View
215
Download
1
Category
Tags:
Preview:
Citation preview
1
Evaluation of the Chemical Mechanismwithin CCMs using a constrained
Photochemical Steady State (PSS) Model
Ross Salawitch1, Ben Johnson1, Tim Canty1, Doug Kinnison2, Martyn Chipperfield3
1University of Maryland; 2NCAR; 3University of Leeds
Will be delighted to work with CCM Investigators to help resolve discrepancies:
rjs@atmos.umd.edu
• Work relies entirely on the *REF-B1*T3I* files submitted to the archive
• Material posted at http://www.atmos.umd.edu/~rjs/ccmval/
• CCM community emailed above URL (and description of this work) week prior to this meeting
CCMVal Workshop, Toronto
9 September 2009 Update
2
Evaluation of the Chemical Mechanismwithin CCMs using a constrained
Photochemical Steady State (PSS) Model
• This work builds on methods developed for NASA Models and Measurements I & II evaluations
included mainly 2D models but some 3D models
• Process for evaluation of chemical mechanism:
designed during breakouts at “Process-orientated validation of coupled- chemistry climate models” workshop, Grainau, Germany, Nov, 2003
described in Eyring et al. (BAMS, 2005)
facilitated by design of the T3I file specifications (a lot of email and phone calls between Doug and Ross, and between Doug and many others)
• First public “unveiling” of PSS Chem Mech Eval, due to confluence of:
availability of “core dumps” (T3I) files from many models sufficient lead time, commitment, and “gentle prodding” to transfer more than 600 Gb of info to our home computers and develop software to interface files from each group (which differ in subtle ways!) to our PSS model
has been “straightforward” but certainly not “easy”
3
Chemical Mechanism
Chemical Mechanism
Computes: O(3P), O(1D), OH, HO2,
NO, NO2, ClO, BrO, etc
Physical properties of atmosphere:
p, T, aerosol surface area, overhead O3
Solar Declination, Latitude, Longitude
Long-lived radical precursors:
O3, H2O, CH4, CO, NOy, Cly, Bry
Once we know: O(3P), O(1D), OH, HO2,
NO, NO2, ClO, BrO, etc
Compute:
dO3/dt terms, etc
4
Chemical Mechanism
Chemical Mechanism
Computes: O(3P), O(1D), OH, HO2,
NO, NO2, ClO, BrO, etc
Physical properties of atmosphere:
p, T, aerosol surface area, overhead O3
Solar Declination, Latitude, Longitude
Long-lived radical precursors:
O3, H2O, CH4, CO, NOy, Cly, Bry
Contend: If all models are using “JPL 2006” kinetics, which defines their chemical mechanism, then all models should compute the same values of O(3P), OH, NO, ClO, BrO, etc for the same specification of long-lived radical precursors & physical properties
5
Chemical Mechanism
Chemical Mechanism
Computes: O(3P), O(1D), OH, HO2,
NO, NO2, ClO, BrO, etc
Physical properties of atmosphere:
p, T, aerosol surface area, overhead O3
Solar Declination, Latitude, Longitude
Long-lived radical precursors:
O3, H2O, CH4, CO, NOy, Cly, Bry
Contend: If all models are using “JPL 2006” kinetics, which defines their chemical mechanism, then all models should compute the same values of O(3P), OH, NO, ClO, BrO, etc for the same specification of long-lived radical precursors & physical properties
How to test? In Grainau (Nov 2003), we stated that definition of long-lived radical precursors and physical properties of the atmosphere from zonal monthly mean sampling of a CCM could be used as input to a well established chemical model, which if run in photochemical steady state (PSS) over a full diel cycle (integrated PL =0 for all species), would result in 24 hr avg’d profiles of radicals that should closely approximate the zonal monthly mean profiles of radicals from the CCM
6
Chemical Mechanism
Chemical Mechanism
Computes: O(3P), O(1D), OH, HO2,
