Stratospheric Dehydration, Residence Time, and Age-of-Air Inferred from the SDW Forward Trajectory Model – An Intercomparison of using MERRA, ERA interim, JRA55, and CFSR Oct. 19, 2016 SPARC RIP Tao Wang (JPL, former aggie) Andrew Dessler (Texas A&M Univ.) Mark Schoeberl (STC Corp.)
56
Embed
Stratospheric Dehydration, Residence Time, and Age-of-Air Inferred ...
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Stratospheric Dehydration, Residence Time, and Age-of-Air
Inferred from the SDW Forward Trajectory Model – An Intercomparison of using MERRA, ERA interim, JRA55, and CFSR
Oct. 19, 2016SPARC RIP
Tao Wang (JPL, former aggie)
Andrew Dessler (Texas A&M Univ.)
Mark Schoeberl (STC Corp.)
Motivation
1/24
ContributetoS-RIP
• Running forward (TRAJ3D by Ken Bowman)straightforward; longer (16 years) integration period.
(Schoeberl et al., 2003 JGR)
these are backward integrations, Figure 1b shows tropicalparcels descending relative to those in Figure 1a. In aforward integration the parcels would be rising in responseto tropical heating.)[24] The two diabatic simulations, using assimilated
winds from UKMO and the FVDAS, show parcels rapidlymoving to middle latitudes after 50 days. The diabaticdistributions are generally similar even after 200 daysalthough the FVDAS integration is beginning to show anupward plume in the north polar region not seen in theUKMO case. In contrast, the UKMO and FVDAS kine-matic integrations show large vertical dispersion of parcelsafter 50 days; some parcels have already moved into thetroposphere and have been removed from the model. Strik-ingly, the FVGCM kinematic integration shows almost nomeridional or vertical dispersion after 50 days and thedistribution is still confined to middle and low latitudes
after 200 days, while the four DAS experiments havemoved parcels to the polar regions.[25] In order to quantify the initial dispersion of parcels
from the tropics, we have computed the decay rate for thenumber of parcels in the tropics during the first six monthsof the integration. This short period insures that the parcelcount is representative of the initial dispersion and notcontaminated by parcels recirculated from midlatitudes. Ofcourse, the calculation includes the effects of both verticaland horizontal dispersion. The decay rates a (in years!1) fornumber of parcels between 15!S and 15!N in the lowerstratosphere for the first five experiments shown in Table 1are as follows (the experiment labeling in Figure 2 is used):UKM D., 3.7; UKM K., 5.2; FVDAS D., 2.2; FVDAS K.,4.2; FVGCM, 0.35. The data is least-squares fit to theexponential form exp(!at). The rates reflect the impressiongiven in Figure 1. Higher values of a mean more rapid
Figure 1. The distribution of parcels 50 days (part a) and 200 days (part b) after the beginning of theback trajectory calculation (Dt = !50, !200 days). The lower thin white lines show the zonal meanaltitude of the tropopause, the upper thin white line shows the zonal mean altitude of the 380K isentrope.The short thin vertical gray line segment at 20 km in each figure over the equator shows the initialposition of the parcels. Grayscale indicates zonal mean temperature. Parcels are shown as white dots. Thefar left panel shows the results using the UKMO DAS wind fields, diabatic trajectories (UKM D.). Thenext panel (left to right) uses the same wind fields, but is a kinematic trajectory calculation UKM K.).The third panel uses the FVDAS wind fields and diabatic trajectories (FVDAS D.). The fourth panel usesthe FVDAS with kinematic trajectory calculation (FVDAS K.). The fifth panel shows the kinematictrajectory calculation using the FVGCM (FVGCM K.). The percent of parcels remaining in thestratosphere at the time are indicated in each panel.
