Introduction
Some 30 % of the temperature change since 1980 can be attributed to the
radiative forcing of increased stratospheric water vapour
and 10 % of the global total ozone decline from 1960 to 1999 can be
explained by the water vapour increase . However, the
mechanisms driving long-term changes in stratospheric water vapour are not
well understood . Stratospheric water vapour is
determined by in situ methane oxidation and the intrusion of water vapour
through the tropical tropopause layer (TTL). The seasonal cycle of lower
stratospheric water vapour in the Tropics is characterized by the tape
recorder , which exhibits a hydrated lower stratosphere in
boreal summer and fall and a dry lower stratosphere in boreal winter and
spring. Thus, most of the water vapour enters the stratosphere during boreal
summer when the tropopause temperatures are higher and monsoon systems
e.g. as well as enhanced deep convection over the
Tropics e.g. cause augmented transport of water vapour
into the TTL. The contribution of the individual mechanisms to the overall
budget of stratospheric water vapour, however, is poorly quantified.
The application of water isotopologues in tracer studies has the potential to answer
open questions concerning the stratospheric water vapour budget.
The different vapour pressures and binding energies of the respective
water isotopologues lead to fractionation effects during phase transitions and
chemical reactions.
The individual processes, which control the water vapour budget in the stratosphere,
thus leave an isotopic fingerprint in the respective water vapour compound .
Direct injection of ice into the stratosphere through
deep convection, dehydration
of air during ascent by in situ formation of cirrus clouds
and the in-mixing of older stratospheric air
lead to an offset of the isotope ratios
in the upper troposphere and lower stratosphere (UTLS) from simple
Rayleigh distillation see.
The mechanisms which are responsible for the intrusion of water vapour into the stratosphere
can hence be separated through the isotopic signature.
Recent satellite retrievals from three different instruments SMR,
MIPAS, ACE-FTS, see allow to obtain
a global picture of the water isotopologue HDO and the water vapour isotope
ratio δD(H2O) – where δD(H2O)=[HDO]/[H2O]RVSMOW-1⋅1000;
RVSMOW=155.76×10-6, ; VSMOW: Vienna
Standard Mean Ocean Water, – in the stratosphere and in the
transition region to the troposphere. In some regions the different satellite
retrievals, however, show considerable deviations when comparing profiles and
annual cycles of HDO mixing ratios and isotope ratios, respectively
. Furthermore, found a tape
recorder signal in δD(H2O) in the tropical stratosphere in the
MIPAS observations, comparable to the known tape recorder in water vapour
mixing ratios. In contrast, did not find an analogue
upward propagation of the seasonal cycle of water vapour isotope ratios in
the stratosphere when analysing ACE-FTS data. Speculations about this
discrepancy mainly focus on the different sampling techniques and retrieval
algorithms. Consequently, in order to improve the understanding of the basic
structure of δD(H2O) in the stratosphere, comprehensive modelling
of the physical and chemical isotope processes in a well-resolved global
chemistry climate model (CCM) with explicit stratospheric dynamics is
necessary.
Hitherto, a GCM (general circulation model) study by
including HDO and focusing on the UTLS region only featured a relatively
coarse vertical resolution in the stratosphere and only a prescribed tendency
for chemical influences on HDO. Here, the water isotopologue HDO is
implemented into the EMAC (ECHAM MESSy Atmospheric Chemistry; MESSy: Modular
Earth Submodel System) model . EMAC provides the
opportunity to accurately analyse troposphere–stratosphere exchange processes
on a global scale. When used with 90 layers in the vertical, its base model
ECHAM5; possesses a well-resolved tropopause
region the vertical resolution near the tropopause is about 500 m;
see and explicit stratospheric dynamics. Moreover, EMAC
optionally represents chemical processes, which are needed for the
computation of methane oxidation, which is crucial for the representation of
H2O and HDO in the stratosphere.
Confirmation of the correct representation of the fractionation processes
during phase transitions in EMAC is achieved through the evaluation of the
isotope ratios in precipitation. The consideration of the influence of
methane oxidation on atmospheric HDO requires the computation of the methane
isotopologue CH3D. In a next step, this tracer is evaluated with respect
to chemical transport models 1-D, CHEM2D, and
measurements from stratospheric balloon flights . Finally, the HDO mixing ratios and δD(H2O) in the stratosphere are
compared with the observations from satellites. This approach yields a more
complete picture of the isotopic composition of stratospheric water vapour
and provides insights into the discrepancies between the respective satellite
retrievals. Comprehensive analyses of stratospheric δD(H2O) in EMAC
will be presented in the companion part 2 paper
. These model results will be investigated with respect
to the sensitivity and the origin of the δD(H2O) tape recorder, and
the role of convective ice lofting on the pattern is analysed.
Model description and simulation setup
The MESSy submodel H2OISO (H2O ISOtopologues) has been incorporated into
the EMAC model system. This submodel comprises tracers
see for the three stable water isotopologues
H216O (“normal” water, hereafter denoted as H2O), H218O and
HDO for all three phases (vapour, liquid and ice), respectively. Moreover, it
contains an additional hydrological cycle, identical to the model's actual
hydrological cycle, which includes all processes that modify the tracers and
the corresponding fractionation effects for the isotopologues during phase
transitions. The modular approach of MESSy allows the optional usage of the
H2OISO submodel for all users in future EMAC versions. Besides this
structural difference, the implementation of the water isotopologues follows
previous studies with ECHAM3 , ECHAM4
and ECHAM5 . Supplementary to these previous studies, the
chemical fractionation effects during the formation of water vapour through
methane oxidation are considered. Since observations of water isotopologues
in the stratosphere are mostly available for H2O and HDO, the more
elaborate accounting for the chemical fractionation of H218O was not
conducted in this study. Hence, here H218O basically only serves the
evaluation of the model in the troposphere. The physical isotope effects of
HDO and H218O only differ by the corresponding fractionation factors.