NO, NO2, ClO, BrO, etc
Physical properties of atmosphere:
p, T, aerosol surface area, overhead O3
Solar Declination, Latitude, Longitude
Long-lived radical precursors:
O3, H2O, CH4, CO, NOy, Cly, Bry
Contend: If all models are using “JPL 2006” kinetics, which defines their chemical mechanism, then all models should compute the same values of O(3P), OH, NO, ClO, BrO, etc for the same specification of long-lived radical precursors & physical properties
How to test? In Boulder (Oct 2005), we discussed the file specification needed to carry out this task
Since then, Doug Kinnison has been instrumental in implementing the specifications of the T3I files needed to carry out these comparisons
7
PSS Model Comparisons: T3I files (shown in Toronto)
Model Comment
CAM 3.5 Highest model level is 3.55 hPa†
CCSR NIES All files present
CMAM Missing sad_sulf*; ClO on different eta grid
CNRM-ACM All files present
GEOS CCM Missing sad_sulf*
MRI Missing sad_sulf*
UMSLIMCAT Missing sad_sulf*, O1D, & O3P
BrO missing; Brx saved as Bry
WACCM Years 1990 to 1994 not posted on BADC; files for these years obtained privately from Doug Kinnison
† O3 and T from WACCM above 3.55 hPa used for J value computation* sad_sulf from WACCM used for analysis
Sorry our color schemes are not yet consistent with the prescribedcolors we were asked to use.
8
PSS Model Comparisons: T3I files (post Toronto)
Model Comment
UMUKCA-METO
Missing sad_sulf* & O1D
Mass mixing ratio written rather than volume mixing ratio
AMTRAC All files present for PSS comp days only
EMAC Missing sad_sulf*
Output written in hybrid coordinate system
SOCOL Missing sad_sulf*, O1D, & O3P
ULAQ All files present
LMDZrepro Missing O1D & O3P
Niwa_SOCOL Not enough files available to conduct comparisons
E39CA No T3I files yet posted
UMUKCA-UCAM
No T3I files yet posted
9
A Few More Details• Focus on periods of time when atmospheric observations are available
• Will examine:
a) profiles of radical precursors
b) “tracer – tracer” relations of radical precursors
c) CCM vs PSS radical profiles
For radicals, the comparison is between 24 hour average output of the PSS model versus the ZMM of the CCM model
If the chemistry is properly represented in both models, this comparison should look very good (but will not be perfect!)
Powerful method to diagnose representation of fast chemistry in models
• Initial focus on 35N, Sept 1993:a) time of high aerosol loadingb) atmosphere sampled by a high altitude balloon flight that resulted in many papers documenting atmospheric composition (e.g., Osterman et al., GRL, 1997)
• Have also examined 22N, Feb 1996:a) measurement of HOx, NOx, ClO & precursors in the tropical UT/LS regionb) focus of Wennberg et al. (Science, 1998) and numerous other papers
10
Aerosol Surface Area• Many modeling groups did not provide sulfate aerosol surface area, which was prescribed for this run
• Value had been requested (variable sad_sulf) so that we could be sure to run the
• If sad_sulf is missing, then profiles of sad_sulf vs pressure from WACCM are used to interpolate onto the pressure grid of each model (the prescribed climatology are time series at various altitudes; since altitude was not output (or given) by the various modeling groups, use of this prescribed climatology is a challenge
• For groups that did provide sad_sulf, we compare to the climatology by:
a) calculating geometric altitude by integrating the model pressure/temperature (there is some ambiguity in this calculation, such as surface pressure, surface topography, and whether or not g is allowed to vary with height)
b) compare to the climatological time series at 5 altitudes
11