SCHOEBERL ET AL.: LOWER STRATOSPHERIC AGE SPECTRA ACL 5 - 5
these are backward integrations, Figure 1b shows tropicalparcels descending relative to those in Figure 1a. In aforward integration the parcels would be rising in responseto tropical heating.)[24] The two diabatic simulations, using assimilated
winds from UKMO and the FVDAS, show parcels rapidlymoving to middle latitudes after 50 days. The diabaticdistributions are generally similar even after 200 daysalthough the FVDAS integration is beginning to show anupward plume in the north polar region not seen in theUKMO case. In contrast, the UKMO and FVDAS kine-matic integrations show large vertical dispersion of parcelsafter 50 days; some parcels have already moved into thetroposphere and have been removed from the model. Strik-ingly, the FVGCM kinematic integration shows almost nomeridional or vertical dispersion after 50 days and thedistribution is still confined to middle and low latitudes
after 200 days, while the four DAS experiments havemoved parcels to the polar regions.[25] In order to quantify the initial dispersion of parcels
from the tropics, we have computed the decay rate for thenumber of parcels in the tropics during the first six monthsof the integration. This short period insures that the parcelcount is representative of the initial dispersion and notcontaminated by parcels recirculated from midlatitudes. Ofcourse, the calculation includes the effects of both verticaland horizontal dispersion. The decay rates a (in years!1) fornumber of parcels between 15!S and 15!N in the lowerstratosphere for the first five experiments shown in Table 1are as follows (the experiment labeling in Figure 2 is used):UKM D., 3.7; UKM K., 5.2; FVDAS D., 2.2; FVDAS K.,4.2; FVGCM, 0.35. The data is least-squares fit to theexponential form exp(!at). The rates reflect the impressiongiven in Figure 1. Higher values of a mean more rapid
Figure 1. The distribution of parcels 50 days (part a) and 200 days (part b) after the beginning of theback trajectory calculation (Dt = !50, !200 days). The lower thin white lines show the zonal meanaltitude of the tropopause, the upper thin white line shows the zonal mean altitude of the 380K isentrope.The short thin vertical gray line segment at 20 km in each figure over the equator shows the initialposition of the parcels. Grayscale indicates zonal mean temperature. Parcels are shown as white dots. Thefar left panel shows the results using the UKMO DAS wind fields, diabatic trajectories (UKM D.). Thenext panel (left to right) uses the same wind fields, but is a kinematic trajectory calculation UKM K.).The third panel uses the FVDAS wind fields and diabatic trajectories (FVDAS D.). The fourth panel usesthe FVDAS with kinematic trajectory calculation (FVDAS K.). The fifth panel shows the kinematictrajectory calculation using the FVGCM (FVGCM K.). The percent of parcels remaining in thestratosphere at the time are indicated in each panel.
SCHOEBERL ET AL.: LOWER STRATOSPHERIC AGE SPECTRA ACL 5 - 5
10
20
30
Height(km)
Diabatic run: realistic upwelling
vertical motion = dθ/dt
Kinematic run vertical motion = dP/dt
• Domain-filling: statistically robust
• Diabatic Run (vertical coordinate θ)
The S-D-W Forward Trajectory MODEL
50-days integration
2/24
Parcels are thinned out by a factor of 3
3yrs
The concept of domain filling
Pressure(h
Pa)
2/24
Endingpoint
Startingpoint Atmost16years
u Along each parcel’s integration path, we record:location, T, H2O, dehydration events,age …
u Parcels travelled below 345 K (~10km) and above 1800-K (~40km) are removed.
This model has been used to study:• Dehydration/H2O (Schoeberl and Dessler, 2011; Schoeberl et al., 2012, 2013; Wang et al., 2015);
• Transport of O3/CO (by using chemical prod/loss rates, Wang et al., 2014);• Age spectrum (Schoeberl and Dessler, 2011; Schoeberl et al., 2012; Ray et al., 2014);• Cloud Formation (Schoeberl et al., 2014, 2016);• Water vapor feedback (Dessler et al., 2013);• Water vapor long-term variability (Dessler et al., 2014) & future projection (Dessler et al., 2015);• Indian/North American monsoon (Zhang et al., 2016; Schwartz et al., in progress; Randel et al., in progress);• MJO (Wright et al., in progress);• Convective influences (Su et al., in progress) ;… 3/24
!
!52
Figure 3.5. Water vapor tape recorder signal averaged over 15o N-S from August 2004 to December 2009. The black contours are MLS H2O overlaid in each panel to emphasize the comparison in propagation of this signal.