Water isotopologues in the hydrological cycle of EMAC
The hydrological cycle in the H2OISO submodel reproduces the actual
hydrological cycle of the EMAC model. The tracers for H2O in the gaseous
(H2Ovap), the liquid (H2Oliq) and the ice
(H2Oice) phase were tested to be equal to the standard ECHAM
variables for the water vapour (Q), the liquid water (XL) and the ice water
(XI) content, respectively. Small numerical errors (the maxima are of the
order of 10-20 kg kg-1 s-1 for
water vapour and 10-22 kg kg-1 s-1 for liquid water and ice), which arise
due to the coding design, are corrected after each time step, in order to
prevent the two hydrological cycles from diverging. Moreover, we assured that
these numerical errors are small enough to not deteriorate the calculation of
the water isotopologues.
For the water isotopologues, fractionation effects occur during phase changes.
Equilibrium and kinetic fractionation during the evaporation of water
from oceans is described by the bulk formula of .
The surface flux for the water isotopologues depends on the isotope ratio of the ocean,
the wind speed above the surface, the sea surface temperature,
the specific humidity and the isotopic composition of the vapour above the
surface .
As in the study by , the isotope ratio of the ocean is prescribed with
a global gridded data set based on the 18O
isotopic composition in sea water by .
Since there is no equivalent data set for the deuterium isotopic composition,
the HDO content is prescribed as eight times the H218O isotope ratios.
This is in accordance with global observations .
Due to the limitations of the applied land surface scheme in the EMAC model,
we neglect any isotope fractionation from land surfaces.
The land surface scheme in EMAC is identical to the scheme in the ECHAM5 model.
A more detailed description of this simplification is given in .
As for water vapour, liquid water and ice in the actual hydrological cycle,
the advection of the new water isotopologue tracers follows the flux form
semi-Lagrangian (FFSL) scheme by .
The cloud and convection parameterisations (CLOUD and CONVECT) in EMAC
include a number of phase transitions and therefore several different
fractionation effects. Again, the implementation follows the study of
. During the formation of clouds, condensation of water
vapour to liquid water and deposition of vapour to cloud ice take place. For
condensation within clouds, a closed system is assumed. The condensate is
assumed to be in contact, and hence in isotopic equilibrium, with the
surrounding vapour during the entire process. This also applies for the
evaporation of cloud water, where, in contrast to evaporation from the ocean,
a closed system is assumed. An open system is used for the deposition of
water vapour to ice. Due to the low diffusivities of the isotopologues in the
ice phase, no exchange happens between ice and vapour. The effective
fractionation factor αeff is used here, including a function
for the supersaturation S. adjusted this function to
S=1.01-0.0045⋅Tcond (with Tcond as the condensation
temperature during ice crystal formation, given in ∘C), in order to
attain realistic isotope ratios in Antarctic snow. Since the focus of the
present study lies on the tropopause region, where also low temperatures have
a major effect on kinetic fractionation through deposition, as a first
estimate, the values from have been taken. During the
melting of ice and the freezing of water, no fractionation is assumed. Other
in-cloud processes like sedimentation of ice, autoconversion, accretion and
aggregation include no fractionation effects either. The isotopic
(non-)equilibrium factor αeff, which describes a fractionation
process between a falling raindrop and the surrounding water vapour, may vary
with time. Its value depends on the humidity, the temperature, the
diffusivity of the water molecules and the droplet size. Since these
processes are not resolved in GCMs, the fractionation during re-evaporation of
raindrops falling through undersaturated air can only be approximated.
Following , an isotopical equilibration of 45 % is
assumed for large drops from convective rain and 95 % for small drops
falling from stratiform clouds. Due to their low exchange rates, snow and ice
do not re-equilibrate at all, which leads to more depleted isotope ratios in
solid precipitation.
Stratospheric isotope chemistry for water and methane
The EMAC model contains the submodel CH4, which provides a simplified
methane oxidation mechanism to take into account the chemical production of
water vapour. It includes a tracer for methane (the CH4 tracer), which
experiences a source from the surface – here as lower boundary conditions from
the submodel TNUDGE, see next section; alternatively as methane fluxes,
provided by the submodel OFFEMIS see – and a sink in
form of methane oxidation. Solutions are calculated for the four oxidation
reactions, which are determined by the mixing ratios of the three oxidation
partners (Cl, OH, O(1D)) and the photolysis rate. The photolysis rate
jCH4 (=rhν) is here calculated in the MESSy submodel JVAL
for details, see and passed on to the
submodel CH4 (alternatively it can be prescribed). The rates for the
oxidation of methane with the reaction partners Cl, OH and O(1D) are
calculated within the submodel CH4. First, the first-order reaction
coefficients kOH for OH, kCl for Cl and kO1D
for O(1D) are determined. While kO1D=1.75×10-10cm3s-1 is constant, kOH and
kCl are temperature (T in K) dependent and are computed
by
kOH=1.85×10-20⋅exp2.82⋅log(T)-987Tcm3s-1
and
kCl=6.6×10-12⋅exp-1240Tcm3s-1.
Subsequently the pseudo first-order rate coefficients for the reactions with
methane are
rO1D=kO1D⋅cair⋅O(1D)rCl=kCl⋅cair⋅ClrOH=kOH⋅cair⋅OH,
with O(1D), Cl and OH representing the prescribed mixing ratios (in
mol mol-1) of the respective species and cair the
concentration of dry air molecules (in cm-3), which is calculated by
cair=NA×10-6⋅pRgas⋅T⋅1+MairMH2O-1⋅Q.
Here NA denotes the Avogadro constant
(6.022045×1023 mol-1), p the pressure (in Pa),
Rgas the universal gas constant
(8.314409 J K-1 mol-1), T the
temperature (in K), Mair the molar mass of dry air
(28.97 g mol-1), MH2O the molar mass of water
(18.02 g mol-1) and Q the specific humidity (in kg kg-1).