Aerosol Surface Area Time Series• CAM 3.5 sulfate surface area compares will with prescribed climatology
12
Aerosol Surface Area Time Series• CCSR NIES sulfate surface area compares will with prescribed climatology
13
Aerosol Surface Area Time Series• CNRM-ACM sulfate surface area compares will with prescribed climatology
14
Aerosol Surface Area Time Series• ULAQ sulfate surface area appears to be quite different than prescribed climatology
15
Aerosol Surface Area Time Series• WACCM sulfate surface area compares will with prescribed climatology
16
Aerosol Surface Area Time Series• LMDZrepro sulfate surface area is also quite different than prescribed climatology
17
Aerosol Surface Area• LMDZrepro sulfate surface area is also quite different than prescribed climatology
• AMTRAC provided profiles of SA just for days of interest, and hence we can not produce time series plots for this model
• GEOS CCM ran using background aerosol as a function of latitude and altitude; we could produce time series plots using this information, but it does not seem
18
Aerosol Surface Area Profiles• AMTRAC sulfate surface area only provided for two days• Profiles differ somewhat from prescribed climatology
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
19
Aerosol Surface Area Profiles• CAM sulfate surface area agrees quite well with prescribed climatology for the two times and places for which we will conduct PSS comparisons
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
Analysis restricted to modelvalues between tropopause(which we determine foreach model) and 30 km
20
Aerosol Surface Area Profiles• CCSR NIES sulfate surface area looks very reasonable
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
21
Aerosol Surface Area Profiles• CNRM-ACM sulfate surface area higher than prescribed climatology in lower stratos• Strangely, same value of g found for both cases
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
22
Aerosol Surface Area Profiles• GEOS CCM was run using background surface area, resulting in low g for Sept 1993 • Important we use proper sulfate surface area profile for in the PSS comparisons
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
23
Aerosol Surface Area Profiles• LMDZrepro sulfate surface area profile is less than climatology in lower stratos• Unusual shape of sulfate surface area profile for Feb 1996 case
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
24
Aerosol Surface Area Profiles• ULAQ sulfate surface area is quite larger than prescribed climatology in Sept 1993, leading to a very low g value
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
25
Aerosol Surface Area Profiles• WACCM sulfate surface area profile shifted slightly wrt altitude compared to the climatology, perhaps reflecting difficulty of using data on a geometric altitude grid
NCCM CLIMATOLOGY
0 Sulf_sad
abs(Sulf_sad Sulf_sad )1g = 1
N 3
where we have: a) integrated p vs T of each CCM model to arrive at an estimate of geometric altitude because the Sulf_sad climatology is given only as a function of altitude and latitude b) estimated Sulf_sad based on the sum of differences between adjacent 5 latitude bins of the
climatology and the difference of Sulf_sad for a 0.2 km uncertainty in altitude (error bars represent Sulf_sad)
c) floored g at 0 for each altitude (i.e., g is not allowed to go negative) (this follows Waugh & Eyring, ACP, 2008)
26
Aerosol Surface Area Summary
Differences in prescribed sulfate surface area and values of the variable sulf_sad in the *REF-B1*T3I* files submitted by AMTRAC, GEOSCCM, LMDZrepro, and ULAQ underscores the importance of other groups submitting, to the archive, values of sulf_sad for the PSS comparison.