Fig. 3.5 shows that all model runs driven by different reanalyses did a good job
reproducing the tape recorder up to ~10 hPa (~30 km). Apparently, the ERAi run shows a
drier stratospheric entry level of H2O, due to the cold temperature bias displayed in Fig.
3.4b. CFSR on the other hand, shows wetter air of 0.7-1.4 ppmv due to its warm bias
(Fig. 3.4c). The MLS H2O contours are overlaid in each panel to compare the vertical
propagation of the tape recorder signal. It is obvious that ERAi run creates a faster
transport than the MERRA and CFSR runs, caused by the larger diabatic heating in the
ERAi datasets.
The different transport time scales hinted at from three reanalyses are more
clearly shown in Fig. 3.6, which compares the diabatic heating rates and thus the vertical
Traj.MERRA
Traj.ERAi Traj.CFSR
MLS
4/24
1970 1980 1990 2000 2010Time
-1.0
0.0
1.0
2.0
H2O
Ano
m. (
ppm
v)
1970 1980 1990 2000 2010
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[1960.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_B_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_newT_since80_var1_monCat_1980-2013.nc; X02: 120916_sat100_init370_ERAi_6hr_15yr_addGPH_var1_monCat_1980-2013.nc; X03: 150101_s100_i370_mthd_inj1sav3_JRA55windT_day_dhAll_fdhm-un_var1_monCat_1958-2013
MLS MERRA ERAi JRA55
Traj.MERRA Traj.ERAi Traj.JRA55 MLS
Entry level (83-hPa) Water Vapor Anomaly, comparing to MLS
3524 T. Wang et al.: Impact of temperature vertical structure on stratospheric H2O simulation
MLS traj.MER-T traj.GPS-T traj.MER-Twave
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
Weighted%by%AK%
MER-CPT GPS-CPT MER-CPTwave
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Col
d Po
int T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wv
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0H
2O A
nom
. (pp
mv)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.MER-Twave.AKtraj.GPS-T.AK
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Col
d Po
int T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wva) 83-hPa H2O Anomaly
b) Cold-Point Tropopause Anomaly
Figure 8. (a) Trajectory simulated H2O anomalies compared with the MLS observations; and (b) cold-point temperature anomalies fromthree temperature data sets. All time series are averaged over the deep tropics (18� N–18� S). All trajectory results in panel a are weightedby the MLS averaging kernels for fair comparison.
ences become smaller. Thus we conclude that using GPS-Tand MER-Twave decreases simulated stratospheric H2O byan average of⇠ 0.11 and 0.28 ppmv, respectively, accountingfor ⇠ 2.5 and 7% changes given typical stratospheric H2Oabundances of ⇠ 4 ppmv.It is important to point out that, despite these differences
in the absolute value of H2O, there is virtually no differencein the anomalies (residual from the average annual cycle).In Fig. 8a, we compare the time series of H2O anomaliesat 83 hPa from the three different trajectory runs weightedby the MLS averaging kernels to the MLS H2O observa-tions. Note that the interannual variations of approximately±0.5 ppmv in H2O are in good agreement with the interan-nual changes of about±1K in cold-point tropopause temper-atures (Fig. 8b) for all three different runs, further support-ing that the stratospheric entry level of H2O and cold-pointtropopause temperature are strongly coupled (e.g., Randel etal., 2004, 2006; Randel and Jensen, 2013). We also comparedtraj.MER-T and traj.MER-Twave over a longer period (1985–2013), and it shows almost no differences in interannual vari-ability either. Clearly, for studying the interannual variabilityof H2O, MERRA temperatures in coarse vertical resolutionare as good as temperatures at finer vertical resolution.