The tendency for the methane tracer (in mol mol-1 s-1) is then given by
∂(CH4)∂t=-1⋅CH4⋅(rO1D+rCl+rOH+rhν),
where CH4 is the methane mixing ratio (in mol mol-1) of the previous
time step and the -1 accounts for the fact that this is a pure sink
reaction for the methane tracer. To calculate the tendency for the specific
humidity due to methane oxidation,
∂Q∂t|C=-2⋅∂(CH4)∂tMairMH2O11-Q2
is applied. The subscript C denotes that this is the chemical tendency of
Q. This division is to convert the tendency from mol mol-1 s-1 to
kg kg-1 s-1. The negative sign here
accounts for the fact that methane oxidation is a source for water vapour and
the factor 2 for the reaction of the four hydrogen atoms of one methane
molecule into two water molecules.
In order to take into account the chemical production of HDO, analogously a
parameterisation for the oxidation of CH3D has been devised. A tracer for
CH3D, the most abundant deuterium isotopologue of methane, was included
for this purpose. The coefficients for the mass-dependent kinetic isotope
effects (KIE) for the reactions of CH3D with OH, O(1D) and Cl have been
determined in laboratory measurements by
. They are partly temperature dependent
and can be described with the function KIE(T)=A⋅exp(B/T). The
values for A and B and their temperature ranges are given in
Table see also.
Temperature-dependent kinetic isotope fractionation coefficients for
the reaction with CH3D. The kinetic isotope effect is determined by
KIE(T)=A⋅exp(B/T) for the given temperature range
see.
Reactant
T range
A
B
OH
1.097
49 ± 22
O(1D)
224–295
1.066
0
Cl
223–295
1.278
51.31 ± 19.1
The absorption cross-section of CH3D is shifted 0.9 nm blueward
relative to CH4 . For the photodissociation of CH3D,
this results in the fractionation factor KIEhν=0.995 in the atmosphere
of Mars see also. This approach is here applied to the
Earth's atmosphere, since the photodissociation characteristics of methane do
not differ from one planet of the solar system to another.
As for physical kinetic fraction processes, the Rayleigh equation
R=R0NN0KIE-1-1
is applied for the fractionation processes in the chemical reactions.
Inserting the isotope ratios (R0 and R) and the total mixing ratios
(N0 and N) before and after the reaction leads to
CH3D-∂(CH3D)∂tCH4-∂(CH4)∂t=CH3DCH4CH4-∂(CH4)∂tCH4KIE-1-1.
Using Eq. () for ∂(CH4)/∂t
and considering that the KIE is different for each of the reactions, the tendency of the CH3D tracer is given by
∂(CH3D)∂t=CH3D⋅1-1+rOHKIEOH-1+1-1+rClKIECl-1+1-1+rO1DKIEO1D-1+1-1+rhνKIEhν-1.
In order to calculate the tendency of the HDO tracer from the tendency of the
CH3D tracer (i.e. the chemical tendency of HDO),
analogously to Eq. (),
∂(HDO)∂t|C=-1⋅∂(CH3D)∂tMairMHDO11-HDO2
is applied. Here, we use the specific humidity and the molar mass of HDO
(MHDO=19.02 g mol-1) instead of Q and MH2O and the
factor 1 instead of 2, because the oxidation of one CH3D molecule can
only produce one HDO molecule. This approach, however, neglects the
intermediate reactions between CH3D and HDO with molecular hydrogen and
its isotopologue HD. While H2 is fairly constant throughout the
stratosphere and therefore can be neglected for the CH4 conversion to
H2O, δD(H2) increases with altitude due to chemical isotope
fractionation effects see. Hence, some
deuterium is stored in HD during this process and therefore one CH3D
molecule translates into a little less than exactly one HDO molecule
see also. In order to account for this, we apply the
stratospheric relation
HD=-6.32×10-5⋅CH4+0.297
derived from measurements by . The time derivative of HD in dependence
of the time derivative of CH4 can now be used to account for the stored deuterium in HD
and thus for the reduction of chemically formed HDO. Therefore,
Eq. ()
was extended to
∂(HDO)∂t|C=-∂(CH3D)∂t+6.32×10-5⋅∂(CH4)∂tMairMHDO11-HDO2.
This approach still neglects a number of effects that may be important
for the chemical production of HDO:
firstly, the other, rather rare methane
isotopologues CH2D2, CHD3 and CD4 as well as the reaction partner OD
(an isotopologue of the hydroxyl radical) are not considered.
Secondly, the fractionation effects during the formation of HDO by HD
are not taken into consideration.
Most of these, however, are poorly
quantified and therefore neglected for this initial study. These simplifications
have to be kept in mind when evaluating the model results in the stratosphere.
Emissions of CH3D could be defined with the aid of an extension (for
deuterium) of the MESSy isotope scheme of . However, this
task requires the integration of the full chemistry scheme and therefore goes
beyond the scope of our study, which mainly focuses on stratospheric and
upper tropospheric processes. This leads to another simplification: in
accordance with the isotope ratio of methane was fixed to
-86 ‰ in the troposphere (i.e. here, below 500 hPa). This is in
between the stratospheric entry values of δD(CH4)
(the δD(CH4) is also based on VSMOW) found by
(-90 ‰), (-80 ‰) and
(-81 ‰) and also within the range of recent
aircraft measurements presented by .
Simulation setup
For this study, an EMAC (v2.42) model simulation in T42L90MA resolution was
performed. This corresponds to an approximate horizontal grid box size of
2.8∘ × 2.8∘, 90 layers in the vertical and
explicitly resolved stratospheric dynamics. The uppermost model layer is
centred around 0.01 hPa. The MESSy submodels which were applied in this
simulation (in addition to the described H2OISO and CH4 submodels and the
ECHAM5 base model from EMAC version 2.42) are listed and described in the
Appendix. The time step of the simulation was 12 min and the output
was set to produce instantaneous values with an interval of 11 h. The
EMAC model provides the possibility to use several different convection
schemes. In all the simulations conducted for this study, the
“Tiedtke–Nordeng” convection scheme was
applied. Before starting the actual simulation, a 20-year free-running
simulation was carried out to obtain steady-state initial values for water,
methane and their isotopologues. From these initial conditions a simulation
with specified dynamics (i.e. in “nudged” mode) was started, which means a
Newtonian relaxation of the divergence, the vorticity, the temperature and
the logarithm of the surface pressure towards reference data. Here, the
relaxation is performed up to 1 hPa towards the ERA-INTERIM reanalysis data
ECMWF;. This guarantees that not only the climatic state,
but also the meteorological situation of the model simulation corresponds to
the actual states and allows a direct comparison of the model results with
satellite or in situ measurements. The simulation starts at the beginning of
the year 1982 and terminates at the end of the year 2010. From the
steady-state conditions, which are used for the initialisation, the model
again needs several years to adjust to the conditions of the “nudged” mode.
