ACPAtmospheric Chemistry and PhysicsACPAtmos. Chem. Phys.1680-7324Copernicus PublicationsGöttingen, Germany10.5194/acp-17-12495-2017In situ temperature measurements in the upper troposphere and
lowermost stratosphere from 2 decades of IAGOS long-term
routine observationBerkesFlorianf.berkes@fz-juelich.dehttps://orcid.org/0000-0002-4558-3196NeisPatrickSchultzMartin G.https://orcid.org/0000-0003-3455-774XBundkeUlrichhttps://orcid.org/0000-0001-5484-8099RohsSusannehttps://orcid.org/0000-0001-5473-2934SmitHerman G. J.WahnerAndreashttps://orcid.org/0000-0001-8948-1928KonopkaPaulBoulangerDamienhttps://orcid.org/0000-0001-6935-1106NédélecPhilippeThouretValeriePetzoldAndreashttps://orcid.org/0000-0002-2504-1680Forschungszentrum Jülich, IEK-8, Jülich, GermanyForschungszentrum Jülich, IEK-7, Jülich, GermanyLaboratoire d'Aérologie, CNRS and Université de Toulouse, Toulouse, FranceFlorian Berkes (f.berkes@fz-juelich.de)23October20171720124951250824May20176June20175September20177September2017This work is licensed under the Creative Commons Attribution 3.0 Unported License. To view a copy of this licence, visit https://creativecommons.org/licenses/by/3.0/This article is available from https://acp.copernicus.org/articles/17/12495/2017/acp-17-12495-2017.htmlThe full text article is available as a PDF file from https://acp.copernicus.org/articles/17/12495/2017/acp-17-12495-2017.pdf
Despite several studies on temperature trends in the tropopause region,
a comprehensive understanding of the evolution of temperatures in this
climate-sensitive region of the atmosphere remains elusive. Here we present
a unique global-scale, long-term data set of high-resolution in situ
temperature data measured aboard passenger aircraft within the European
Research Infrastructure IAGOS (In-service Aircraft for a Global Observing
System; http://www.iagos.org). This data set is used to investigate
temperature trends within the global upper troposphere and lowermost
stratosphere (UTLS, < 13 km) for the period of 1995–2012 in different geographical regions
and vertical layers of the UTLS. The largest number of observations is
available over the North Atlantic. Here, a neutral temperature trend is found
within the lowermost stratosphere. This contradicts the temperature trend in
the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA-Interim
reanalysis, in which a significant (95 % confidence) temperature increase
of +0.56 Kdecade-1 is found. Differences between trends
derived from observations and reanalysis data can be traced back to changes
in the temperature difference between observation and model data over the
period studied. This study underpins the value of the IAGOS temperature
observations as an anchor point for the evaluation of reanalyses and its
suitability for independent trend analyses.
Introduction
Temperature changes in the lower stratosphere (∼50hPa) obtained from radiosondes and satellite retrievals show
cooling of about 0.5 Kdecade-1 over much of the globe during the
period from 1979 to 1995. Since 1995, the cooling turned into a neutral trend
with a larger increase (not significant) over the Antarctic region than over
the tropics (Randel et al., 2009; Blunden and Arndt, 2014; Seidel et al.,
2016). The robustness of the temperature trends in the lower
stratosphere derived from radiosondes
(since 1958) and from satellites (since 1979) suffers from
instrumental uncertainties such as sensor changes, drifts, etc., implying
large uncertainties in the trend estimates (Simmons et al., 2014). In recent
years, several studies assessed the uncertainty of temperature trends in the
lower stratosphere and the impact on changing trends of radiatively active
constituents (such as ozone) or atmospheric dynamics (e.g., Fueglistaler
et al., 2014; Seidel et al., 2016).
Temperature trends in the upper troposphere–lowermost stratosphere (UTLS)
are even more uncertain due to insufficient regional coverage of in situ
observations. Most studies of UTLS temperature trends are based on the global
radiosonde network, but most of these data are from the Northern Hemisphere
midlatitudes (70 % of radiosonde launches occurred between 30 and
60∘ N). Furthermore, these observations suffer from time-varying
biases, which cannot capture the large variability in the UTLS and
inhomogeneity due to changes in instrumentation (Bencherif et al., 2006;
Seidel and Randel, 2006; Xu and Powell, 2010). Satellite observations cover
the spatial scale but are limited by their coarse vertical resolution,
especially in the UTLS region. A well-suited data source for temperature
profiles is the relatively new Global Positioning System radio
occultation (GPSRO) technique
(Kursinski et al., 1997; Wickert et al., 2001). The reliability of these
measurements within the UTLS region has been demonstrated with trend analyses
of GPSRO temperature (Steiner et al., 2009) or GPSRO-derived trends in the
thermal tropopause temperature and tropopause height (Schmidt et al., 2010;
Rieckh et al., 2014; Khandu et al., 2016). Ho et al. (2017) demonstrated the
usefulness of the GPSRO measurements to correct the temperature bias of
radiosondes with different sensor types in the lowermost stratosphere (LMS),
which is an important task to reduce the temperature uncertainties.
Comparison of tropospheric temperature trends derived from homogenized
satellite data sets and model simulations find more consistency but require
long-term time series (> 17 years) before a robust trend arises from
internal climate variability (Santer et al., 2011).
Since the radiative forcing from greenhouse gases, including water vapor, is
sensitive to changes in the mid-troposphere and the UTLS (Solomon et al.,
2010; Riese et al., 2012), this region is extremely important for climate
change and for controlling dynamical processes governing
stratosphere–troposphere exchange (Gettelman et al., 2011). Furthermore, the
variability and changes in the temperature in the UTLS play an important role
in regulating the exchange of water vapor, ozone, and other trace gases
between the troposphere and the stratosphere.
Continuous in situ observations of these properties in the UTLS region can
only be conducted with satellite and aircraft measurements over a large
spatial region. Automated aircraft temperature observations are collected,
along with an increasing number of humidity data through the World
Meteorological Organization (WMO) Aircraft Meteorological DAta Relay (AMDAR)
program (WMO, 2014; Petersen, 2016). Petersen (2016) showed that at flight
level the errors of temperature and wind in the 3–48 h forecast were
reduced by nearly 50 % when assimilating data from passenger aircraft.
However, Ballish and Kumar (2008) and Drüe et al. (2008) identified that
the AMDAR aircraft temperature is strongly affected by a warm bias, which can
fluctuate by altitude, aircraft type, and phase of flight, while the reason
for this bias is not fully understood (Ingleby et al., 2016).
Previous studies already used the passenger aircraft temperature measurements
from MOZAIC (Measurement of ozone and water vapor by Airbus in-service
aircraft Marenco et al., 1998) for intercomparison with GPSRO
measurements and ECMWF (European Centre for Medium-Range Weather Forecasts)
analyses. Bortz et al. (2006) analyzed MOZAIC temperature measurements at
cruise altitude from 1994 to 2003 within the tropical region. They found no
significant temperature increase within the upper tropical troposphere. The
authors concluded that the temperature measurements are representative enough
to be used in intercomparison studies with satellite (Microwave Sounding
Unit, MSU) and radiosonde measurements within this region, but at this time
the data record was too short for trend estimates. Heise et al. (2008)
compared around 2700 MOZAIC in situ temperature profiles with profiles from
GPSRO and analyses from ECMWF between 2001 and 2006. They concluded that
MOZAIC in situ temperature had no bias against the ECMWF temperature above
300 hPa, whereas GPSRO showed a cold bias of -0.9K
compared to the MOZAIC temperature. Since 2001, GPSRO data have been
assimilated in numerical weather prediction models and reanalysis products.
