 15,
6577–6587, <ext-link xlink:href="http://dx.doi.org/10.5194/acp-15-6577-2015" ext-link-type="DOI">10.5194/acp-15-6577-2015</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bib28"><label>28</label><mixed-citation>Kobayashi, C., Endo, H., Ota, Y., Kobayashi, S., Onoda, H., Harada, Y.,
Onogi, K., and Kamahori, H.: Preliminary results of the JRA-55C, an
atmospheric reanalysis assimilating conventional observations only. Sci.
Online Lett. on the Atmos., 10, 78–82, <ext-link xlink:href="http://dx.doi.org/10.2151/sola.2014-016" ext-link-type="DOI">10.2151/sola.2014-016</ext-link>, 2014.</mixed-citation></ref>
      <ref id="bib1.bib29"><label>29</label><mixed-citation>
Kobayashi, S., Ota, Y., Harada, Y., Ebita, A., Moriya, M., Hirokatsu, O.,
Onogi, K., Kamahori, H., Kobayashi, C., Endo, H., Miyaoka, K., and Takahashi,
K.: The JRA-55 Reanalysis: general specifications and basic characteristics,
J. Meteorol. Soc. Jpn., 93, 5–48, 2015.</mixed-citation></ref>
      <ref id="bib1.bib30"><label>30</label><mixed-citation>Miyazaki, K., Iwasaki, T., Kawatani, Y., Kobayashi, C., Sugawara, S., and
Hegglin, M. I.: Inter-comparison of stratospheric mean-meridional circulation
and eddy mixing among six reanalysis data sets, Atmos. Chem. Phys., 16,
6131–6152, <ext-link xlink:href="http://dx.doi.org/10.5194/acp-16-6131-2016" ext-link-type="DOI">10.5194/acp-16-6131-2016</ext-link>, 2016.</mixed-citation></ref>
      <ref id="bib1.bib31"><label>31</label><mixed-citation>Molod, A., Takacs, L., Suarez, M., and Bacmeister, J.: Development of the
GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2,
Geosci. Model Dev., 8, 1339–1356, <ext-link xlink:href="http://dx.doi.org/10.5194/gmd-8-1339-2015" ext-link-type="DOI">10.5194/gmd-8-1339-2015</ext-link>, 2015.</mixed-citation></ref>
      <ref id="bib1.bib32"><label>32</label><mixed-citation>
Naujokat, B.: An update of the observed quasi-biennial oscillation of the
stratospheric zonal wind over the tropics, J. Atmos. Sci., 43, 1873–1877,
1986.</mixed-citation></ref>
      <ref id="bib1.bib33"><label>33</label><mixed-citation>NOAA: Integrated Global Radiosonde Archive (IGRA), available at:
<uri>https://www.ncdc.noaa.gov/data-access/weather-balloon/integrated-global-radiosonde-archive</uri>,
last access: 1 June 2016.</mixed-citation></ref>
      <ref id="bib1.bib34"><label>34</label><mixed-citation>
Onogi, K., J. Tsutsui, J., Koide, H., Sakamoto, M., Kobayashi, S.,
Hatsushika, H., Matsumoto, T., Yamazaki, N., Kamahori, H., Takahashi, K.,
Kadokura, S., Wada, K., Kato, K., Oyama, R., Ose, T., Mannoji, N., and Taira,
R.: The JRA-25 reanalysis, J. Meteorol. Soc. Jpn., 85, 369–432, 2007.</mixed-citation></ref>
      <ref id="bib1.bib35"><label>35</label><mixed-citation>
Orr, A., Bechtold, P., Scinocca, J., Ern, M., and Janiskova, M.: Improved
middle atmosphere climate and forecasts in the ECMWF model through a
nonorographic gravity wave drag parameterization, J. Climate, 23, 5905–5926,
2010.</mixed-citation></ref>
      <ref id="bib1.bib36"><label>36</label><mixed-citation>Pascoe, C. L., Gray, L. J., Crooks, S. A., Juckes, M. N., and Baldwin, M. P.:
The quasi-biennial oscillation: Analysis using ERA-40 data, J. Geophys. Res.,
110, D08105, <ext-link xlink:href="http://dx.doi.org/10.1029/2004JD004941" ext-link-type="DOI">10.1029/2004JD004941</ext-link>, 2005.</mixed-citation></ref>
      <ref id="bib1.bib37"><label>37</label><mixed-citation>
Pawson, S. and Fiorino, M.: A comparison of reanalyses in the tropical
stratosphere, Part 2: the quasi-biennial oscillation, Clim. Dynam., 14,
645–658, 1998.</mixed-citation></ref>
      <ref id="bib1.bib38"><label>38</label><mixed-citation>
Pawson, S. and Fiorino, M.: A comparison of reanalyses in the tropical
stratosphere, Part 3: Inclusion of the pre-satellite data era, Clim. Dynam.,
15, 241–250, 1999.</mixed-citation></ref>
      <ref id="bib1.bib39"><label>39</label><mixed-citation>Podglajen, A., Hertzog, A., Plougonven, R., and Žagar, N.: Assessment of
the accuracy of (re)analyses in the equatorial lower stratosphere, J.