4 Summary
The dehydration of air entering the stratosphere largely de-pends on the cold-point temperature around the tropopause.This may not be represented accurately by reanalyses due totheir relatively coarse vertical resolution that reports coarsertemperature vertical structure. To investigate the impactsof this, we compare trajectory results from using standardMERRA temperatures at coarse model levels (traj.MER-T)to those using GPS temperatures in higher vertical resolution(traj.GPS-T) and those using adjusted MERRA temperatureswith finer vertical structures induced by waves (traj.MER-Twave).Driven by the same MERRA circulation, with a 100%
saturation assumption we find that on average traj.GPS-Tdries the stratospheric H2O prediction by ⇠ 0.1 ppmv andtraj.MER-Twave dries it by ⇠ 0.2–0.3 ppmv (Fig. 7a–b), ac-counting for at most⇠ 2.5% and 7.5% of changes given typ-ical stratospheric H2O abundances of⇠ 4 ppmv, respectively.However, despite the differences in H2O abundances, the in-terannual variability (residual from the mean annual cycle)exhibits virtually no differences due to the strong couplingbetween the interannual changes of stratospheric H2O andtropical cold-point tropopause temperatures (Fig. 8). There-fore, in terms of studying the interannual changes of strato-spheric H2O, we argue that reanalysis temperatures are moreuseful due to their long-term availability.
3524 T. Wang et al.: Impact of temperature vertical structure on stratospheric H2O simulation
MLS traj.MER-T traj.GPS-T traj.MER-Twave
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
Weighted%by%AK%
MER-CPT GPS-CPT MER-CPTwave
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0Co
ld P
oint
T A
nom
. (K
)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wv
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.MER-Twave.AKtraj.GPS-T.AK
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0Co
ld P
oint
T A
nom
. (K
)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wva) 83-hPa H2O Anomaly
b) Cold-Point Tropopause Anomaly
Figure 8. (a) Trajectory simulated H2O anomalies compared with the MLS observations; and (b) cold-point temperature anomalies fromthree temperature data sets. All time series are averaged over the deep tropics (18� N–18� S). All trajectory results in panel a are weightedby the MLS averaging kernels for fair comparison.
ences become smaller. Thus we conclude that using GPS-Tand MER-Twave decreases simulated stratospheric H2O byan average of⇠ 0.11 and 0.28 ppmv, respectively, accountingfor ⇠ 2.5 and 7% changes given typical stratospheric H2Oabundances of ⇠ 4 ppmv.It is important to point out that, despite these differences
in the absolute value of H2O, there is virtually no differencein the anomalies (residual from the average annual cycle).In Fig. 8a, we compare the time series of H2O anomaliesat 83 hPa from the three different trajectory runs weightedby the MLS averaging kernels to the MLS H2O observa-tions. Note that the interannual variations of approximately±0.5 ppmv in H2O are in good agreement with the interan-nual changes of about±1K in cold-point tropopause temper-atures (Fig. 8b) for all three different runs, further support-ing that the stratospheric entry level of H2O and cold-pointtropopause temperature are strongly coupled (e.g., Randel etal., 2004, 2006; Randel and Jensen, 2013). We also comparedtraj.MER-T and traj.MER-Twave over a longer period (1985–2013), and it shows almost no differences in interannual vari-ability either. Clearly, for studying the interannual variabilityof H2O, MERRA temperatures in coarse vertical resolutionare as good as temperatures at finer vertical resolution.
4 Summary
The dehydration of air entering the stratosphere largely de-pends on the cold-point temperature around the tropopause.This may not be represented accurately by reanalyses due totheir relatively coarse vertical resolution that reports coarsertemperature vertical structure. To investigate the impactsof this, we compare trajectory results from using standardMERRA temperatures at coarse model levels (traj.MER-T)to those using GPS temperatures in higher vertical resolution(traj.GPS-T) and those using adjusted MERRA temperatureswith finer vertical structures induced by waves (traj.MER-Twave).Driven by the same MERRA circulation, with a 100%
saturation assumption we find that on average traj.GPS-Tdries the stratospheric H2O prediction by ⇠ 0.1 ppmv andtraj.MER-Twave dries it by ⇠ 0.2–0.3 ppmv (Fig. 7a–b), ac-counting for at most⇠ 2.5% and 7.5% of changes given typ-ical stratospheric H2O abundances of⇠ 4 ppmv, respectively.However, despite the differences in H2O abundances, the in-terannual variability (residual from the mean annual cycle)exhibits virtually no differences due to the strong couplingbetween the interannual changes of stratospheric H2O andtropical cold-point tropopause temperatures (Fig. 8). There-fore, in terms of studying the interannual changes of strato-spheric H2O, we argue that reanalysis temperatures are moreuseful due to their long-term availability.