Schmidt et al. (2010) characterized the tropopause inversion layer in the
Northern Hemisphere with temperature profiles from in situ measurements and
from the model for the period from 2001 to 2009 and concluded that the cold
point at the tropopause agreed well.
In this study, IAGOS (In-service Aircraft for a Global Observing System)
temperature observations, which are available for almost 2 decades since
1994, are analyzed. The geographical coverage of the measurements is shown in
Sect. 2, in which the reliability of the IAGOS observations and the data
selection is discussed. The UTLS temperature distribution and derived
temperature trends are presented in Sect. 3, and their robustness and
suitability for the evaluation of global-scale reanalyses with the example of
ERA-Interim (ERA-I) are discussed in Sect. 4. Due to varying geographical
coverage of these data, they cannot provide a full global assessment yet, but
as we will show in our study, they can serve as an anchor point for trend
analyses and evaluation of reanalysis at least over the extratropics. This
conclusion is included in Sect. 5.
Data selection and methodsThe IAGOS European Research Infrastructure
Since 1994, IAGOS in situ observations of essential climate variables
(temperature, water vapor, and ozone) in the UTLS are provided on a global
scale by the European Research Infrastructure IAGOS (Petzold et al., 2015).
IAGOS builds on the former EU framework projects MOZAIC and CARIBIC (Civil
Aircraft for the Regular Investigation of the atmosphere Based on an
Instrument Container Brenninkmeijer et al., 2007).
(a) Schematic of the temperature sensor attached to the
humidity sensor mounted in the Rosemount housing (Helten et al., 1998).
(b) Packages 1 and 2 installed aboard the Airbus A340-300, and the
inlet plate including the Rosemount housing (photograph courtesy of
Lufthansa).
Currently (2017), up to 10 passenger aircraft from various international
airlines are equipped with scientific instruments to monitor the
meteorological state (temperature, water vapor, and wind) and atmospheric
chemical composition. In addition to the measurements of temperature and
(relative) humidity, the IAGOS-CORE Package 1 includes instruments to measure
ozone and carbon monoxide. Ozone is measured by UV absorption (Thermo
Scientific for Model 49i Ozone Analyzer), and CO by infrared absorption (Thermo Scientific for Model 48CTL CO Analyzer). Both instruments are
regularly calibrated before and after deployment and the overall uncertainty
for ozone is 2 ppb± 2 % and 5 ppb± 5 %
for CO (Thouret et al., 1998; Nédélec et al., 2015). Additionally,
positions, pressure, ambient temperature (measured from the aircraft),
aircraft speed, wind speed, and wind direction are provided by the A330 and
A340 avionic systems (details are given by Petzold et al., 2015). For
completeness, several additional parameters (NOy, NO, NO2)
are measured or will soon be measured (CH4, CO2) by IAGOS
Package 2a–d, which is described by Petzold
et al. (2015).
(a) Temperature correlation for all measurements below the
pressure altitude 400 hPa and (b) temperature bias of the
IAGOS temperature instrument (TICH) compared to a regularly
quality-checked temperature sensor (TAC) from the research
aircraft for seven flights during the AIRTOSS-ICE campaign in 2013. The grey
area marks the total range of the total uncertainty for the IAGOS temperature
observations. The temperature bias was separated for measurements between 300
and 400 hPa (orange) and 200 and 300 hPa (blue) to highlight
the shift to a smaller bias at lower pressure.
Relative frequency of the IAGOS temperature observations at cruise
altitude (p<350hPa) in different regions (Table 1). The total
number of 1 min averages is 14.8 million. Note, 1 % includes 0.2 million
1 min mean data points within the UTLS region.
Temperature measurements and evaluations
On IAGOS aircraft, temperature is measured in situ with a compact airborne
sensing device AD-FS2 (Aerodata, Braunschweig, Germany), which is installed
in an appropriate aeronautic housing (Helten et al., 1998). From 1994 to 2009
a platinum resistance sensor (Pt100) was attached near the humidity sensing
device. After a design change from MOZAIC to IAGOS, this sensor is now
directly at the humidity sensor (Fig. 1). Intercomparison between both
systems showed a temperature deviation of less than 0.1 K in the
calibration chamber, which is below the sensor uncertainty. The temperature
is measured with an uncertainty of ±0.25 K by
a microprocessor-controlled transmitter unit (model HMT333, Vaisala, Finland),
which passes the signal to the data acquisition system of the IAGOS Package 1
instrument, in which it is recorded with a time resolution of 4 s
(Nédélec et al., 2015; Petzold et al., 2015). Pre- and
post-deployment calibrations from the laboratory are used to evaluate the
temperature signal and to ensure the quality of the temperature measurement
(Helten et al., 1998; Neis et al., 2015). Typical deployment phases are in
the range of 2–3 months. Accounting for corrections of adiabatic compression
at the inlet part of the housing (Stickney et al., 1994; Moninger et al.,
2003), the overall uncertainty of the ambient air temperature is
±0.5 K. More details are given in the standard operating
procedure (SOP) of the IAGOS Capacitive Hygrometer (ICH), available at
http://www.iagos.org. For the purpose of this analysis the data set was
reduced to 1 min averages.
To ensure the reliability of the IAGOS temperature observations, we make use
of temperature measurements from the AIRTOSS-ICE aircraft campaign (Aircraft
Towed Sensor Shuttle – Inhomogeneous Cirrus Experiment), which focused on
midlatitude cirrus clouds (Neis et al., 2015). During this campaign, the
IAGOS temperature instrument was installed on the research aircraft (Learjet
35A) and provides the opportunity to compare both temperature measurements.
The aircraft temperature measurement was made with a Pt100 thermistor mounted
in the same type of Rosemount as for the IAGOS temperature measurements. The
temperature sensor of the research aircraft was calibrated regularly with an
uncertainty of about 0.5–1.0 K. Figure 2 shows the temperature
correlation and the temperature bias (ΔT) at pressures below
400 hPa during seven flights. The general behavior between both
temperature measurements agreed well along the flight tracks. The mean
deviation is ΔT=-0.3 K. The temperature deviation is pressure dependent and decreases towards 200 hPa. The temperature correlation is high and the
temperature bias is smaller than the overall uncertainty of 0.5 K
(sensor and adiabatic compression correction), which demonstrates the
capability of the IAGOS temperature sensor to measure the ambient temperature
at cruise altitude very precisely.