At the present time, we lack sulf_sad from the following groups: CMAM, EMAC, GEOS CCM, MRI, SOCOL, UMUKCA-METO, and UMSLIMCAT
In the absence of sulf_sad submitted by various groups, we are assuming the climatology applies, but this may be an erroneous assumption, as is the case for AMTRAC, GEOSCCM, LMDZrepro, and ULAQ
Submission of files for sulf_sad also allows us to test whether we are properly interpreting date in the submitted files: there are “issues” here due to some model’s use of a 360 day year, whether or not leap years are included, etc
27
First Set of Comparisons• Tracer profile plots to follow for ~35N, Sept 1993 for 13 models
• Results presented in same order as at Toronto meeting (first 8 models) and in the order that calculations were conducted based on receipt of files (last 6 models)
• Sorry that colors do not yet match requested scheme
28
WACCM
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
29
CAM 3.5
Bry low
Model lid at 3.5 hPa might be affectingtransport for species such as N2O
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
30
CMAM
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
31
CNRM-ACM
Bry present in troposphereCly present in troposphere
ColdTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
32
GEOS CCM
LowTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Cly low Bry low
33
MRI
Bry high
Bry present in troposphere
LowTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
34
UMSLIMCAT
LowTropopause
Model Output Not AvailableFor p > 170 hPa
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Bry high: a VSL bromocarbon
source of ~6 ppt was used in this run
35
CCSR NIES
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Bry very high
Cly
high
Lots of Bry present in trop
36
UMUKCA-METO
Bry present in troposphereCly present in troposphere
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
37
AMTRAC
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Cly a litte lowBry somewhat low
38
EMAC
LowTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Bry low
39
SOCOL
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Bry very high
Cly
high
Some Bry present in trop
40
ULAQ
Bry present in troposphere
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Unusual O3
profile
41
LMDZrepro
Bry present in troposphere
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Sept 1993 Upper Strat Bry Sept 1993
Bry high: looks like a
VSL bromocarbon source may have been
used in this run
42
Second Set of Comparisons• Tracer vs tracer plots to follow for ~35N, Sept 1993 for 13 models
• Results presented in same order as at Toronto meeting (first 8 models) and in the order that calculations were conducted based on receipt of files (last 6 models)
• Sorry that colors do not yet match requested scheme
43
WACCM
Woodbridge and Wamsley RelnsScaled to Sept 1993
44
WACCM
Bry differs because Wamsley Reln considers CH2Br2, which is known
to reach the stratosphere
45
CAM 3.5
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Bry low
Model lid at 3.5 hPa might be affectingtransport for species such as N2O
Woodbridge and Wamsley RelnsScaled to Sept 1993
46
CMAM
HTOT slightly low
Bry differs because Wamsley Reln considers CH2Br2, which is known
to reach the stratosphere
47
CNRM-ACM
HTOT high: puzzling due tovery cold tropopause
Cly a bit high throughout stratosphere
48
GEOS CCM
Bry slightly low, even considering neglect of CH2Br2Cly slightly low
49
MRI
Bry very highCly slightly high
50
UMSLIMCAT
Bry high: a VSL bromocarbon source of ~6 ppt was used in this run
51
CCSR NIESScale Change
Bry very high(may actually becorrect, however )
Cly high
HTOT very low
Scale Change
O3 high
52
UMUKCA-METO
O3 vs N2Oa bit different
than data;all other relns
look fine
53
AMTRAC
Cly a bit low Bry a bit low
54
EMAC
Bry low, even considering neglect of CH2Br2
HTOT slightly low
55
SOCOL
Cly high
Bry very high(may actually becorrect, however )
O3 vs N2O“shifted”
56
ULAQ
All other tracer relations look good
O3 vs N2O quitedifferent than data
(and most other models)
57
LMDZrepro
Bry high: looks like a VSL bromocarbon was used in this run
58
Third Set of Comparisons• Radical (CCMVal) and radial (PSS) profile plots for ~35N, Sept 1993 for 13 models
• Results presented in same order as at Toronto meeting (first 8 models) and in the order that calculations were conducted based on receipt of files (last 6 models)
• Sorry that colors do not yet match requested scheme
59