Geophys. Res., 119, 11166–11188, <ext-link xlink:href="http://dx.doi.org/10.1002/2014JD021849" ext-link-type="DOI">10.1002/2014JD021849</ext-link>, 2014.</mixed-citation></ref>
      <ref id="bib1.bib40"><label>40</label><mixed-citation>
Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J.,
Liu, E., Bosilovich, M. G., Schubert, S. D., Takacs, L., Kim, G-K., Bloom,
S., Chen, J., Collins, D., Conaty, A., da Silva, A., Gu, W., Joiner, J.,
Koster, R. D., Lucchesi, R., Molod, A., Owens, T., Pawson, S., Pegion, P.,
Redder, C. R., Reichle, R., Robertson, F. R., Ruddick, A. G., Sienkiewicz,
M., and Woollen, J.: MERRA: NASA's Modern-Era Retrospective Analysis for
Research and Applications, J. Climate, 24, 3624–3648, 2011.</mixed-citation></ref>
      <ref id="bib1.bib41"><label>41</label><mixed-citation>
Randel, W. J., Wu, F., Swinbank, R., Nash, J., and O'Neill, A.: Global QBO
circulation derived from UKMO stratospheric analyses, J. Atmos. Sci., 56,
457–474, 1999.</mixed-citation></ref>
      <ref id="bib1.bib42"><label>42</label><mixed-citation>
Randel, W. J., Udelhofen, F., Fleming, E., Geller, M., Gelman, M., Hamilton,
K., Karoly, D., Ortland, D., Pawson, S., Swinbank, R., Wu, F., Baldwin, M.,
Chanin, M., Keckhut, P., Labitzke, K., Remsberg, E., Simmons, A., and Wu, D.:
The SPARC intercomparison of middle-atmosphere climatologies, J. Climate, 17,
986–1003, 2004.</mixed-citation></ref>
      <ref id="bib1.bib43"><label>43</label><mixed-citation>
Reed, R. J.: Evidence of geostrophic motion in the equatorial stratosphere,
Q. J. R. Meteor. Soc., 88, 324–327, 1962.</mixed-citation></ref>
      <ref id="bib1.bib44"><label>44</label><mixed-citation>Saha, S., Nadiga, S., Thiaw, C., Wang, J., Wang, W., Zhang, Q., van den Dool,
H. M., Pan, H.-L., Moorthi, S., Behringer, D., Stokes, D., Peña, M.,
Lord, S., White, G., Ebisuzaki, W., Peng, P., and Xie, P.: The NCEP climate
forecast system, J. Climate, 19, 3483–3517, 2006.
 </mixed-citation></ref><?xmltex \hack{\newpage}?>
      <ref id="bib1.bib45"><label>45</label><mixed-citation>
Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., Tripp, P.,
Kistler, R., Woollen, J., Behringer, D., Liu, H., Stokes, D., Grumbine, R.,
Gayno, G., Wang, J., Hou, Y.-T., Chuang, H.-Y., Juang, H.-M. H., Sela, J.,
Iredell, M., Treadon, R., Kleist, D., Delst, P. V., Keyser, D., Derber, J.,
Ek, M., Meng, J., Wei, H, Yang, R., Lord, S., Dool, H. V. D., Kumar, A.,
Wang, W., Long, C., Chelliah, M., Xue, Y., Huang, B., Schemm, J.-K.,
Ebisuzaki, W., Lin, R., Xie, P., Chen, M., Zhou, S., Higgins, W., Zou, C.-Z.,
Liu, Q., Chen, Y., Han, Y., Cucurull, L., Reynolds, R. W., Rutledge, G., and
Goldberg, M.: The NCEP Climate Forecast System Reanalysis, B. Am. Meteorol.