The temperature measured from the aircraft (TAC) is based on
total air temperature (TAT) designed for subsonic aircraft (Goodrich
Corporation, formerly Rosemount Aerospace). The total air temperature is
defined as the ambient air temperature plus the temperature increase due to
adiabatic compression in the Rosemount housing. Typically three TAT sensors
(platinum resistance sensor) are installed at the nose region of the
aircraft, but in general only one is used for the pilots and stored for
IAGOS. The other two sensors are used to monitor the differences between all
TAT sensors. In general, the airlines follow the AMDAR quality
recommendations (WMO, 2003), and the TAT sensors are regularly checked by
visible inspection. An exchange of one TAT sensor is performed if it differs
more than 3 K from the other two TAT sensors.
Definition of different regions covered by IAGOS flight tracks.
Data density at cruise altitude
(p<350hPa) for all defined regions on a logarithmic scale (color)
for the period analyzed.
Spatial and temporal data coverage
In this study all temperature measurements from the IAGOS-CORE flights at
cruise altitude (p<350hPa) from January 1995 to December 2012 are
used. More recent measurements are not yet validated and therefore not
included in this study. Therefore, most of the IAGOS temperature observations
rely on the former instrument design. Following previous IAGOS or MOZAIC
analyses (Thouret et al., 2006; Dyroff et al., 2015; Thomas et al., 2015;
Stratmann et al., 2016), we divide the data into 14 geographical regions
(Fig. 3 and Table 1). Seasonal and regional differences in the temperature
behavior can then be linked to the different dynamical patterns. The largest
number of measurements is obtained in the midlatitudes (North America, North
Atlantic, Europe, and Central Asia), whereas the number of measurements in
the higher latitudes (northern
Canada, Greenland, Scandinavia, northern Asia) and the tropical regions
(Middle America, tropical Atlantic, North Africa, tropical Asia, South
America, southern Africa) is much smaller and does not provide continuous coverage
over the period presented. The data coverage for each region over the
analyzed period is shown in Fig. 4.
Changes in radiatively active chemical constituents (e.g., ozone and water
vapor) within the LMS and the upper troposphere (UT) have different impacts
on the ambient temperature in these layers. In general, the tropopause layer
(TPL) separates the stable stratified LMS and the unstable UT. In the present
study, the pressure of the thermal tropopause (pTPHWMO) is
derived from ERA-I (see below), and it is used to determine the
position of the aircraft relative to this layer and to distinguish if the
aircraft flew within the UT, the TPL, or the LMS. This is achieved using
the following criteria:
LMS:p<pTPHWMO-15hPa,which is limited by themaximum cruise altitude(p∼190hPa);TPL:p=pTPHWMO±15hPa;UT:p>pTPHWMO+15hPa,limited to350hPa.
A comparable definition has been used by Thouret et al. (2006), in which the
pressure of the 2 PVU (potential vorticity unit) surface was used to define
the dynamical tropopause layer with a vertical depth of ±15hPa.
The thermal tropopause is valid within all latitude bands, whereas the
dynamical tropopause cannot be used in tropical regions because the Coriolis
parameter is zero. Therefore, potential vorticity (PV) goes to zero and the
2 PVU surface is not defined
(Boothe and Homeyer, 2017).
Cumulative distribution of (a) ozone, (b) PV from
ERA-I, and (c) CO for the different vertical layers (LMS, dark grey; TPL,
grey; UT, light grey) over the North Atlantic region from January 1995 to
December 2012. The dashed lines mark the values of each species when the
cumulative distribution reached 50 %.
Temperature time series of IAGOS observations (black) and ERA-I
(orange) for the (a) lowermost stratosphere, (b) tropopause
layer, and (c) upper troposphere over the North Atlantic region. The
grey lines show the SD (1σ) of the mean using the IAGOS observations
for each month.
Within the IAGOS project, measurements of ozone (since 1994) and carbon
monoxide (since 2002) are available and are used to justify our layer
classification scheme. Figure 5 shows the cumulative distribution of ozone,
carbon monoxide, and PV. All distributions demonstrate a clear separation
between the three layers based on the thermal tropopause. The median ozone
mixing ratio in the UT is 60 ppb, median CO mixing ratio is
90 ppb, and median PV is 0.5 PVU. Within the LMS, the median values
for O3 (310 ppb), for CO (40 ppb), and for PV
(7 PVU) are consistent with previous studies (Pan et al., 2004; Kunz
et al., 2008; Brioude et al., 2009; Schmidt et al., 2010).
Meteorological reanalysis
ERA-I covers the period from 1979 until present, assimilating
observational data from various satellites, radiosondes, buoys, commercial
aircraft, and others (Dee et al., 2011; Simmons et al., 2014). Note that the
IAGOS temperature observations are not assimilated in any numerical weather
prediction model or reanalysis product, which makes it unique for model
evaluation. For this study, the 6-hourly outputs from ERA-I (0.75∘×0.75∘) were interpolated onto a 1∘×1∘ horizontal grid and on 60 vertical levels of constant pressure
and potential temperature (Kunz et al., 2014). Additionally, the variables of
the PV, and the pressure of the thermal tropopause (pTPHWMO)
based on the WMO criteria were calculated (WMO, 1957; Reichler et al., 2003).
The ERA-I data were linearly interpolated (longitude, latitude, pressure,
time) onto each flight track with 4 s resolution as described by Kunz
et al. (2014). As for IAGOS temperature observations, the ERA-I data set was
reduced to 1 min averages.
ResultsIAGOS temperature measurements over different regions
Approximately 69 % of the IAGOS temperature measurements in the UTLS were
obtained in the midlatitude band between 30 and 60∘ N. Therefore, we
show our detailed analysis within all three layers of the UTLS only over the
North Atlantic region between January 1995 and December 2012 and provide
additional material in the Supplement. Figure 6 shows monthly median
temperature of the IAGOS observations in the LMS, TPL, and UT for the period
from 1995 to 2012 over the North Atlantic region. All layers show a seasonal
temperature variation of 5–10 K in each year. The warmest
temperatures are observed within the LMS and the coldest (as expected) within
the TPL. Between 2006 and 2008, the wintertime temperature minima at the TPL
and the UT are almost 3–4 K warmer compared to the other years.
These enhanced temperatures can also be observed over Europe and Central Asia
within these layers. All other regions in the extratropics show time series
in amplitude and seasonal variations mostly comparable to the North Atlantic
region, if enough measurements were available for each vertical layer.
Throughout all regions in the extratropics, the summer (winter) months are
always the warmest (coldest) within all three layers (see Tables S1–S3 in
the Supplement). Over the tropical regions, no measurements are available in
the LMS and TPL because the TPL is mostly above cruise altitude. In the UT,
the temperature is mostly constant throughout the year for each tropical
region.
Anomalies in monthly averaged temperature in the (a) LMS,
(b) TPL, and (c) UT over the North Atlantic region from
IAGOS observations (black), ERA-I (orange), and aircraft measurements (AC,
light blue). The aircraft measurements are assumed to be comparable to AMDAR
data. The anomalies are smoothed with a 12-month running mean to highlight
the behavior of the time series, which reflects the temperature trends by
linear regression analyses (dashed lines).