Third Set of Comparisons• As promised in Toronto, we have now computed “error bar” for PSS model values of O(3P), O(1D), HOx, NOx/NOy, ClO/Cly, and BrO/Bry by propagating the variance,
about the zonal monthly mean, of inputs to the PSS model (i.e., T, O3, H2O,
etc derived from each CCM model black error bars on following plots (PSS)
• We also computed the variance of the zonal mean of the CCM profiles of O(3P), O(1D), HOx, NOx/NOy, ClO/Cly, and BrO/Bry based on analysis of output for various
days in the month used to define the zonal mean colored error bars on following plots (CCM)
• When error bars overlap, or nearly overlap, as is often the case in the lowermost stratosphere, this means the way we sampled the CCM might be the cause of the difference between PSS and CCM values
60
Third Set of Comparisons• We then use PSS and CCM in the computation of g, which is shown for each panel
of each plot (this took a fair amount of effort to setup and formulate ):
NCCM PSS
0 CCM PSS
abs(Mixing Ratio Mixing Ratio )1g = 1
N 3 ( )
where we have: a) floored g at 0 for each altitude (i.e., g is not allowed to go negative) b) PSS floored at 5% of baseline value from PSS model at all altitudes (i.e, in the calculation of g, we
assume the PSS model is in error by at least 5% at all levels … this is not represented in our plots, but is represented in the computation of g and the log files we have posted
c) evaluated g between: the tropopause (which we determine) and 1 hPa for all species except BrO/Br y, the tropopause (which we determine) and 5 hPa for all BrO/Bry (there seem to be
issues with model representation of BrO at high altitude, where it simply does not matter, for BrO we focus on only the altitudes that matter
d) evaluated g for JPL 2002 kinetics and JPL 2006 kinetics for O(1D) and BrO/Bry, and propagate the higher value of g (many CCMs clearly did not add the O+BrONO2 reaction, new for JPL 2006, into their chemical mechanism; since we can account for this nuance, we will show the best value of g, but will indicate on the metric summary which kinetic set was used)
(again, this builds on the work of Waugh & Eyring, ACP, 2008)
61
WACCM
As good as it gets ! If we are actually solving the same set of chemical reactions,as we aspire, then all comparisons should look nearly this good
Values of g are not unity, however, because the multi-decadal log plots obscure small differences
There is ared line
under theblack
There is ared line
under theblack
There is ared line
under theblack
TropopausePSS run only
above this level
62
CAM 3.5
SuspectBrONO2+O rxn,new for JPL-06,
has not beenadded NOx a little low
compared to PSS
Diff in ClO consistentwith diff in NOx
Looks good. NOx a little low. Would like to determine whether BrONO2+O rxn is included in model
Diff in BrO consistent
with diff in NOx
For this model, neglectedhighest altitude pointin computation of g
due to uncertainties inoverhead conditions
63
CMAM
Have confirmedBrONO2+O rxn,new for JPL-06,
has not beenincluded !
Looks great! Sulfate SA file from CMAM is not yet submitted but no reason to suspect any difficulty.
64
CNRM-ACM
NOx suppressed,perhaps due to cold T.Not picked up by PSS
ClOx slightly activatedperhaps due to PSC like T
Looks likeBrONO2+Onot used
Looks good; differences in lowermost stratosphere likely due to response in CNRM-ACM to very cold T.Would like to determine whether BrONO2+O rxn is included in model.
65
GEOS CCM
ClO slightlyhigh
Looks good. However, the chemical mechanism in this model seems to over-estimate ClO in theupper stratosphere.
Looks likeBrONO2+Onot used
BrO not imp.at high alt
Very lowtropopause
in this model
66
MRI
HOx low
NOx high
ClO high
The chemical mechanism used in this model seems to differ from PSS in numerous important manners.Would like to run with MRI sulfate surface area! However, this will not affect comparisons at high altitude.
Scale Change
67
UMSLIMCAT
O(3P) filenot submitted
O(1D) filenot submitted
BrO file not submitted;have to compare BrO+Br
NOx slightly lower than PSS,perhaps due to surface area
mismatch ClO offset consistent with NOx
HOx offset consistentwith NOx
Looks like BrONO2+Ohas been included
Looks good. Would like to examine O(3P), O(1D), BrO,and run with UMSLIMCAT sulfate surface area.
Scale Change
Difference inconseq.;
BrO does notmatter
at high altitude
68
CCSR NIES
Looks likeBrONO2+Onot used
NOx highClO activation due to
very cold T, picked up by PSS !