Soc., 91, 1015–1057, 2010.</mixed-citation></ref>
      <ref id="bib1.bib46"><label>46</label><mixed-citation>Trenberth, K. E.: Global analyses from ECMWF and atlas of 1000 to 10 Mb
circulation statistics, NCAR Technical Note NCAR/TN-373<inline-formula><mml:math display="inline"><mml:mo>+</mml:mo></mml:math></inline-formula>STR,
<ext-link xlink:href="http://dx.doi.org/10.5065/D6WM1BCN" ext-link-type="DOI">10.5065/D6WM1BCN</ext-link>, 1992.</mixed-citation></ref>
      <ref id="bib1.bib47"><label>47</label><mixed-citation>
Uppala S. M., Kallberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D.,
Fiorino, M., Gibson, J. K., Haseler, J., Hernandez, A., Kelly, G. A., Li, X.,
Onogi, K., Saarinen, S., Sokka, N., Allan, R. P., Andersson, E., Arpe, K.,
Balmaseda, M. A., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann,
N., Caires, S., Chevallier, F., Dethof, A., Dragosavac, M., Fisher, M.,
Fuentes, M., Hagemann, S., Holm, E., Hoskins, B. J., Isaksen, L., Janssen, P.
A. E. M., Jenne, R., McNally, A. P., Mahfouf, J. F., Morcrette, J. J.,
Rayner, N. A., Saunders, R. W., Simon, P., Sterl, A., Trenberth, K. E.,
Untch, A., Vasiljevic, D., Viterbo, P., and Woollen, J.: The ERA-40
re-analysis, Q. J. Roy. Meteor. Soc., 131, 2961–3012, 2005.</mixed-citation></ref>
      <ref id="bib1.bib48"><label>48</label><mixed-citation>
Wallace, J. M.: General circulation of the tropical lower stratosphere, Rev.
Geophys., 11, 191–222, 1973.</mixed-citation></ref>
      <ref id="bib1.bib49"><label>49</label><mixed-citation>Yoo, C. and Son, S.-W.: Modulation of the boreal wintertime Madden-Julian
oscillation by the stratospheric quasi-biennial oscillation, Geophys. Res.
Lett., 43, 1392–1398, <ext-link xlink:href="http://dx.doi.org/10.1002/2016GL067762" ext-link-type="DOI">10.1002/2016GL067762</ext-link>, 2016.</mixed-citation></ref>

  </ref-list><app-group content-type="float"><app><title/>

    </app></app-group></back>
    <!--<article-title-html>Representation of the tropical stratospheric zonal wind in global
atmospheric reanalyses</article-title-html>
<abstract-html><p class="p">This paper reports on a
project to compare the representation of the monthly-mean zonal wind in the
equatorial stratosphere among major global atmospheric reanalysis data sets.
The degree of disagreement among the reanalyses is characterized by the
standard deviation (SD) of the monthly-mean zonal wind and this depends on
latitude, longitude, height, and the phase of the quasi-biennial oscillation
(QBO). At each height the SD displays a prominent equatorial maximum,
indicating the particularly challenging nature of the reanalysis problem in
the low-latitude stratosphere. At 50–70 hPa the geographical distributions
of SD are closely related to the density of radiosonde observations. The
largest SD values are over the central Pacific, where few in situ
observations are available. At 10–20 hPa the spread among the reanalyses
and differences with in situ observations both depend significantly on the
QBO phase. Notably the easterly-to-westerly phase transitions in all the
reanalyses except MERRA are delayed relative to those directly observed in
Singapore. In addition, the timing of the easterly-to-westerly phase
transitions displays considerable variability among the different reanalyses
and this spread is much larger than for the timing of the
westerly-to-easterly phase changes. The eddy component in the monthly-mean
zonal wind near the Equator is dominated by zonal wavenumber 1 and 2
quasi-stationary planetary waves propagating from midlatitudes in the
westerly phase of the QBO. There generally is considerable disagreement among
the reanalyses in the details of the quasi-stationary waves near the Equator.