Temperature anomalies and trends
The temperature trends are derived from using a robust regression analysis
over the full period of de-seasonalized temperature observations. In all
regions, 18-year monthly averages (e.g., mean of all January values, mean of
all February values, etc.) were subtracted from the time series of each
layer. The temperature trends exhibit no significant non-linear
contributions. We ensured that the data are homoscedastic, and they are
nearly normally distributed (median and mean values are very close; see
Tables S1, S2, and S3). Additionally, we checked that the data are not
auto-correlated. Temperature trends are reported only if at least 90 % of
all months have at least 200 data points each (Fig. 4). The
Mann–Kendall test is used to identify the trend significance (Mann, 1945;
Kendall, 1975; Gilbert, 1987). The 90 % threshold was chosen because the
Mann–Kendall significance of the trend analysis did not change when 10 %
of the data were randomly excluded from the trend calculation. The robustness
of temperature trends was tested by skipping the first or last year of the
18-year period (Table S4). Within all layers and all regions, each trend kept
the same sign and the trend values varied within the standard error. The only
exception was the UT over North America where the temperature trend changed
from slight positive trend (18 years) to neutral when the final year was
removed.
Temperature trends over the analyzed period for each region obtained
from the IAGOS observations and ERA-I for the (a) LMS, (b)
TPL, and (c) UT, for which at least 90 % of all months were
available (see text for details).
Figure 7 shows the time series of temperature anomalies and linear trend
lines over the North Atlantic region using the IAGOS observations. The
monthly temperature anomalies in each layer vary by ±3 K
(strongest within the UT). Temperature trends are
+0.22(±0.20)Kdecade-1 in the UT,
+0.25(±0.16)Kdecade-1 in the TPL, and
-0.05(±0.17)Kdecade-1 in the LMS. None of the trends are
significant at the 95 % level. In the UT, the amplitudes of the smoothed
time series are larger and decrease towards the upper levels. They show two
warmer phases and two colder phases during the period analyzed.
Figure 8 and Table 3 summarize the temperature trends for different regions
in the different layers. Within the LMS, significant cooling is observed over
Greenland (-1.39(±0.29)Kdecade-1) and North America
(-0.71(±0.21)Kdecade-1). The smaller trend over Europe
(-0.53(±0.20)Kdecade-1) is not significant. The other
regions do not have enough data for a meaningful trend analysis in the LMS.
Within the TPL, there are only three regions with sufficient data coverage,
and no significant trend is detected, although there is a weak indication for
a small warming over North America and the North Atlantic and for a small
cooling over Europe. In the UT, significant cooling is found over North
America (-1.08(±0.18)Kdecade-1), while temperatures over
Europe (-0.59(±0.1)Kdecade-1) insignificantly decrease.
Over the North Atlantic (+0.22(±0.20)Kdecade-1) and over
Central Asia (+0.32(±0.33)Kdecade-1), temperatures
increase insignificantly. Over
tropical Asia (-0.54(±0.04)Kdecade-1) temperatures show
a tendency to decrease, but without significance at the 95 % level. This
result is puzzling because it is assumed that the temperature increases in
the tropics in the UT, equivalent to the surface temperature (Khandu et al.,
2016). One reason could be related to the tropopause definition, but
O3 (40–80 ppb) and PV (always below 1 PVU) indicate
that the selected air mass is tropospheric. Another reason could be related
to a higher variability in the temperature due to large-scale influence or
simply that the data coverage is still too poor in this region, which might
lead to a higher variability of the local temperatures which then mask the
temperature trend.
Temperature correlation from the observations and ERA-I over the
North Atlantic region. The distribution is averaged in bin widths of
0.5 K and the color shows the number frequency with each bin on
a logarithmic scale. The numbers included show the number of data points, the
absolute bias, the corresponding SD, and the linear robust regression fit
(black line). The 1:1 line is shown in red, and its variation with
2 K is shown as dashed red lines. The statistical values for all
other regions are summarized in Table 3.
DiscussionIntercomparison with ERA-I reanalysis
ERA-I reanalysis is widely used for intercomparisons and is currently
the latest global atmospheric reanalysis product provided by ECMWF (Dee
et al., 2011; Fujiwara et al., 2017). Simmons et al. (2014) indicated that
there is a problem with the temperature near the tropopause in the tropics
and extratropics. Here, we demonstrate the use of IAGOS observations to
evaluate the ERA-I reanalysis within the UTLS region.
Figure 9 shows the intercomparison between the temperature of the IAGOS
observations and the ERA-I reanalysis of all 1 min mean data over the North
Atlantic region at cruise altitude (below 350 hPa). The good
agreement between both temperatures is reflected by the high correlation
coefficient (R2=0.97) and the slope of 0.94 from the regression fit. The
bias between the temperature of the IAGOS measurements and ERA-I is
-0.02 K over the entire period. The absolute bias is
0.81(±0.72)K, and 93 % of the data are close to the 1:1
line (±2K). The spread of the correlation corresponds to
a larger variability in the IAGOS temperature measurements when the aircraft
flew through clouds, or from the interpolation of the gridded ERA-I
temperature to the aircraft position. IAGOS measurements with a backscatter
cloud probe have been available since 2011 and these measurements will be
used in the future to distinguish between clear-sky and in-cloud measurements
to further reduce the uncertainties. The good agreement between the 1 min
mean temperature data from ERA-I and the IAGOS observation in the other
regions is documented in Table 2, in which R2 is always larger than
0.97, the slope of each regression fit is between 0.94 and 0.98, and the
absolute bias varies between 0.60 and 0.98 K, with a SD of
0.56–0.79 K.
Statistical parameters summarized for all regions studied (see
Fig. 9). The slopes and intercepts are determined from the linear robust
regression fit between ERA-I and IAGOS temperature. Additionally, for each
region the absolute bias and its SD are given.
RegionNumber of dataR2SlopeInterceptAbsolute biasSDpoints ×[106][K][K]Higher latitudesNorthern Canada0.230.980.969.680.730.64Greenland0.500.970.969.830.810.70Scandinavia0.160.970.969.240.790.68Northern Asia0.610.970.9510.340.920.77MidlatitudesNorth America1.420.980.9510.300.790.69North Atlantic3.230.970.9412.260.810.72Europe2.800.970.9510.270.850.72Central Asia1.760.980.976.350.920.79Northern tropicsMiddle America0.030.990.976.640.810.57Tropical Atlantic0.390.970.977.290.770.63North Africa0.710.980.976.530.660.61Tropical Asia0.500.980.9610.490.980.79Southern tropicsSouth America0.140.980.976.620.870.65Southern Africa0.410.990.984.740.600.56Anomalies and trends using ERA-I
The ERA-I temperature time series and temperature anomalies agree with the
IAGOS measurements mostly in phase and amplitude within all vertical layers
over the North Atlantic region (Figs. 6 and 7). Larger deviations can be
found at the beginning of the time series in the mid-1990s, when the IAGOS
temperatures are warmer than ERA-I, and after 2008, when the opposite occurs.
The temperature trends from ERA-I show a slight warming over the North
Atlantic region (+0.38(±0.18)Kdecade-1) in the UT and
0.46(±0.15)Kdecade-1 in the TPL. Neither the trends in
the UT nor in the TPL
are significant.