The first 8 years are hence not considered for the analysis. Only the
21 years from 1990 to 2010 are evaluated. Transient greenhouse gas
concentrations are prescribed throughout the atmosphere. Methane is
prescribed at the lower boundary through the submodel TNUDGE, based on
observations. The mixing ratios of OH, Cl, O(1D) are prescribed (monthly
averages) from a previous “nudged”, transient EMAC simulation with full
chemistry. The same applies for ozone, which is needed to calculate the
photolysis rate in the submodel JVAL. As mentioned above, the methane
oxidation is calculated in the CH4 submodel.
Model evaluation
Before evaluating the model simulation with respect to the isotopic
composition of water vapour in the stratosphere, we show that the
basis for this is established in the EMAC model. This includes the correct
representation of the physical water isotope effects in the troposphere
and the chemical HDO precursor CH3D
in the stratosphere, as well as its sink reactions.
Subsequently stratospheric HDO mixing ratios and δD(H2O) are compared
to satellite observations.
Water isotope ratios in precipitation
The isotopic composition in precipitation makes it possible to evaluate various
isotope fractionation effects .
Hence, the representation of the isotope physics in the model can
be evaluated by examining the isotope ratios in precipitation.
This has been conducted in many studies with isotopologue-enabled GCMs
e.g..
The GNIP (Global Network of Isotopes in Precipitation) measurement
survey provides a comprehensive database for this purpose.
The International Atomic Energy Agency (IAEA) and the World Meteorological Organisation
(WMO) started GNIP in 1961. Since its start, more than 800 meteorological stations in
101 countries have collected samples of the water isotopologues HDO and H218O in precipitation
. Some of these stations provided monthly precipitation measurements for
several decades. Most of them, however, operated for a much shorter period.
This data set serves as a basis for the evaluation of
the simulated isotope ratios in the troposphere.
Good qualitative agreement was achieved when comparing the general global patterns in
annual mean precipitation, as well as in the seasonal cycles between GNIP and EMAC.
Details can be found in the Supplement.
The isotopologue-enabled GCM ECHAM5-wiso (water isotopologues)
has also been tested successfully against GNIP data, as
well as against other water isotopologue measurements see.
Several model resolutions have been applied for this study.
The results of the lowest of these horizontal resolutions
(T31 ∼ 3.75∘ × 3.75∘)
was largely reproduced with the EMAC model (see Supplement).
Since the model physics and dynamics of the two models,
including the hydrological cycle and the implementation
of the water isotopologues (as described in Sect. ),
are almost identical,
this is not surprising. However, it is a prerequisite for assuring the correct representation
of the tropospheric isotope composition of water.
A detailed comparison between EMAC and ECHAM5-wiso
is given in the Supplement.
The good quantitative agreement of this comparison supports the conclusion,
that the EMAC
model with the H2OISO submodel represents the state of
the art of GCMs with an explicit representation of the water isotopologues
HDO and H218O in the troposphere.
CH<inline-formula><mml:math display="inline"><mml:msub><mml:mi/><mml:mn mathvariant="normal">3</mml:mn></mml:msub></mml:math></inline-formula>D in the stratosphere
The simulated CH3D
is compared to results from the 1-D model by
and the CHEM2D model by .
These models comprise an oxidation
scheme, where CH3D produces HDO through a number of chemical reactions.
This oxidation scheme was developed for the 1-D model.
For the CHEM2D model, it was extended for higher altitudes and
included into the Naval Research Laboratory 2-D
chemical/dynamical model.
In and the two chemical transport
models have shown good general agreement with measurements from
the ATMOS (Atmospheric Trace MOlecule Spectroscopy) instrument . ATMOS provides
global data for CH3D and HDO. However, with large uncertainties.
The equatorial values of δD(CH4) of EMAC,
CHEM2D and the 1-D model are presented in Fig. .
Comparison of equatorial averages of δD(CH4) with altitude
between
EMAC (red), CHEM2D (blue) by and the 1-D model
(dashed purple) by .
![]()
The tropospheric values of δD(CH4) in the CHEM2D and the 1-D model
are fixed to -68 and -65 ‰, respectively. For a better
comparison with these models, an additional EMAC simulation was carried out,
with the only difference that this value was set from -86 to
-68 ‰ (see Sect. ). Therefore, the model results do
not differ below the tropopause. Moreover, the overall dependence of
δD(CH4) on altitude qualitatively agrees in all the three model
simulations. Between 20 and 50 km altitude the methane isotope ratio
increases from -68‰ to around +120 ‰ in CHEM2D and the 1-D
model, and to around +130 ‰ in EMAC. Especially the increase in
the lower stratosphere is much stronger in EMAC, which leads to rather large
discrepancies in the altitude range between 30 and 40 km. Between 50 and
60 km, both CHEM2D and EMAC show almost no change in δD(CH4)
with altitude, and the 1-D model does not extend above 50 km. This is the
transition region between the altitudes of the chemical and the photolytic
methane oxidation. The photodissociation, which becomes important above
60 km and increases continuously above, is much stronger in CHEM2D. This
is somewhat surprising, because there is no fractionation included for the
photolysis of CH3D in the CHEM2D model. Even though the fractionation for
photolysis in EMAC is very small, the photolysis of CH3D is expected to be
of similar strength as in CHEM2D. The discrepancy is hence likely caused by
the differences in the calculation of the photolysis rates in EMAC and
CHEM2D. Since the mid of the uppermost layer of the EMAC model in the applied
resolution is at 80 km, a comparison further above is not possible.