At each level, there is a tendency for the agreement to be best near the
longitude of Singapore, suggesting that the Singapore observations act as a
strong constraint on all the reanalyses. Our measures of the quality of the
reanalysis clearly show systematic improvement over the period considered
(1979–2012). The SD among the reanalysis declines significantly over the
record, although the geographical pattern of SD remains nearly constant.</p></abstract-html>
<ref-html id="bib1.bib1"><label>1</label><mixed-citation>
Abalos, M., Legras, B., Ploeger, F. and Randel, W. J.: Evaluating the
advective Brewer-Dobson circulation in three reanalyses for the period
1979–2012, J. Geophys. Res.-Atmos., 120, 7534–7554,
<a href="http://dx.doi.org/10.1002/2015JD023182" target="_blank">doi:10.1002/2015JD023182</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib2"><label>2</label><mixed-citation>
Anstey, J. A., Scinocca, J. F., and Keller, M.: Simulating the QBO in an
atmospheric general circulation model: sensitivity to resolved and
parameterized forcing, J. Atmos. Sci., 73, 1649–1665, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib3"><label>3</label><mixed-citation>
Baldwin, M. P. and Gray, L. J.: Tropical stratospheric zonal winds in ECMWF
ERA-40 reanalysis, rocketsonde data, and rawinsonde data, Geophys. Res.
Lett., 32, L09806, <a href="http://dx.doi.org/10.1029/2004GL022328" target="_blank">doi:10.1029/2004GL022328</a>, 2005.
</mixed-citation></ref-html>
<ref-html id="bib1.bib4"><label>4</label><mixed-citation>
Baldwin, M. P., Gray, L. J., Dunkerton, T. J., Hamilton, K., Haynes, P. H.,
Randel, W. J., Holton, J. R., Alexander, M. J., Hirota, I., Horinouchi, T.,
Jones, D. B. A., Kinnersley, J. S., Marquardt, C., Sato, K., and Takahashi,
M.: The quasi-biennial oscillation, Rev. Geophys., 39, 179–229, 2001.
</mixed-citation></ref-html>
<ref-html id="bib1.bib5"><label>5</label><mixed-citation>
Baker, W. E., Atlas, R., Cardinali, C., Clement, A., Emmitt, G., Gentry, B.
M., Hardesty, R. M., Källén, E., Kavaya, M. J., Langland, R., Ma,
Z.,Masutani, M., McCarty, W., Pierce, R. B., Pu, Z., Riishojgaard, L. P.,
Ryan, J., Tucker, S., Weissmann, M., and Yoe, J. G.: LIDAR-MEASURED WIND
PROFILES The Missing Link in the Global Observing System, B. Am. Meteorol.
Soc., 95, 543–564, <a href="http://dx.doi.org/10.1175/BAMS-D-12-00164.1" target="_blank">doi:10.1175/BAMS-D-12-00164.1</a>, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib6"><label>6</label><mixed-citation>
Boer, G. and Hamilton, K.: QBO influence on extratropical predictive skill,
Clim. Dynam., 31, 987–1000, <a href="http://dx.doi.org/10.1007/s00382-008-0379-5" target="_blank">doi:10.1007/s00382-008-0379-5</a>, 2008.
</mixed-citation></ref-html>
<ref-html id="bib1.bib7"><label>7</label><mixed-citation>
Coy, L. and Swinbank, R.: Characteristics of stratospheric winds and
temperatures produced by data assimilation, J. Geophys. Res., 102,
25763–25781, 1997.
</mixed-citation></ref-html>
<ref-html id="bib1.bib8"><label>8</label><mixed-citation>
Coy, L., Wargan, K., Molod, A., McCarty, W., and Pawson, S: Structure and
Dynamics of the Quasi-Biennial Oscillation in MERRA-2, J. Climate,
<a href="http://dx.doi.org/10.1175/JCLI-D-15-0809.1" target="_blank">doi:10.1175/JCLI-D-15-0809.1</a>, online first, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib9"><label>9</label><mixed-citation>
Dee, D. P., Uppala, S. M., Simmons, A. J., Berrisford, P., Poli, P.,
Kobayashi, S., Andrae, U., Balmaseda, M. A., Balsamo, G., Bauer, P.,
Bechtold, P., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann, N.,
Delsol, C., Dragani, R., Fuentes, M., Geer, A. J., Haimberger, L., Healy, S.
B., Hersbach, H., Hólm, E. V., Isaksen, L., Kållberg, P., Köhler,
M., Matricardi, M., McNally, A. P., Monge-Sanz, B. M., Morcrette, J.-J.,
Park, B.-K., Peubey, C., de Rosnay, P., Tavolato, C., Thépaut, J.-N., and
Vitart, F.: The ERA-Interim reanalysis: configuration and performance of the
data assimilation system, Q. J. Roy. Meteor. Soc., 137, 553–597, 2011.