Temperature trends in ERA-I and from the IAGOS observations within
the LMS, TPL, and UT, for which at least 90 % of measurements were available over the
entire period (1995–2012). SE is the standard error of the
temperature trend. The temperature trend is significant (Sig = 1) if the
p value (consistency) is smaller than 0.06 (> 94 %), which was
derived from the Mann–Kendall test.
RegionERA-I IAGOS ΔT18yrSESig.p valueΔT18yrSESig.p value(Kdecade-1) (Kdecade-1) LMS Greenland-0.790.2900.12-1.390.2910.01North America-0.250.2100.46-0.710.2110.02North Atlantic+0.560.1710.05-0.050.1700.98Europe+0.110.1900.83-0.530.2000.17TPL North America+0.290.1900.19+0.230.2000.28North Atlantic+0.460.1500.16+0.250.1600.33Europe+0.200.1500.55-0.440.1700.20UT North America-0.920.1710.04-1.080.1810.01North Atlantic+0.380.1800.20+0.220.2000.52Europe-0.240.1400.73-0.590.1500.26Central Asia+0.660.3300.53+0.320.3300.85Tropical Asia-0.580.3900.72-0.540.0400.63
In the LMS, ERA-I temperatures show a significant increase of
+0.56(±0.17)Kdecade-1 over the 18 years of the study
period at the 95 % significance level, which is not the case for the
IAGOS observations, in which the temperature shows
a insignificant decrease of
-0.05(±0.17)Kdecade-1 over the period analyzed (see
Sect. 4.1 above). We calculated the temperature trends in the lower (5th
percentile, colder temperatures) and upper (95th percentile, warmer
temperatures) ranges of the data to determine the robustness of these trends
within the LMS. For the 5th percentile, the temperature trend from ERA-I
remains significant (99 % confidence) at +0.48 Kdecade-1,
while the trend from the IAGOS observations still does not significantly
decrease at -0.22Kdecade-1 (87 % confidence). In the
upper range, the differences between the temperature trends diminish to
+0.12Kdecade-1 (53 % confidence) using ERA-I and
-0.07Kdecade-1 (73 % confidence) using the IAGOS
observations. This shows that within the colder temperature regimes, the
trend difference is significant, while for the warmest temperatures regimes,
the variability is too large to obtain a significant temperature trend.
Figure 8 and Table 3 also contain the temperature trend estimates for
ERA-I. In contrast to the IAGOS observations, a cooling tendency in the
LMS can only be found over Greenland
(-0.79(±0.29)Kdecade-1), and over North America
(-0.25(±0.21)Kdecade-1), but both are not significant.
The strong cooling trend over Greenland results from elevated temperatures in
the late 1990s. Over Europe, where IAGOS observations showed an insignificant
decrease, ERA-I exhibits a very small (and not significant) warming of
+0.11(±0.19)Kdecade-1.
Within the TPL, the temperature trends are mostly comparable to the IAGOS
observations (except over Europe), but all trends are not significant. The
best consistency between the temperature trends from IAGOS observations and
ERA-I is seen for the UT layer, except over Europe, where ERA-I shows
slightly less cooling than IAGOS observations, and Central Asia, where ERA-I
has a stronger warming trend.
Possible sources of the differing temperature trends
There is a remarkable coherence of the temperature time series and the
anomalies of the observations and ERA-I in all regions and layers, but the
linear temperature trends reveal a deviation between the two data sets,
especially in the LMS. In order to investigate this in more detail, Fig. 10
shows the annual deviation between the IAGOS observations and ERA-I at the
LMS over eight regions. From 1995 to 2002 the observations are always warmer
than ERA-I within the Northern Hemisphere. The sign changes to neutral until
2008, and since 2009 the IAGOS observations have been colder than ERA-I. This
temporally varying temperature bias is one reason why the temperature trends
are not equal between the two data sets. The differences cannot be explained
by the chosen definition to determine the different layers within the UTLS
region. For example, ozone observations show seasonal variations with values
between 350 and 640 ppb, which are robust values for stratospheric
air masses with a typical seasonality (Brioude et al., 2009). Additionally we
use CO observations (from 2002 to 2008) and PV with similar results.
Annual mean of the monthly mean difference between the observations
and ERA-I for different regions in the Northern Hemisphere in the lowermost
stratosphere from 1995 to 2014. The red colors show that ERA-I temperature is warmer than IAGOS temperature and in blue colors vice versa. The dashed lines show
clear break points within the time series. The cross marks a year when the
annual mean could not be calculated.
Furthermore, we demonstrated that temperature measurements from IAGOS agree
well with quality-controlled measurements aboard research aircraft (Fig. 2).
Therefore, we expect that the trends from IAGOS are robust, which leads to
the hypothesis that ERA-I exhibits temperature biases that vary with time.
Some of the observed deviations can be explained by changes in the data
assimilation sources in ERA-I, which employed several new satellite products
at various times after the year 2000. Simmons et al. (2014) showed that the
source of input data (e.g., inconsistent sea surface
temperature bias) changed in ERA-I after 2002. Since 2001, temperature
profiles have been assimilated from GPSRO measurements into ERA-I, which led
to a cold bias in ERA-I, where the largest effects appeared after 2006 in
ERA-I, when the number of assimilated data increased. This could lead to
a warming of the lower stratosphere and a cooling at around 200 hPa
(Poli et al., 2008, 2010).
Another source of the deviation could be expected from the increasing number
of temperature measurements from the AMDAR system onboard passenger
aircraft. Ballish and Kumar (2008) showed that the temperatures from
passenger aircraft are warm-biased at cruise altitude (200–300 hPa).
This has not been accounted for in ERA-I. Furthermore, the WMO reported that the
number of aircraft reports increased from 100 000 in early 2000 to more than
350 000 after 2012 and became the third most important data source of
assimilation for short-term forecast in numerical weather prediction (WMO,
2014; Petersen, 2016).
In order to test the second hypothesis, Fig. 7 also includes the temperature
trend derived over the North Atlantic region using the aircraft temperature
measurements (TAC), which we assume to be comparable to AMDAR
measurements. The temperature trend in the LMS is also positive, and it is in
between the neutral temperature trend using the IAGOS temperature
observations and the positive temperature trend using ERA-I. This gives an
indication that the assimilated aircraft measurements could indeed be a cause
for bias in the ERA-I data. It is therefore planned to introduce a bias
correction for AMDAR aircraft observations in the next ERA reanalysis (ERA5,
D. Dee, personal communication, 2016).
Conclusion
In this study, nearly
2 decades of in situ IAGOS
temperature measurements by the European Research Infrastructure IAGOS using
passenger aircraft were presented and used to determine regional trends in
the UTLS. The quality of the temperature measurements is regularly evaluated
in the laboratory through pre- and post-deployment calibration, and it has
been assessed by intercomparison with temperature observations from research
aircraft.
UTLS temperature time series and trends are analyzed for 14 different regions
within the northern latitudes, midlatitudes, and tropics and are separated
into the lowermost stratosphere, the tropopause layer, and the upper
troposphere using the thermal tropopause from ERA-I as a reference layer.