Measurements of CH3D in the stratosphere are sparse. ,
however, collected 13 altitude profiles from stratospheric balloon borne
measurements, which were provided by the Max Planck Institute (MPI) for Solar
System Research and by the Institute for Atmospheric and Environmental
Sciences of the University of Frankfurt. The mixing ratios and the isotopic
composition of CH4 were measured using a high-precision continuous-flow
isotope ratio mass spectrometry system . Twelve of these
balloon flights can be used for direct intercomparison with the data from the
EMAC simulation. One flight (Flight ID: HYD-87-03) was conducted in 1987.
Since the model results are considered to be in steady state only from 1990
on, this profile is not taken into account for the evaluation. In
Fig. , 12 panels are presented showing the balloon-borne data and the EMAC results of δD(CH4) as function of altitude,
between 5 and 35 km. The flight IDs included in the panels provide
information about the location, the month and the year of the balloon flights
(see caption). To provide an estimate of the average and the annual
variability of δD(CH4) in the model simulation, additionally the
averages, maxima and minima of the 21 simulation years of the respective
months at the location of the launch are included in the panels.
Comparison of δD(CH4) vertical profiles between
EMAC and balloon-borne data by .
The red lines show the
observations and the blue lines show the EMAC data of the same day at the
location of the balloon launch. The black lines represent the simulated averages, minima and maxima
of the 21 monthly averages of the respective month
at the location of
the balloon launch. The flight IDs included in the panels denote the location, the year
and the month of the balloon flight. KIR: Kiruna, Sweden (67.9∘ N, 21.10∘ E);
GAP: Gap, France (44.44∘ N, 6.14∘ E);
HYD: Hyderabad, India (17.5∘ N, 78.60∘ E);
ASA: Aire sur l'Adour, France (43.70∘ N, 0.30∘ W).
![]()
In general, a good qualitative agreement is achieved between the measured and
the simulated data. Both simulation and measurements show an increase of the
methane isotope ratios from tropospheric values to values between 100 and
200 ‰ at 25 to 35 km in the Arctic region and to values between
0 and 100 ‰ in the mid-latitude and tropical regions. The
balloon-borne profiles are mostly lying within the extremes of the simulation
and close to the simulated values from the same day and location. The
measured δD(CH4) values in the troposphere and lower stratosphere
are in good agreement with the chosen value of -86 ‰ from the
study by . The steep increase of δD(CH4) above
25 km in the simulation can only be seen in the measurements of the Arctic
region (KIR). In the mid-latitude (GAP and ASA) profiles, the
δD(CH4) increase with height is smaller, which leads to partly
large deviations at these altitudes. The differences in the KIR-00-01 and the
KIR-03-03 profiles between simulation and balloon flights are exceptional.
Apart from the two uppermost data points in the KIR-03-03 profile, the
measured δD(CH4) values are constantly higher than the simulated
values. These two profiles are associated with special meteorological and
thus chemical situations. The KIR-03-03 data comprise a mesospheric
enclosure and during the sampling of the KIR-00-01 data, a strong Arctic
vortex was present . These phenomena can also be observed
in the simulation – due to the “nudging” these special meteorological
situations and the associated chemical situations are broadly represented.
However, the horizontal resolution of the model simulation is too coarse to
resolve sharp horizontal gradients around the site of the balloon launch.
Moreover, the balloon's drift off the launching site while ascending can
cause deviations of such magnitudes. Also, this may be related to omitting
the chemical cycle of H2 and HD, since isotopically very heavy H2 might
have been reconverted to CH4 in these profiles.
Another method for evaluating the methane isotope chemistry is assessing the
relation of δD(CH4) to the CH4 mixing ratio. The
δD(CH4) values of the same data as in Fig. ,
from 5 to 35 km altitude, are plotted versus the CH4 mixing ratios in
Fig. . The figure is divided into the launches in the
polar region (KIR) in the left panel and the launches in mid- (ASA, GAP) and
tropical (HYD) latitudes in the right panel.
Again, overall consistency between the simulation and the measurements is visible.
Apart from a single exception (the measurement from the HYD-99-04 profile, right panel)
the simulated δD(CH4) as function of the methane mixing ratios agrees
very well with the measurements.
The slope of increasing isotope ratios with decreasing methane mixing ratios
is almost identical. Since these compact tracer–tracer correlations are generally
found for trace gases whose life times are longer than the transport times
, it implies that the chemical removal of the
CH3D tracer in relation to the removal of the CH4 tracer
is well represented, despite the simplified chemistry parameterisation.
HDO in the stratosphere
During the first decade of the 21st century, three satellite missions
collected data applicable for the retrieval of the water isotopologue HDO in
the stratosphere. The MIPAS (Michelson Interferometer for Passive Atmospheric
Sounding) instrument on Envisat (Environmental Satellite) allowed the
retrieval of HDO by measuring the thermal emission in the mid-infrared. This
high-resolution Fourier transform spectrometer measured at the atmospheric
limb and provided data for HDO retrievals in full spectral resolution from
July 2002 to March 2004, roughly in the altitude range between 10 and
50 km. It orbits the Earth sun-synchronously 14 times a day. The vertical
resolution for the retrieval of HDO is around 5 km between the tropopause
and 30 km altitude; above it degrades (∼ 8 km at 40 km)
. The Odin satellite also orbits the Earth
sun-synchronously and carries the SMR (Sub-Millimetre Radiometer) instrument,
among other purposes to passively measure HDO on the global scale roughly on
1 day per week. It operates in the microwave range. Data has been retrieved
from the start of the mission in 2001 until today, at altitudes between
roughly 20 and 70 km with a vertical resolution of around 3 km
. The ACE-FTS (Atmospheric Chemistry Experiment Fourier
Transform Spectrometer) instrument circularly orbits the Earth on the SCISAT
satellite and obtains Fourier transform absorption spectra from solar
occultation measurements. It has a vertical resolution between 2 and 6 km
and a comparably limited spatial sampling. The ACE orbit is optimised for
measurements in mid- and high latitudes; the Tropics are only covered during
the 4 months of February, April, August and October
.