3524 T. Wang et al.: Impact of temperature vertical structure on stratospheric H2O simulation
MLS traj.MER-T traj.GPS-T traj.MER-Twave
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
Weighted%by%AK%
MER-CPT GPS-CPT MER-CPTwave
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Col
d Po
int T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wv
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0H
2O A
nom
. (pp
mv)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.MER-Twave.AKtraj.GPS-T.AK
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Col
d Po
int T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wva) 83-hPa H2O Anomaly
b) Cold-Point Tropopause Anomaly
Figure 8. (a) Trajectory simulated H2O anomalies compared with the MLS observations; and (b) cold-point temperature anomalies fromthree temperature data sets. All time series are averaged over the deep tropics (18� N–18� S). All trajectory results in panel a are weightedby the MLS averaging kernels for fair comparison.
ences become smaller. Thus we conclude that using GPS-Tand MER-Twave decreases simulated stratospheric H2O byan average of⇠ 0.11 and 0.28 ppmv, respectively, accountingfor ⇠ 2.5 and 7% changes given typical stratospheric H2Oabundances of ⇠ 4 ppmv.It is important to point out that, despite these differences
in the absolute value of H2O, there is virtually no differencein the anomalies (residual from the average annual cycle).In Fig. 8a, we compare the time series of H2O anomaliesat 83 hPa from the three different trajectory runs weightedby the MLS averaging kernels to the MLS H2O observa-tions. Note that the interannual variations of approximately±0.5 ppmv in H2O are in good agreement with the interan-nual changes of about±1K in cold-point tropopause temper-atures (Fig. 8b) for all three different runs, further support-ing that the stratospheric entry level of H2O and cold-pointtropopause temperature are strongly coupled (e.g., Randel etal., 2004, 2006; Randel and Jensen, 2013). We also comparedtraj.MER-T and traj.MER-Twave over a longer period (1985–2013), and it shows almost no differences in interannual vari-ability either. Clearly, for studying the interannual variabilityof H2O, MERRA temperatures in coarse vertical resolutionare as good as temperatures at finer vertical resolution.
4 Summary
The dehydration of air entering the stratosphere largely de-pends on the cold-point temperature around the tropopause.This may not be represented accurately by reanalyses due totheir relatively coarse vertical resolution that reports coarsertemperature vertical structure. To investigate the impactsof this, we compare trajectory results from using standardMERRA temperatures at coarse model levels (traj.MER-T)to those using GPS temperatures in higher vertical resolution(traj.GPS-T) and those using adjusted MERRA temperatureswith finer vertical structures induced by waves (traj.MER-Twave).Driven by the same MERRA circulation, with a 100%
saturation assumption we find that on average traj.GPS-Tdries the stratospheric H2O prediction by ⇠ 0.1 ppmv andtraj.MER-Twave dries it by ⇠ 0.2–0.3 ppmv (Fig. 7a–b), ac-counting for at most⇠ 2.5% and 7.5% of changes given typ-ical stratospheric H2O abundances of⇠ 4 ppmv, respectively.However, despite the differences in H2O abundances, the in-terannual variability (residual from the mean annual cycle)exhibits virtually no differences due to the strong couplingbetween the interannual changes of stratospheric H2O andtropical cold-point tropopause temperatures (Fig. 8). There-fore, in terms of studying the interannual changes of strato-spheric H2O, we argue that reanalysis temperatures are moreuseful due to their long-term availability.
3524 T. Wang et al.: Impact of temperature vertical structure on stratospheric H2O simulation
MLS traj.MER-T traj.GPS-T traj.MER-Twave
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
Weighted%by%AK%
MER-CPT GPS-CPT MER-CPTwave
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0Co
ld P
oint
T A
nom
. (K
)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wv
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.MER-Twave.AKtraj.GPS-T.AK
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0Co
ld P
oint
T A
nom
. (K
)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wva) 83-hPa H2O Anomaly
b) Cold-Point Tropopause Anomaly
Figure 8. (a) Trajectory simulated H2O anomalies compared with the MLS observations; and (b) cold-point temperature anomalies fromthree temperature data sets. All time series are averaged over the deep tropics (18� N–18� S). All trajectory results in panel a are weightedby the MLS averaging kernels for fair comparison.