The interannual variability within all regions and layers is mostly
consistent between the IAGOS observations and ERA-I during the past
18 years (1995–2012), which is not the case for the temperature trends.
For the temperature trend, regions are only considered when at least 90 %
of all months are available. Over the North Atlantic, where the largest
number of measurements are available, we found a significant (95 %
confidence level) positive trend (+0.56Kdecade-1) in ERA-I
and a small negative trend (-0.05Kdecade-1) in the
observations (not significant) in the LMS. A significant (95 %
confidence) negative temperature trend over Greenland and North America in
the LMS for the IAGOS temperature is found. Over Europe both temperature
trends are not significant, but the temperature trend from the IAGOS
observations is negative and for positive ERA-I. Within the tropopause layer
we found mostly comparable trends, except over Europe, where the sign of
trends is different, but all trends are not significant. The calculated
temperature trends in the UT show all the same signs for each region, whereas
only a significant trend is found over North America, with a large cooling
rate of about -0.92Kdecade-1 (ERA-I) and
-1.08 Kdecade-1 (IAGOS) in the last 18 years.
The large deviation between the different LMS temperature trends from IAGOS
observations and ERA-I data is mostly related to the temporally varying bias
between both temperature time series. As we have no reason to assume an
evolving bias in the IAGOS observations, we conclude that ERA-I temperatures
are too cold between 1995 and 2001 and too warm after 2007. These dates
roughly correspond with changes in the data streams that were used in the
ERA-I data assimilation. The evolution of ERA-I temperature can be explained
by additional assimilation of GPSRO data and data from the AMDAR passenger
aircraft network, which have been shown to be warm-biased. These data sources
play an increasing role in the ERA-I data assimilation after 2006. Our
recommendation is therefore to include a bias correction for these
temperature measurements in future versions of the European reanalysis and
use IAGOS observations as an anchor point.
The IAGOS temperature measurements highlight the need of independent global
measurements with high and long-term accuracy to quantify long-term changes,
especially in the UTLS region, and to help identify inconsistencies between
different data sets of observations and models. Due to the expansion of the
IAGOS aircraft fleet with airlines from other continents, we are looking
forward to investigating temperature changes in the UTLS over several more
regions in a few years from now.
The measurements are available free of charge on
www.iagos.org.
The Supplement related to this article is available online at https://doi.org/10.5194/acp-17-12495-2017-supplement.
FB conducted the analysis of the measurements and wrote the
paper. HGJS, PN, SR, and UB were in charge of the instrument setup,
calibration, and processing of the measurements. AP, MGS, and AW helped with
the interpretation of the data and in the paper-writing process. PN, DB,
and VT were in charge of providing the CO and ozone measurements of MOZAIC and
IAGOS.
The authors declare that they have no conflict of
interests.
Acknowledgements
We gratefully acknowledge all partners for their continuous support for more
than 20 years: Lufthansa, Air France, China Airlines, Cathay Pacific,
Iberia, CNRS, University of Manchester, Meteo-France, Sabena Technics, and
Enviscope GmbH, including all former coordinators of MOZAIC and IAGOS: Alain
Marenco, Andreas Volz-Thomas and Jean-Pierre Cammas. We acknowledge ECMWF for
providing the ERA-Interim data and Nicole Thomas for excellent programming
support. Part of this project was funded by BMBF in IAGOS-D contract
01LK1301A. The IAGOS database is supported by AERIS (CNES and
INSU-CNRS).
The article processing charges for this open-access
publication were covered by a Research Centre of the
Helmholtz Association.
Edited by: Farahnaz Khosrawi
Reviewed by: Heini Wernli and one anonymous referee
ReferencesBallish, B. A. and Kumar, V. K.: Systematic differences in aircraft and radiosonde temperatures, B. Am. Meteorol. Soc., 89,
1689–1707, 10.1175/2008BAMS2332.1, 2008.Bencherif, H., Diab, R. D., Portafaix, T., Morel, B., Keckhut, P., and Moorgawa, A.: Temperature climatology and trend
estimates in the UTLS region as observed over a southern subtropical site, Durban, South Africa, Atmos. Chem. Phys., 6, 5121–5128,
10.5194/acp-6-5121-2006, 2006.Blunden, J. and Arndt, D. S. (Eds.): State of the Climate in 2013, B. Am.
Meteorol. Soc., 95, S1–S279, 10.1175/2014BAMSStateoftheClimate.1,
2014.Boothe, A. C. and Homeyer, C. R.: Global large-scale stratosphere–troposphere exchange in modern reanalyses,
Atmos. Chem. Phys., 17, 5537–5559, 10.5194/acp-17-5537-2017, 2017.Bortz, S. E., Prather, M. J., Cammas, J.-P., Thouret, V., and Smit, H.: Ozone, water vapor, and temperature in the upper
tropical troposphere: variations over a decade of MOZAIC measurements, J. Geophys. Res., 111, D05305, 10.1029/2005JD006512,
2006.Brenninkmeijer, C. A. M., Crutzen, P., Boumard, F., Dauer, T., Dix, B.,
Ebinghaus, R., Filippi, D., Fischer, H., Franke, H., Frieß, U.,
Heintzenberg, J., Helleis, F., Hermann, M., Kock, H. H., Koeppel, C.,
Lelieveld, J., Leuenberger, M., Martinsson, B. G., Miemczyk, S., Moret, H.
P., Nguyen, H. N., Nyfeler, P., Oram, D., O'Sullivan, D., Penkett, S., Platt,
U., Pupek, M., Ramonet, M., Randa, B., Reichelt, M., Rhee, T. S., Rohwer, J.,
Rosenfeld, K., Scharffe, D., Schlager, H., Schumann, U., Slemr, F., Sprung,
D., Stock, P., Thaler, R., Valentino, F., van Velthoven, P., Waibel, A.,
Wandel, A., Waschitschek, K., Wiedensohler, A., Xueref-Remy, I., Zahn, A.,
Zech, U., and Ziereis, H.: Civil Aircraft for the regular investigation of
the atmosphere based on an instrumented container: The new CARIBIC system,
Atmos. Chem. Phys., 7, 4953–4976, 10.5194/acp-7-4953-2007, 2007.Brioude, J., Cammas, J. P., Cooper, O. R., and Nedelec, P.: Characterization of the composition, structure, and seasonal
variation of the mixing layer above the extratropical tropopause as revealed by MOZAIC measurements, J. Geophys. Res.-Atmos., 114,
1–17, 10.1029/2007JD009184, 2009.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, 10.1002/qj.828, 2011.Drüe, C., Frey, W., Hoff, A., and Hauf, T.: Aircraft type-specific errors in AMDAR weather reports from commercial
aircraft, Q. J. Roy. Meteor. Soc., 134, 229–239, 10.1002/qj.205, 2008.Dyroff, C., Zahn, A., Christner, E., Forbes, R., Tompkins, A. M., and van Velthoven, P. F. J.: Comparison of ECMWF
analysis and forecast humidity data with CARIBIC upper troposphere and lower stratosphere observations, Q. J. Roy. Meteor. Soc., 141,
833–844, 10.1002/qj.2400, 2015.Fueglistaler, S., Abalos, M., Flannaghan, T. J., Lin, P., and Randel, W. J.: Variability and trends in dynamical forcing
of tropical lower stratospheric temperatures, Atmos. Chem. Phys., 14, 13439–13453, 10.5194/acp-14-13439-2014, 2014.Fujiwara, M., Wright, J. S., Manney, G. L., Gray, L. J., Anstey, J., Birner, T., Davis, S., Gerber, E. P., Harvey, V. L.,
Hegglin, M. I., Homeyer, C. R., Knox, J. A., Krüger, K., Lambert, A., Long, C. S., Martineau, P., Molod, A., Monge-Sanz, B. M.,
Santee, M. L., Tegtmeier, S., Chabrillat, S., Tan, D. G. H., Jackson, D. R., Polavarapu, S., Compo, G. P., Dragani, R., Ebisuzaki,
W., Harada, Y., Kobayashi, C., McCarty, W., Onogi, K., Pawson, S., Simmons, A., Wargan, K., Whitaker, J. S., and Zou, C.-Z.:
Introduction to the SPARC Reanalysis Intercomparison Project (S-RIP) and overview of the reanalysis systems, Atmos. Chem. Phys., 17,
1417–1452, 10.5194/acp-17-1417-2017, 2017.Gettelman, A., Pan, L. L., Randel, W. J., Hoor, P., Birner, T., and Hegglin, M. I.: The extratropical upper troposphere
and lower stratosphere, Rev. Geophys., 49, 1–31, 10.1029/2011RG000355, 2011.