Comparison of the relations
of δD(CH4) to CH4 between EMAC (black)
and balloon-borne data (red; data as in Fig. ). The
left panel shows the Arctic data (KIR, △: vortex, □: non-vortex)
and the right panel shows the mid-latitude (∗) and tropical data (∘).
The altitude range is 5 to 35 km.
![]()
collected data of the three instruments for
intercomparison and concluded a good consistency between MIPAS and ACE-FTS at
altitudes above 20 km. Below this altitude, issues like different cloud
filtering and measurement techniques as well as different vertical
resolutions cause large deviations. Generally, in the stratosphere the MIPAS
and the ACE-FTS data agree favourably; the SMR data show considerably drier
conditions, especially below 30 km. This is mainly due to uncertainties in
the different spectroscopies of the instruments .
These data could now also be used to evaluate the HDO simulated by
EMAC. In Fig. , the tropical
(15∘ S to 15∘ N) HDO mixing ratios
of the three satellite instruments and the EMAC model
are presented. Additionally to the data of MIPAS, SMR and version 2.2 of ACE-FTS shown by
, here also version 3.0 of ACE-FTS is included, which reaches higher up
in the stratosphere, compared to version 2.2.
Since ACE-FTS only provides data for 4 months per year in this region, the panels show averages for
February, April, August and October.
Since the years of the ACE-FTS and the MIPAS satellite retrievals do not overlap,
a direct comparison is not possible.
However, tests have shown that the averages of the periods of the individual retrievals do not show
substantial differences between each other and between the average of the entire simulation.
In order to get a good estimate for the comparison of the model with all retrievals
and for the annual variability,
the averages, the minima and the maxima of the respective months
have been taken from the entire 21 years of the EMAC simulation.
Since, at first order, the HDO mixing ratios follow those of H2O, the profiles of H2O
of EMAC and of the mentioned satellite retrievals are shown in
Fig. in the same manner, in order to
allow a more detailed discussion of the results.
Comparison of HDO mixing ratio–altitude profiles between EMAC and
various satellite observations.
Black: averages, minima and maxima of 21 monthly averages of the EMAC simulation
for the respective months;
green: Odin/SMR; blue: ENVISAT/MIPAS; red: SCISAT/ACE-FTS-2.2;
purple: SCISAT/ACE-FTS-3.0.
![]()
Comparison of H2O mixing ratio–altitude profiles between EMAC and
various satellite observations.
Black: averages, minima and maxima of 21 monthly averages of the EMAC simulation
for the respective months;
green: Odin/SMR; blue: ENVISAT/MIPAS; red: SCISAT/ACE-FTS-2.2;
purple: SCISAT/ACE-FTS-3.0.
![]()
The EMAC data are generally drier in HDO and in H2O compared to the MIPAS
and the ACE-FTS profiles in each of the presented months at all altitudes.
For water vapour in the UTLS, this offset has already been shown by
and is consistent with the slightly too cold hygropause in
the nudging data see e.g.. Only between 30 and 35 km do
the H2O profiles of EMAC increase more strongly than in the satellite data and
reach the level of ACE-FTS H2O mixing ratios. For HDO, this is not the
case – its increase with height at these altitudes is rather similar between
EMAC and the satellite retrievals and therefore the offset remains. In the
altitude range between 16 and 30 km, HDO in the EMAC simulation
quantitatively corresponds well with the Odin retrieval. In this region also
local maxima and minima, which can be seen in all four satellite profiles, are
reproduced qualitatively in EMAC. These reveal the seasonal cycle of HDO and
H2O. Due to the spectral band that is used to derive these parameters by
SMR, in this special case, H2O cannot be retrieved below 20 km while
the limit for HDO is between 17 and 18 km. Especially for April and
August, the local HDO minimum between 25 and 30 km in the EMAC data is not
as pronounced as in the satellite retrievals. Above the local minima the HDO
mixing ratios of all satellite profiles increase strongly with altitude to
values around 1.1 nmol mol-1 at 50 km, while the EMAC simulation
shows HDO mixing ratios of only 0.6 nmol mol-1 at these altitudes.
In H2O, this behaviour is observable as well. However, only at altitudes
above around 40 km and at a significantly smaller magnitude. Therefore, to
some extent, this is most likely due to the assumptions made in the chemistry
parameterisation for HDO, which does not include the influence of the
fractionation effects during the reactions concerning molecular hydrogen.
The simplified methane oxidation scheme itself, however, also plays a role
here.
The stratospheric <inline-formula><mml:math display="inline"><mml:mi mathvariant="italic">δ</mml:mi></mml:math></inline-formula>D tape recorder
The tape recorder signal in H2O, HDO and δD(H2O) in the EMAC simulation is evaluated
with respect to the MIPAS data.
The satellite and the model data are compared in Fig. .
The left panels show the EMAC results and the right panels show the MIPAS retrieval.
Altitude–time diagrams of H2O, HDO and δD(H2O) in the Tropics (15∘ S–15∘ N).
Left: EMAC simulation, right: MIPAS observations.