</mixed-citation></ref-html>
<ref-html id="bib1.bib10"><label>10</label><mixed-citation>
Durre, I., Russell S. V., and David B. W.: Overview of the Integrated Global
Radiosonde Archive, J. Climate, 19, 53–68, 2006.
</mixed-citation></ref-html>
<ref-html id="bib1.bib11"><label>11</label><mixed-citation>
Errera, Q., Fujiwara, M., Long, C., and Jackson, D.: Report on the 10th SPARC
data assimilation workshop and the 2014 SPARC Reanalysis Intercomparison
Project (S-RIP) workshop, SPARC Newsletter, 44, 31–37, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib12"><label>12</label><mixed-citation>
Freie Universität Berlin: The Quasi-Biennial-Oscillation (QBO) Data
Serie, available at:
<a href="http://www.geo.fu-berlin.de/en/met/ag/strat/produkte/qbo/" target="_blank">http://www.geo.fu-berlin.de/en/met/ag/strat/produkte/qbo/</a>, last access:
1 June 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib13"><label>13</label><mixed-citation>
Fujiwara, M. and Jackson, D.: SPARC Reanalysis Intercomparison Project
(S-RIP) Planning Meeting, 29 April–1 May 2013, Exeter, UK: SPARC Newsletter,
41, 52–55, 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib14"><label>14</label><mixed-citation>
Fujiwara, M., Hibino, T., Mehta, S. K., Gray, L., Mitchell, D., and Anstey,
J.: Global temperature response to the major volcanic eruptions in multiple
reanalysis data sets, Atmos. Chem. Phys., 15, 13507–13518,
<a href="http://dx.doi.org/10.5194/acp-15-13507-2015" target="_blank">doi:10.5194/acp-15-13507-2015</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib15"><label>15</label><mixed-citation>
Hamilton, K.: Mean wind evolution through the quasi-biennial cycle of the
tropical lower stratosphere, J. Atmos. Sci., 41, 2113–2125, 1984.
</mixed-citation></ref-html>
<ref-html id="bib1.bib16"><label>16</label><mixed-citation>
Hamilton, K. and Yuan, L.: Experiments on tropical stratospheric mean wind
variations in a spectral general circulation model, J. Atmos. Sci., 49,
2464–2483, 1992.
</mixed-citation></ref-html>
<ref-html id="bib1.bib17"><label>17</label><mixed-citation>
Hamilton, K., Hertzog, A., and Vial, F.: and Stenchikov, G., Longitudinal
variation of the stratospheric quasi-biennial oscillation, J. Atmos. Sci.,
61, 383–402, 2004.
</mixed-citation></ref-html>
<ref-html id="bib1.bib18"><label>18</label><mixed-citation>
Huesmann, A. S. and Hitchman, M. H.: The stratospheric quasi-biennial
oscillation in the NCEP reanalyses: Climatological structures, J. Geophys.
Res., 106, 11859–11874, 2001.
</mixed-citation></ref-html>
<ref-html id="bib1.bib19"><label>19</label><mixed-citation>
Huesmann, A. S. and Hitchman, M. H.: The 1978 shift in the NCEP reanalysis
stratospheric quasi-biennial oscillation, Geophys. Res. Lett., 30, 1048,
<a href="http://dx.doi.org/10.1029/2002GL016323" target="_blank">doi:10.1029/2002GL016323</a>, 2003.
</mixed-citation></ref-html>
<ref-html id="bib1.bib20"><label>20</label><mixed-citation>
Iida, C., Hirooka, T., and Eguchi, N.: Circulation changes in the
stratosphere and mesosphere during the stratospheric sudden warming event in
January 2009, J. Geophys. Res.-Atmos., 119, 7104–7115, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib21"><label>21</label><mixed-citation>
Kalnay, E.: Atmospheric Modeling, Data Assimilation and Predictability,
Cambridge University Press, 341 pp., 2003.
</mixed-citation></ref-html>
<ref-html id="bib1.bib22"><label>22</label><mixed-citation>
Kalnay, E., Kanamitsu, M., Kistler, R., Collins, W., Deaven, D., Gandin, L.,
Iredell, M., Saha, S., White, G., Woollen, J., Zhu, Y., Chelliah, M.,
Ebisuzaki, W., Higgins, W., Janowiak, J., Mo, K. C., Ropelewski, C., Wang,
J., Leetmaa, A., Reynolds, R., Jenne, R., and Joseph, D.: The NCEP/NCAR
40-year reanalyses project, B. Am. Meteorol. Soc., 77, 437–471, 1996.