Gilbert, R. O.: Statistical Methods for Environmental Pollution Monitoring, Wiley, New York, 1987.Heise, S., Wickert, J., Beyerle, G., Schmidt, T., Smit, H., Cammas, J.-P., and Rothacher, M.: Comparison of water vapor
and temperature results from GPS radio occultation aboard CHAMP with MOZAIC aircraft measurements, IEEE T. Geosci. Remote, 46,
3406–3411, 10.1109/TGRS.2008.920268, 2008.Helten, M., Smit, H. G. J., Sträter, W., Kley, D., Nedelec, P., Zöger, M., and Busen, R.: Calibration and
performance of automatic compact instrumentation for the measurement of relative humidity from passenger
aircraft, J. Geophys. Res.-Atmos., 103, 25643, 10.1029/98JD00536, 1998.Ho, S.-P., Peng, L., and Vömel, H.: Characterization of the long-term radiosonde temperature biases in the upper
troposphere and lower stratosphere using COSMIC and Metop-A/GRAS data from 2006 to 2014, Atmos. Chem. Phys., 17, 4493–4511,
10.5194/acp-17-4493-2017, 2017.Ingleby, B., Isaksen, L., and Dahoui, M.: ECMWF steps up assimilation of
aircraft weather data, available at:
http://www.ecmwf.int/en/newsletter/148/news/ecmwf-steps-assimilation-aircraft-weather-data,
last access: October 2016.
Kendall, M. G.: Rank Correlation Methods, 4th edn., Griffin, Charles, London, 1975.Khandu, Awange, J. L., and Forootan, E.: Interannual variability of
temperature in the UTLS region over Ganges–Brahmaputra–Meghna river basin
based on COSMIC GNSS RO data, Atmos. Meas. Tech., 9, 1685–1699,
10.5194/amt-9-1685-2016, 2016.Kunz, A., Schiller, C., Rohrer, F., Smit, H. G. J., Nedelec, P., and Spelten, N.: Statistical analysis of water vapour and
ozone in the UT/LS observed during SPURT and MOZAIC, Atmos. Chem. Phys., 8, 6603–6615, 10.5194/acp-8-6603-2008, 2008.Kunz, A., Spelten, N., Konopka, P., Müller, R., Forbes, R. M., and Wernli, H.: Comparison of Fast In situ
Stratospheric Hygrometer (FISH) measurements of water vapor in the upper troposphere and lower stratosphere (UTLS) with ECMWF
(re)analysis data, Atmos. Chem. Phys., 14, 10803–10822, 10.5194/acp-14-10803-2014, 2014.Kursinski, E. R., Hajj, G. A., Schofield, J. T., Linfield, R. P., and Hardy, K. R.: Observing Earth's atmosphere with
radio occultation measurements using the Global Positioning System, J. Geophys. Res.-Atmos., 102, 23429–23465,
10.1029/97JD01569, 1997.
Mann, H. B.: Nonparametric tests against trend, Econometrica, 13, 163–171, 1945.Marenco, A., Thouret, V., Nédélec, P., Smit, H., Helten, M., Kley, D., Karcher, F., Simon, P., Law, K., Pyle, J.,
Poschmann, G., Von Wrede, R., Hume, C., and Cook, T.: Measurement of ozone and water vapor by Airbus in-service aircraft: the MOZAIC
airborne program, an overview, J. Geophys. Res.-Atmos., 103, 25631–25642, 10.1029/98JD00977, 1998.Moninger, W. R., Mamrosh, R. D., and Pauley, P. M.: Automated meteorological
reports from commercial aircraft, B. Am. Meteorol. Soc., 84, 203–216,
10.1175/BAMS-84-2-203, 2003.Nédélec, P., Blot, R., Boulanger, D., Athier, G., Cousin, J.-M., Gautron, B., Petzold, A., Volz-Thomas, A., and
Thouret, V.: Instrumentation on commercial aircraft for monitoring the atmospheric composition on a global scale: the IAGOS system,
technical overview of ozone and carbon monoxide measurements, Tellus B, 67(MOZAIC/IAGOS), 1–16, 10.3402/tellusb.v67.27791,
2015.Neis, P., Smit, H. G. J., Rohs, S., Bundke, U., Krämer, M., Spelten, N., Ebert, V., Buchholz, B., Thomas, K., and
Petzold, A.: Quality assessment of MOZAIC and IAGOS capacitive hygrometers: insights from airborne field studies, Tellus B, 67,
1–11, 10.3402/tellusb.v67.28320, 2015.Pan, L. L., Randel, W. J., Gary, B. L., Mahoney, M. J., and Hintsa, E. J.: Definitions and sharpness of the extratropical
tropopause: a trace gas perspective, J. Geophys. Res.-Atmos., 109, 1–11, 10.1029/2004JD004982, 2004.Petersen, R. A.: On the impact and benefits of AMDAR observations in operational forecasting – Part I: A review of the
impact of automated aircraft wind and temperature reports, B. Am. Meteorol. Soc., 97, 585–602, 10.1175/BAMS-D-14-00055.1,
2016.Petzold, A., Thouret, V., Gerbig, C., Zahn, A., Brenninkmeijer, C. A. M., Gallagher, M., Hermann, M., Pontaud, M.,
Ziereis, H., Boulanger, D., Marshall, J., Nédélec, P., Smit, H. G. J., Friess, U., Flaud, J.-M., Wahner, A., Cammas, J.-P.,
and Volz-Thomas, A.: Global-scale atmosphere monitoring by in-service aircraft – current achievements and future prospects of the
European Research Infrastructure IAGOS, Tellus B, 67, 1–24, 10.3402/tellusb.v67.28452, 2015.Poli, P., Healy, S. B., Rabier, F., and Pailleux, J.: Preliminary assessment of the scalability of GPS radio occultations
impact in numerical weather prediction, Geophys. Res. Lett., 35, L23811, 10.1029/2008GL035873, 2008.Poli, P., Healy, S. B., and Dee, D. P.: Assimilation of Global Positioning System radio occultation data in the ECMWF
ERA-Interim reanalysis, Q. J. Roy. Meteor. Soc., 136, 1972–1990, 10.1002/qj.722, 2010.Randel, W. J., Shine, K. P., Austin, J., Barnett, J., Claud, C., Gillett, N. P., Keckhut, P., Langematz, U., Lin, R.,