![]()
Overall, there is only a rather weak agreement between EMAC and MIPAS in all
three quantities. As already shown in the previous section, a persistent (all
altitudes, all seasons) dry bias in H2O and HDO is visible in the EMAC
simulation. Also δD(H2O) is generally too low throughout the
stratosphere. In the lower stratosphere, some of these differences can be
explained with the coarse vertical resolution of the MIPAS retrieval, which
smoothes the hygropause see. In the upper
stratosphere, the underestimation by the model can be associated to the
assumptions in the chemistry parameterisation. In the lower stratosphere, the
strongest deviations in δD(H2O) can be observed during Northern
Hemisphere (NH) summer
when MIPAS observations show δD(H2O) values around
-500 ‰ and the δD(H2O) values in EMAC do not exceed
-600 ‰. A tape recorder signal can be seen in all three quantities
for both the model and observations, although with different amplitudes and a
phase shift of 2–3 months. While the maxima of the tape recorder in
the lower stratosphere in EMAC are found during summer, the satellite data
show them at the beginning of autumn. This can partly (at maximum 1 month)
be an artefact of the MIPAS retrieval and its coarse vertical sampling, but
has to be kept in mind and compared with other data sets. The tape recorder
signals in HDO and H2O fade out at around 30 km in both model and
observations. The δD(H2O) tape recorder signal in MIPAS reaches
these altitudes as well, the EMAC δD(H2O) tape recorder, in
contrast, fades out somewhat lower.
The amplitude of the tape recorder in EMAC is larger for H2O and smaller
for HDO and δD(H2O), respectively, compared to the MIPAS data.
Above 28 km, the δD(H2O) tape recorder becomes overshadowed by
chemically produced high δD(H2O) values. This effect is further
investigated in the companion part 2 paper . The MIPAS
δD(H2O) tape recorder might be amplified artificially by the
offsets in vertical resolution between H2O and HDO . A
correction of this error is likely to reveal a δD(H2O) tape
recorder signal with smaller amplitude in the MIPAS retrieval.
By analysing ACE-FTS data, found a tape recorder signal in H2O and in HDO, but could
not find a corresponding pattern
in δD(H2O). The lower stratosphere in Fig. 7 of
shows distinct seasonally varying maxima and minima of δD(H2O); these, however, hardly
propagate upwards in time. For comparison with the ACE-FTS data, the
tropical H2O, HDO and δD(H2O) in the stratosphere of the
EMAC simulation are displayed in Fig. for the
same period and altitudes as in .
Altitude–time diagrams of H2O, HDO and δD(H2O) in the
Tropics (15∘ S–15∘ N) from the EMAC simulation.
Left: monthly averages; right: averages of only February, April, August and October, as in
. The plotting algorithm linearly interpolates between the available months.
![]()
The left panels show the monthly averages of the respective values.
Qualitatively these agree fairly well with the ACE-FTS observations by
. The increase of HDO and therewith δD(H2O) with
altitude in the stratosphere is too weak, however. At 30 km
δD(H2O) exhibits values below -600 ‰, while in the
ACE-FTS retrieval values of around -500 ‰ can be seen. Similarly
to the comparison with the MIPAS retrieval, also the lower stratosphere in
EMAC is drier in H2O and in HDO. Still, a clear tape recorder signal can
be observed in all three panels, with minimum values in the lower
stratosphere during boreal winter and maxima during boreal summer.
For the right panels, the EMAC data have been filtered, using only the 4
months (February, April, August and October) which are also available in the
ACE-FTS retrieval, to estimate the influence of the sparse temporal sampling
on the tape recorder signals. This filtering somewhat blurs the tape recorder
in all three panels, compared to the full data set. In particular, the tape
recorder in δD(H2O) appears to lose some of its upward motion at
around 20 km and generally becomes less obvious. Therefore, it can be
assumed that the sparse temporal sampling of ACE-FTS data is an issue in the
evaluation of the δD(H2O) tape recorder and may well contribute to
the indistinctness of the signal in the study by .
Due to its instrumental design, the SMR satellite instrument measures H2O and HDO
orbitally shifted. This means that H2O and HDO are never measured at the same time and
implies inaccuracies for the calculation of δD(H2O). A comparison of the δD(H2O)
tape recorder signal between EMAC and SMR has therefore not been conducted.
Summary and discussion
The results of the EMAC simulations were first assessed with respect to
the isotope ratios in precipitation,
in order to ensure the correct representation of the physical processes
in the troposphere.
The EMAC results agree very well with GNIP data and are
almost identical with results from the ECHAM5-wiso model.
Hence the physical processes and fractionation
effects of the water isotopologues in
the hydrological cycle of EMAC are represented satisfactorily (see the Supplement).
The chemical HDO precursor CH3D in the stratosphere was evaluated next.
Here, a good qualitative agreement with chemically elaborate transport models by
and , as well as with balloon flights by is achieved.
Differences in the model dynamics as well as in the fractionation coefficients can
contribute to disagreements between the model results.
The coarse horizontal resolution of the model,
the possible drift of the balloon and associated local meteorological features
and especially the sparsity of the measurements
are the main issues in the comparison of δD(CH4) in EMAC with observations.
As desired by , this modelling approach can assist further studies,
especially concerning the investigation of the isotope effects of the chemical
sink reactions of methane isotopologues.
For instance, and discuss the application
of apparent rather than laboratory-based fractionation factors (KIEapp=KIE)
for long-lived trace gases that are removed in the stratosphere.
While in the EMAC simulation the laboratory-based fractionation factors lead to very good
agreement with measurements, according to turbulent diffusion and mixing of
air masses lead to significantly smaller fractionation factors.
Anyhow, more extensive measurements are desired for further evaluation
of the methane isotope ratios in the stratosphere.
The mixing ratios of HDO and H2O were compared with the results of
satellite retrievals. Inconsistencies between the individual satellite
retrievals see render it difficult to define a distinct
result. In the UTLS, the HDO profiles of EMAC agree well with the SMR
satellite observations, but the H2O retrieval of SMR only reaches down to
around 20 km. Compared to the ACE-FTS and the MIPAS satellite profiles
retrieved from measurements in the mid-infrared, the UTLS appears to be too
dry in H2O and in HDO in the EMAC model. Since for H2O this dry bias
has already been discussed see and HDO is, at first
order, determined by H2O, this dry bias for HDO is not surprising. Reasons
for this are the slightly too cold hygropause in the nudging data see
e.g. and the coarse horizontal resolution of the model. However,
due to the rather low vertical resolution of the satellite observations, the
hygropause is blurred in the retrievals, and therefore H2O as well as HDO
is overestimated in MIPAS see e.g..