ences become smaller. Thus we conclude that using GPS-Tand MER-Twave decreases simulated stratospheric H2O byan average of⇠ 0.11 and 0.28 ppmv, respectively, accountingfor ⇠ 2.5 and 7% changes given typical stratospheric H2Oabundances of ⇠ 4 ppmv.It is important to point out that, despite these differences
in the absolute value of H2O, there is virtually no differencein the anomalies (residual from the average annual cycle).In Fig. 8a, we compare the time series of H2O anomaliesat 83 hPa from the three different trajectory runs weightedby the MLS averaging kernels to the MLS H2O observa-tions. Note that the interannual variations of approximately±0.5 ppmv in H2O are in good agreement with the interan-nual changes of about±1K in cold-point tropopause temper-atures (Fig. 8b) for all three different runs, further support-ing that the stratospheric entry level of H2O and cold-pointtropopause temperature are strongly coupled (e.g., Randel etal., 2004, 2006; Randel and Jensen, 2013). We also comparedtraj.MER-T and traj.MER-Twave over a longer period (1985–2013), and it shows almost no differences in interannual vari-ability either. Clearly, for studying the interannual variabilityof H2O, MERRA temperatures in coarse vertical resolutionare as good as temperatures at finer vertical resolution.
4 Summary
The dehydration of air entering the stratosphere largely de-pends on the cold-point temperature around the tropopause.This may not be represented accurately by reanalyses due totheir relatively coarse vertical resolution that reports coarsertemperature vertical structure. To investigate the impactsof this, we compare trajectory results from using standardMERRA temperatures at coarse model levels (traj.MER-T)to those using GPS temperatures in higher vertical resolution(traj.GPS-T) and those using adjusted MERRA temperatureswith finer vertical structures induced by waves (traj.MER-Twave).Driven by the same MERRA circulation, with a 100%
saturation assumption we find that on average traj.GPS-Tdries the stratospheric H2O prediction by ⇠ 0.1 ppmv andtraj.MER-Twave dries it by ⇠ 0.2–0.3 ppmv (Fig. 7a–b), ac-counting for at most⇠ 2.5% and 7.5% of changes given typ-ical stratospheric H2O abundances of⇠ 4 ppmv, respectively.However, despite the differences in H2O abundances, the in-terannual variability (residual from the mean annual cycle)exhibits virtually no differences due to the strong couplingbetween the interannual changes of stratospheric H2O andtropical cold-point tropopause temperatures (Fig. 8). There-fore, in terms of studying the interannual changes of strato-spheric H2O, we argue that reanalysis temperatures are moreuseful due to their long-term availability.
3524 T. Wang et al.: Impact of temperature vertical structure on stratospheric H2O simulation
MLS traj.MER-T traj.GPS-T traj.MER-Twave
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0H
2O A
nom
. (pp
mv)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0H
2O A
nom
. (pp
mv)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.GPS-T.AK traj.MER-Twave.AK
Weighted%by%AK%
MER-CPT GPS-CPT MER-CPTwave
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Col
d Po
int T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wv
2007 2008 2009 2010 2011 2012 2013Time
-1.0
-0.5
0.0
0.5
1.0
H2O
Ano
m. (
ppm
v) 100 hPa
-1.0
-0.5
0.0
0.5
1.0H
2O A
nom
. (pp
mv)
2007 2008 2009 2010 2011 2012 2013
83 hPa
Time Series: lon[0,360],lat[-18,0,0,18],,vert[100,31],time[2007.042,2013.958],X00: monthly_gridded_H2O_lon360_monCat_2004-2013.nc; X01: 140415_s100_i370_mthd_inj1sav3_MERwindT_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X02: 140416_s100_i370_mthd_inj1sav3_MERwindTwave_day_dhAll_fdhm-un_fineV_3isob_mlsAK_monCat_2006-2013.nc; X03: 140417_s100_i370_mthd_inj1sav3_MERwind-GPST_day
MLS traj.MER-T.AK traj.MER-Twave.AKtraj.GPS-T.AK
2007 2008 2009 2010 2011 2012 2013Time
-2.0
-1.0
0.0
1.0
2.0
Cold
Poi
nt T
Ano
m. (
K)
2007 2008 2009 2010 2011 2012 2013
Time Series: lon[0,360],lat[-18,0,0,18],,time[2007.042,2013.958],X00: GPS_xtropo_day2mon_2007-2013.nc; X01: MER_xtropo_day2mon_1979-2013.nc; X02: MER_Norminal_wv_xtropo_day2mon_2007-2013.nc;
GPS MER MER-wva) 83-hPa H2O Anomaly
b) Cold-Point Tropopause Anomaly
Figure 8. (a) Trajectory simulated H2O anomalies compared with the MLS observations; and (b) cold-point temperature anomalies fromthree temperature data sets. All time series are averaged over the deep tropics (18� N–18� S). All trajectory results in panel a are weightedby the MLS averaging kernels for fair comparison.