</mixed-citation></ref-html>
<ref-html id="bib1.bib23"><label>23</label><mixed-citation>
Kanamitsu, M., Ebisuzaki, W., Woollen, J., Yang, S.-K., Hnilo, J. J.,
Fiorino, M., and Potter, G. L.: NCEP–DOE AMIP-II Reanalysis (R-2), B. Am.
Meteorol. Soc., 83, 1631–1643, 2002.
</mixed-citation></ref-html>
<ref-html id="bib1.bib24"><label>24</label><mixed-citation>
Kawatani, Y. and Hamilton, K.: Weakened stratospheric quasibiennial
oscillation driven by increased tropical mean upwelling, Nature, 497,
478–481, 2013.
</mixed-citation></ref-html>
<ref-html id="bib1.bib25"><label>25</label><mixed-citation>
Kawatani, Y., Sato, K., Dunkerton, T. J., Watanabe, S., Miyahara, S., and
Takahashi, M.: The roles of equatorial trapped waves and internal
inertia-gravity waves in driving the quasi-biennial oscillation. Part I:
Zonal mean wave forcing, J. Atmos. Sci., 67, 963–980, 2010.
</mixed-citation></ref-html>
<ref-html id="bib1.bib26"><label>26</label><mixed-citation>
Kawatani, Y., Hamilton, K., and Watanabe, S.: The quasi-biennial oscillation
in a double CO<sub>2</sub> climate, J. Atmos. Sci., 68, 265–283, 2011.
</mixed-citation></ref-html>
<ref-html id="bib1.bib27"><label>27</label><mixed-citation>
Kim, Y.-H. and Chun, H.-Y.: Momentum forcing of the quasi-biennial
oscillation by equatorial waves in recent reanalyses, Atmos. Chem. Phys., 15,
6577–6587, <a href="http://dx.doi.org/10.5194/acp-15-6577-2015" target="_blank">doi:10.5194/acp-15-6577-2015</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib28"><label>28</label><mixed-citation>
Kobayashi, C., Endo, H., Ota, Y., Kobayashi, S., Onoda, H., Harada, Y.,
Onogi, K., and Kamahori, H.: Preliminary results of the JRA-55C, an
atmospheric reanalysis assimilating conventional observations only. Sci.
Online Lett. on the Atmos., 10, 78–82, <a href="http://dx.doi.org/10.2151/sola.2014-016" target="_blank">doi:10.2151/sola.2014-016</a>, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib29"><label>29</label><mixed-citation>
Kobayashi, S., Ota, Y., Harada, Y., Ebita, A., Moriya, M., Hirokatsu, O.,
Onogi, K., Kamahori, H., Kobayashi, C., Endo, H., Miyaoka, K., and Takahashi,
K.: The JRA-55 Reanalysis: general specifications and basic characteristics,
J. Meteorol. Soc. Jpn., 93, 5–48, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib30"><label>30</label><mixed-citation>
Miyazaki, K., Iwasaki, T., Kawatani, Y., Kobayashi, C., Sugawara, S., and
Hegglin, M. I.: Inter-comparison of stratospheric mean-meridional circulation
and eddy mixing among six reanalysis data sets, Atmos. Chem. Phys., 16,
6131–6152, <a href="http://dx.doi.org/10.5194/acp-16-6131-2016" target="_blank">doi:10.5194/acp-16-6131-2016</a>, 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib31"><label>31</label><mixed-citation>
Molod, A., Takacs, L., Suarez, M., and Bacmeister, J.: Development of the
GEOS-5 atmospheric general circulation model: evolution from MERRA to MERRA2,
Geosci. Model Dev., 8, 1339–1356, <a href="http://dx.doi.org/10.5194/gmd-8-1339-2015" target="_blank">doi:10.5194/gmd-8-1339-2015</a>, 2015.
</mixed-citation></ref-html>
<ref-html id="bib1.bib32"><label>32</label><mixed-citation>
Naujokat, B.: An update of the observed quasi-biennial oscillation of the
stratospheric zonal wind over the tropics, J. Atmos. Sci., 43, 1873–1877,
1986.
</mixed-citation></ref-html>
<ref-html id="bib1.bib33"><label>33</label><mixed-citation>
NOAA: Integrated Global Radiosonde Archive (IGRA), available at:
<a href="https://www.ncdc.noaa.gov/data-access/weather-balloon/integrated-global-radiosonde-archive" target="_blank">https://www.ncdc.noaa.gov/data-access/weather-balloon/integrated-global-radiosonde-archive</a>,
last access: 1 June 2016.