Long, C., Mears, C., Miller, A., Nash, J., Seidel, D. J., Thompson, D. W. J., Wu, F., and Yoden, S.: An update of observed
stratospheric temperature trends, J. Geophys. Res.-Atmos., 114, D02107, 10.1029/2008JD010421, 2009.Reichler, T., Dameris, M., and Sausen, R.: Determining the tropopause height from gridded data, Geophys. Res. Lett., 30,
1–5, 10.1029/2003GL018240, 2003.Rieckh, T., Scherllin-Pirscher, B., Ladstädter, F., and Foelsche, U.: Characteristics of tropopause parameters as
observed with GPS radio occultation, Atmos. Meas. Tech., 7, 3947–3958, 10.5194/amt-7-3947-2014, 2014.Riese, M., Ploeger, F., Rap, A., Vogel, B., Konopka, P., Dameris, M., and
Forster, P.: Impact of uncertainties in atmospheric mixing on simulated UTLS
composition and related radiative effects, J. Geophys. Res., 117, D16305,
10.1029/2012JD017751, 2012.Santer, B. D., Mears, C., Doutriaux, C., Caldwell, P., Gleckler, P. J., Wigley, T. M. L., Solomon, S., Gillett, N. P.,
Ivanova, D., Karl, T. R., Lanzante, J. R., Meehl, G. A., Stott, P. A., Taylor, K. E., Thorne, P. W., Wehner, M. F., and Wentz, F. J.:
Separating signal and noise in atmospheric temperature changes?: the importance of timescale, J. Geophys. Res.-Atmos., 116, 1–19,
10.1029/2011JD016263, 2011.Schmidt, T., Cammas, J.-P., Smit, H. G. J., Heise, S., Wickert, J., and Haser, A.: Observational characteristics of the
tropopause inversion layer derived from CHAMP/GRACE radio occultations and MOZAIC aircraft data, J. Geophys. Res.-Atmos., 115,
D24304, 10.1029/2010JD014284, 2010.Seidel, D. J. and Randel, W. J.: Variability and trends in the global tropopause estimated from radiosonde
data, J. Geophys. Res.-Atmos., 111, D21101, 10.1029/2006JD007363, 2006.Seidel, D. J., Li, J., Mears, C., Moradi, I., Nash, J., Randel, W. J.,
Saunders, R., Thompson, D. W. J., and Zou, C.: Stratospheric temperature
changes during the satellite era, J. Geophys. Res.-Atmos., 121, 664–681,
10.1002/2015JD024039, 2016.Simmons, A. J., Poli, P., Dee, D. P., Berrisford, P., Hersbach, H.,
Kobayashi, S., and Peubey, C.: Estimating low-frequency variability and
trends in atmospheric temperature using ERA-Interim, Q. J. Roy. Meteor. Soc.,
140, 329–353, 10.1002/qj.2317, 2014.Solomon, S., Rosenlof, K. H., Portmann, R. W., Daniel, J. S., Davis, S. M.,
Sanford, T. J., and Plattner, G.-K.: Contributions of stratospheric water
vapor to decadal changes in the rate of global warming, Science, 327,
1219–1223, 10.1126/science.1182488, 2010.
Steiner, A. K., Kirchengast, G., Lackner, B. C., Pirscher, B., Borsche, M., and Foelsche, U.: Atmospheric temperature
change detection with GPS radio occultation 1995 to 2008, Geophys. Res. Lett., 36, L18702, 10.1029/2009GL039777, 2009.
Stickney, T. M., Shedlov, M. W., and Thompson, D. I.: Total Temperature
Sensors, Rosemount Aerosp., Goodrich, 1994.Stratmann, G., Ziereis, H., Stock, P., Brenninkmeijer, C. A. M., Zahn, A., Rauthe-Schöch, A., Velthoven, P. V.,
Schlager, H., and Volz-Thomas, A.: NO and NOy in the upper troposphere: nine years of CARIBIC measurements onboard a passenger
aircraft, Atmos. Environ., 133, 93–111, 10.1016/j.atmosenv.2016.02.035, 2016.Thomas, K., Berg, M., Boulanger, D., Houben, N., Gressent, A., Nédélec, P., Pätz, H.-W., Thouret, V., and
Volz-Thomas, A.: Climatology of NOy in the troposphere and UT/LS from measurements made in MOZAIC, Tellus B, 67,
1–16, 10.3402/tellusb.v67.28793, 2015.Thouret, V., Marenco, A., Logan, J. A., Nedelec, P., and Grouhel, C.: Comparisons of ozone measurements from the MOZAIC
airborne program and the ozone sounding network at eight locations, J. Geophys. Res.-Atmos., 103, 25695–25720,
10.1029/98JD02243, 1998.Thouret, V., Cammas, J.-P., Sauvage, B., Athier, G., Zbinden, R.,
Nédélec, P., Simon, P., and Karcher, F.: Tropopause referenced ozone
climatology and inter-annual variability (1994–2003) from the MOZAIC
programme, Atmos. Chem. Phys., 6, 1033–1051, 10.5194/acp-6-1033-2006,
2006.Wickert, J., Reigber, C., Beyerle, G., König, R., Marquardt, C.,
Schmidt, T., Grunwaldt, L., Galas, R., Meehan, T. K., Melbourne, W. G., and
Hocke, K.: Atmosphere sounding by GPS radio occultation: first results from
CHAMP, Geophys. Res. Lett., 28, 3263–3266, 10.1029/2001GL013117, 2001.
WMO: Meteorology a three-dimensional science, WMO, Geneva, WMO Bulletin, VI,
134–138, 1957.WMO: Aircraft meteorological data relay (AMDAR) – Reference manual, WMO-No.
958 available at:
https://www.wmo.int/pages/prog/www/GOS/ABO/AMDAR//publications/AMDAR_Reference_Manual_2003.pdf
(last access: Mai 2016), 2003.WMO: The Benefits of AMDAR Data to Meteorology and Aviation, Technical-Report
No. 2014-01, available at:
https://library.wmo.int/pmb_ged/wigos-tr_2014-01_en.pdf (last access:
June 2016), 2014.Xu, J. and Powell, A. M.: Ensemble spread and its implication for the
evaluation of temperature trends from multiple radiosondes and reanalyses
products, Geophys. Res. Lett., 37, 1–6, 10.1029/2010GL044300, 2010.