ences become smaller. Thus we conclude that using GPS-Tand MER-Twave decreases simulated stratospheric H2O byan average of⇠ 0.11 and 0.28 ppmv, respectively, accountingfor ⇠ 2.5 and 7% changes given typical stratospheric H2Oabundances of ⇠ 4 ppmv.It is important to point out that, despite these differences
in the absolute value of H2O, there is virtually no differencein the anomalies (residual from the average annual cycle).In Fig. 8a, we compare the time series of H2O anomaliesat 83 hPa from the three different trajectory runs weightedby the MLS averaging kernels to the MLS H2O observa-tions. Note that the interannual variations of approximately±0.5 ppmv in H2O are in good agreement with the interan-nual changes of about±1K in cold-point tropopause temper-atures (Fig. 8b) for all three different runs, further support-ing that the stratospheric entry level of H2O and cold-pointtropopause temperature are strongly coupled (e.g., Randel etal., 2004, 2006; Randel and Jensen, 2013). We also comparedtraj.MER-T and traj.MER-Twave over a longer period (1985–2013), and it shows almost no differences in interannual vari-ability either. Clearly, for studying the interannual variabilityof H2O, MERRA temperatures in coarse vertical resolutionare as good as temperatures at finer vertical resolution.
4 Summary
The dehydration of air entering the stratosphere largely de-pends on the cold-point temperature around the tropopause.This may not be represented accurately by reanalyses due totheir relatively coarse vertical resolution that reports coarsertemperature vertical structure. To investigate the impactsof this, we compare trajectory results from using standardMERRA temperatures at coarse model levels (traj.MER-T)to those using GPS temperatures in higher vertical resolution(traj.GPS-T) and those using adjusted MERRA temperatureswith finer vertical structures induced by waves (traj.MER-Twave).Driven by the same MERRA circulation, with a 100%
saturation assumption we find that on average traj.GPS-Tdries the stratospheric H2O prediction by ⇠ 0.1 ppmv andtraj.MER-Twave dries it by ⇠ 0.2–0.3 ppmv (Fig. 7a–b), ac-counting for at most⇠ 2.5% and 7.5% of changes given typ-ical stratospheric H2O abundances of⇠ 4 ppmv, respectively.However, despite the differences in H2O abundances, the in-terannual variability (residual from the mean annual cycle)exhibits virtually no differences due to the strong couplingbetween the interannual changes of stratospheric H2O andtropical cold-point tropopause temperatures (Fig. 8). There-fore, in terms of studying the interannual changes of strato-spheric H2O, we argue that reanalysis temperatures are moreuseful due to their long-term availability.
Our Definition: Average time it takes for parcels to cross a given height for the very first time.(different from chemical residence time or life time)
• Trajectory is a useful tool to (indirectly) validate and quantify differences reanalyses;
• Vertical structures of dehydration show large differences due to differences in CPT; but the interannual variability of CPT are basically the same, so as H2O predicted;
• Residence time is anti-correlated with vertical upwelling. ERA interim has the strongest upwelling, resulting ~2 months of residence time within the TTL since passing the tropopause; MERRA, JRA55, and CFSR indicates of at least 3 months;
• Using ERAi, JRA55, and CFSR produces much younger air than using MERRA. In mid-latitude the 4-5 years old air is close to what observed fom SF6 and CO2 (personal communication with Eric Ray)