</mixed-citation></ref-html>
<ref-html id="bib1.bib34"><label>34</label><mixed-citation>
Onogi, K., J. Tsutsui, J., Koide, H., Sakamoto, M., Kobayashi, S.,
Hatsushika, H., Matsumoto, T., Yamazaki, N., Kamahori, H., Takahashi, K.,
Kadokura, S., Wada, K., Kato, K., Oyama, R., Ose, T., Mannoji, N., and Taira,
R.: The JRA-25 reanalysis, J. Meteorol. Soc. Jpn., 85, 369–432, 2007.
</mixed-citation></ref-html>
<ref-html id="bib1.bib35"><label>35</label><mixed-citation>
Orr, A., Bechtold, P., Scinocca, J., Ern, M., and Janiskova, M.: Improved
middle atmosphere climate and forecasts in the ECMWF model through a
nonorographic gravity wave drag parameterization, J. Climate, 23, 5905–5926,
2010.
</mixed-citation></ref-html>
<ref-html id="bib1.bib36"><label>36</label><mixed-citation>
Pascoe, C. L., Gray, L. J., Crooks, S. A., Juckes, M. N., and Baldwin, M. P.:
The quasi-biennial oscillation: Analysis using ERA-40 data, J. Geophys. Res.,
110, D08105, <a href="http://dx.doi.org/10.1029/2004JD004941" target="_blank">doi:10.1029/2004JD004941</a>, 2005.
</mixed-citation></ref-html>
<ref-html id="bib1.bib37"><label>37</label><mixed-citation>
Pawson, S. and Fiorino, M.: A comparison of reanalyses in the tropical
stratosphere, Part 2: the quasi-biennial oscillation, Clim. Dynam., 14,
645–658, 1998.
</mixed-citation></ref-html>
<ref-html id="bib1.bib38"><label>38</label><mixed-citation>
Pawson, S. and Fiorino, M.: A comparison of reanalyses in the tropical
stratosphere, Part 3: Inclusion of the pre-satellite data era, Clim. Dynam.,
15, 241–250, 1999.
</mixed-citation></ref-html>
<ref-html id="bib1.bib39"><label>39</label><mixed-citation>
Podglajen, A., Hertzog, A., Plougonven, R., and Žagar, N.: Assessment of
the accuracy of (re)analyses in the equatorial lower stratosphere, J.
Geophys. Res., 119, 11166–11188, <a href="http://dx.doi.org/10.1002/2014JD021849" target="_blank">doi:10.1002/2014JD021849</a>, 2014.
</mixed-citation></ref-html>
<ref-html id="bib1.bib40"><label>40</label><mixed-citation>
Rienecker, M. M., Suarez, M. J., Gelaro, R., Todling, R., Bacmeister, J.,
Liu, E., Bosilovich, M. G., Schubert, S. D., Takacs, L., Kim, G-K., Bloom,
S., Chen, J., Collins, D., Conaty, A., da Silva, A., Gu, W., Joiner, J.,
Koster, R. D., Lucchesi, R., Molod, A., Owens, T., Pawson, S., Pegion, P.,
Redder, C. R., Reichle, R., Robertson, F. R., Ruddick, A. G., Sienkiewicz,
M., and Woollen, J.: MERRA: NASA's Modern-Era Retrospective Analysis for
Research and Applications, J. Climate, 24, 3624–3648, 2011.
</mixed-citation></ref-html>
<ref-html id="bib1.bib41"><label>41</label><mixed-citation>
Randel, W. J., Wu, F., Swinbank, R., Nash, J., and O'Neill, A.: Global QBO
circulation derived from UKMO stratospheric analyses, J. Atmos. Sci., 56,
457–474, 1999.
</mixed-citation></ref-html>
<ref-html id="bib1.bib42"><label>42</label><mixed-citation>
Randel, W. J., Udelhofen, F., Fleming, E., Geller, M., Gelman, M., Hamilton,
K., Karoly, D., Ortland, D., Pawson, S., Swinbank, R., Wu, F., Baldwin, M.,
Chanin, M., Keckhut, P., Labitzke, K., Remsberg, E., Simmons, A., and Wu, D.:
The SPARC intercomparison of middle-atmosphere climatologies, J. Climate, 17,
986–1003, 2004.