For H2O, this offset weakens between 30 and 40 km and becomes stronger
again further above. At around 35 km, the simulated H2O increases
more strongly with altitude than the satellite observations show. Above 40 km,
simulated H2O increases at much smaller rates compared to the satellite
observations. These inconsistencies are most likely caused by the simplified
methane oxidation scheme, since they are obvious during all 4 months where
observations of ACE-FTS are available.
HDO in contrast shows a too weak increase with altitude from around 30 km
upwards. Largely, this is attributable to the lack of the intermediate
reactions containing HD in the CH3D oxidation chain in the model. Although
we accounted for the deuterium storage in HD using a relation
by, the isotope effects of the intermediate reactions on HDO
are not taken into account in the model. As has been shown by
and , stratospheric δD(H2)
increases with altitude, while the mixing ratios of molecular hydrogen are
rather constant. The conversion of isotopically very heavy molecular hydrogen
to water would therefore increase the HDO mixing ratios strongly, while
leaving H2O largely unaffected. Additionally, the influence of the
oxidation of CH3D itself also increases with altitude. This implies that
the importance of the intermediate reaction with HD on HDO, which is not
considered here, increases with altitude too. Our simplifications can
therefore explain the offset in HDO mixing ratios between the EMAC simulation
and the satellite retrievals and the different behaviour of H2O and HDO.
This leads to too low δD(H2O) values in the upper parts of the
stratosphere.
In future studies, the chemical deficiencies of the H2OISO submodel can be reduced by
including an explicit representation of HD to take into account the intermediate
reactions from CH3D to HDO.
However, according to the reaction rates of HD and especially the
fractionation effects of these reactions are
poorly quantified and hence can be subject to large uncertainties.
As an alternative, however more sophisticated modelling approach, the water isotopologues can be included
in the isotope scheme of , by extending it to the deuterium cycle.
This features a number of isotopic species and reactions, which
also affect the hydrogen reactions.
The tape recorder signals of H2O, HDO and δD(H2O) of the EMAC simulation
were compared to the MIPAS retrieval see.
In the lower stratosphere, EMAC and MIPAS
are closest during winter, where the differences can mostly
be explained with the low vertical resolution of the MIPAS sampling.
The summer months, however, show much stronger deviations between model and
observations, with too low values in all quantities in the simulation, especially
in HDO and δD(H2O).
This suggests deficiencies in the model physics, for example
the under-representation of overshooting convection in the convection scheme
Tiedtke–Nordeng; applied here.
According to the studies by and
a more complete representation of this effect can
enhance the mixing ratios of HDO, while H2O is hardly affected.
This would increase δD(H2O) in the UTLS especially during NH summer,
since convective events are more frequent during that time of the year.
Thus, a better representation of the δD(H2O) tape recorder in the model
could be achieved.
However, the inconsistent
vertical resolutions between H2O and HDO in the MIPAS retrieval
may be the reason for a spurious amplification of the δD(H2O)
tape recorder signal in the observations . The correction of this artefact
in the MIPAS data is subject of current investigations and may also lead to
a better agreement between the EMAC and MIPAS amplitudes of the
δD(H2O) tape recorders.
The influence of convective ice lofting on the isotopic composition of
stratospheric water vapour in this simulation is analysed in the companion part 2 paper .
In future studies, the quality of other convection schemes could be tested with respect to
the results of HDO in the UTLS and a possibly better representation of overshooting convection.
This, however, requires the
implementation of the water isotopologues in the alternative convection schemes.
The phase shift of 2 to 3 months between the tape recorders of MIPAS and EMAC,
which can be observed in all three quantities, is not easily explainable.
Due to the coarse vertical resolution of the MIPAS retrieval there might be a delay in the
retrieved tape recorder signals of up to 1 month.
There is no evidence that the seasonality of convection in EMAC is phase shifted.
Simulated patterns of moisture in the UTLS compare well with observations and analyses
of the seasonal cycle of zonal mean precipitation (convective and large-scale clouds)
and integrated water vapour conducted by and also
show accordance with observations.
The seasonal cycle of lower stratospheric δD(H2O) in the ACE-FTS
retrieval see shows a different behaviour than that of
the MIPAS retrieval and the EMAC simulation. The too low δD(H2O)
values in EMAC compared to ACE-FTS, especially during summer and in the upper
stratosphere, are consistent; however, did not find the
tape recorder signal in δD(H2O). The δD(H2O) tape
recorder in EMAC is weaker than the corresponding signals in H2O and HDO.
Still, the pattern is clearly recognisable. A possible reason for the lack of
the upward propagation of the seasonal cycle of δD(H2O) in the
ACE-FTS observations is the sparse temporal sampling of the instrument in the
Tropics. A filtering of the EMAC data to the sampling reduces the apparent
temporal upward motion of the δD(H2O) tape recorder in the UTLS
between around 17 and 20 km. Since δD(H2O) does not increase
strongly enough with altitude in the EMAC simulation due to the incomplete
chemistry, a weaker δD(H2O) tape recorder signal is expected for
the correction of this effect. This would blur the tape recorder signal in EMAC
above 20 km and therewith suppress its upward propagation. Below
around 25 km, however, the chemical influence is very weak and thus
some tape recorder signal will remain. The effect of methane oxidation on the
δD(H2O) tape recorder is analysed more deeply in the companion
part 2 paper.
For more quantitative comparisons between models and observations,
more sophisticated methods will be applied in future evaluations.
The vertical
resolution of the EMAC data can be transformed to the resolution of the respective satellite retrieval
using their averaging kernels. Also, the cloud filtering methods
used for the satellite data can be applied to
the model data. This elaborate evaluation can possibly assist to reduce the
discrepancies between model results and observations and reveal the model
and the measurement insufficiencies more precisely.
In the present and in the companion (part 2) study, however, a more comprehensive understanding
of the δD(H2O) tape recorder is desired, which can best be
achieved with the complete data record.