</mixed-citation></ref-html>
<ref-html id="bib1.bib43"><label>43</label><mixed-citation>
Reed, R. J.: Evidence of geostrophic motion in the equatorial stratosphere,
Q. J. R. Meteor. Soc., 88, 324–327, 1962.
</mixed-citation></ref-html>
<ref-html id="bib1.bib44"><label>44</label><mixed-citation>
Saha, S., Nadiga, S., Thiaw, C., Wang, J., Wang, W., Zhang, Q., van den Dool,
H. M., Pan, H.-L., Moorthi, S., Behringer, D., Stokes, D., Peña, M.,
Lord, S., White, G., Ebisuzaki, W., Peng, P., and Xie, P.: The NCEP climate
forecast system, J. Climate, 19, 3483–3517, 2006.

</mixed-citation></ref-html>
<ref-html id="bib1.bib45"><label>45</label><mixed-citation>
Saha, S., Moorthi, S., Pan, H.-L., Wu, X., Wang, J., Nadiga, S., Tripp, P.,
Kistler, R., Woollen, J., Behringer, D., Liu, H., Stokes, D., Grumbine, R.,
Gayno, G., Wang, J., Hou, Y.-T., Chuang, H.-Y., Juang, H.-M. H., Sela, J.,
Iredell, M., Treadon, R., Kleist, D., Delst, P. V., Keyser, D., Derber, J.,
Ek, M., Meng, J., Wei, H, Yang, R., Lord, S., Dool, H. V. D., Kumar, A.,
Wang, W., Long, C., Chelliah, M., Xue, Y., Huang, B., Schemm, J.-K.,
Ebisuzaki, W., Lin, R., Xie, P., Chen, M., Zhou, S., Higgins, W., Zou, C.-Z.,
Liu, Q., Chen, Y., Han, Y., Cucurull, L., Reynolds, R. W., Rutledge, G., and
Goldberg, M.: The NCEP Climate Forecast System Reanalysis, B. Am. Meteorol.
Soc., 91, 1015–1057, 2010.
</mixed-citation></ref-html>
<ref-html id="bib1.bib46"><label>46</label><mixed-citation>
Trenberth, K. E.: Global analyses from ECMWF and atlas of 1000 to 10 Mb
circulation statistics, NCAR Technical Note NCAR/TN-373+STR,
<a href="http://dx.doi.org/10.5065/D6WM1BCN" target="_blank">doi:10.5065/D6WM1BCN</a>, 1992.
</mixed-citation></ref-html>
<ref-html id="bib1.bib47"><label>47</label><mixed-citation>
Uppala S. M., Kallberg, P. W., Simmons, A. J., Andrae, U., Bechtold, V. D.,
Fiorino, M., Gibson, J. K., Haseler, J., Hernandez, A., Kelly, G. A., Li, X.,
Onogi, K., Saarinen, S., Sokka, N., Allan, R. P., Andersson, E., Arpe, K.,
Balmaseda, M. A., Beljaars, A. C. M., van de Berg, L., Bidlot, J., Bormann,
N., Caires, S., Chevallier, F., Dethof, A., Dragosavac, M., Fisher, M.,
Fuentes, M., Hagemann, S., Holm, E., Hoskins, B. J., Isaksen, L., Janssen, P.
A. E. M., Jenne, R., McNally, A. P., Mahfouf, J. F., Morcrette, J. J.,
Rayner, N. A., Saunders, R. W., Simon, P., Sterl, A., Trenberth, K. E.,
Untch, A., Vasiljevic, D., Viterbo, P., and Woollen, J.: The ERA-40
re-analysis, Q. J. Roy. Meteor. Soc., 131, 2961–3012, 2005.
</mixed-citation></ref-html>
<ref-html id="bib1.bib48"><label>48</label><mixed-citation>
Wallace, J. M.: General circulation of the tropical lower stratosphere, Rev.
Geophys., 11, 191–222, 1973.
</mixed-citation></ref-html>
<ref-html id="bib1.bib49"><label>49</label><mixed-citation>
Yoo, C. and Son, S.-W.: Modulation of the boreal wintertime Madden-Julian
oscillation by the stratospheric quasi-biennial oscillation, Geophys. Res.
Lett., 43, 1392–1398, <a href="http://dx.doi.org/10.1002/2016GL067762" target="_blank">doi:10.1002/2016GL067762</a>, 2016.
</mixed-citation></ref-html>--></article>