HOx consistentwith NOx
Looks reasonable except for O1D. NOx is higher than PSS for most altitudes. Differences inHOx consistent with NOx. Looks like BrONO2+O not included.
Scale Change
69
UMUKCA-METO
NOx very high ClOx quite low,consistent with high NOx
Looks likeBrONO2+Onot used
The chemical mechanism used in this model seems to differ from PSS in important manners throughout the lower stratosphere. Would like to run with UMUKCA-METO sulfate surface area,
which could affect interpretation in regions of large discrepancy.
HOx very low,consistent with
forcing from NOx
70
AMTRAC
Looks likeBrONO2+Onot used
Looks great except for some differences in ClO in the upper stratosphereand the likely neglect of BrONO2+O.
71
EMAC
Looks likeBrONO2+Onot used
Looks great. Suspect BrONO2+O has not been considered in EMAC. Also, appearsO(1D) is found using JPL 2002 kinetics.
72
SOCOL
The chemical mechanism used in this model seems to differ from PSS in a somewhat important manner throughoutthe lower stratosphere. Would like to run with SOCOL sulfate surface area, which could affect interpretation
of results in region of large discrepancy, and also compare to O(3P) and O(1D).
NOx lowClO high, some activation
at tropopuse
HOx spot on.Somewhat surprising
given mismatchfor NOx
BrO sig lowerthan PSS
O(3P) filenot submitted
O(1D) filenot submitted
73
ULAQ
The chemical mechanism used in this model seems to differ from PSS in a somewhat important mannerthroughout the stratosphere. Differences in ClOx mirror those of NOx
(similar magnitude, opposite sign), which is reassuring.
NOx low
ClO quite low in upper strat
HOx close.A bit surprising
given mismatchesfor O(1D) & NOx
BrO sig lowerthan PSS
O(1D) low
NOx high
ClOx high
Scale Change
74
LMDZrepro
Looks likeBrONO2+O
was used !!!
Chemical mechanism looks just like the PSS simulation. Impressive .Would like to see O(3P) and O(1D) for completeness.
O(3P) filenot submitted
O(1D) filenot submitted
75
The Metric Plot• This plot show, in a single image, the values of g from all of the prior plots• So, can ignore previous 74 slides and focus just on this one
We’ve decided to use the same colorfor 0.8 to 0.9 as used for 0.9 to 1.0,because we feel g > 0.8 is an excellentevaluation and there is little scientificmeaning in splitting hairs once thisthreshold has been reached
Please see g*log files posted atwww.atmos.umd.edu/~rjs/ccmval/g_valuesfor a record of how the values of gshown in this metric plot were tabulated.
76
New Comparison Plots• Following series of plots show profiles of sulfate SA, HNO3, N2O5, Lifetime of NOx •and HCl (for diagnostic purposes from PSS), ClNO3, and HCl
• Error bars are variability of the zonal mean from CCM model; error bars are large for N2O5 because this species varies diurnally. Nonetheless, the 24 hr avg value of N2O5 from PSS goes right through the middle of the zonal monthly mean value for many of the CCM models, as it is generally expected to do if we are correctly implementing the same chemical mechanism
• These plots provided additional insight and “support” for inconsistencies seen in prior plots for a few models
• Have resisted extensive written comments on the plots that follow
77
WACCM
78
CMAM
79
CNRM-ACM
80
GEOS CCM
81
MRI
Significant Differences
82
UMSLIMCAT
83
CCSR NIES
84
UMUKCA-METO
Signif differences
85
AMTRAC
Sulfate SA profileslook quite different(see earlier slides
comparing to inputclimatology)
Diffs consistentwith those for
ClO
86
EMAC
87
SOCOL
Diffs consistentwith those for
ClO & NOx
88
ULAQ
Yikes! Not sure whatis happening here(see earlier slides
comparing to inputclimatology)
Significant Differences
89
LMDZrepro
Yikes!!! Not sure what isgoing on here. But, as shown
in the SA time series plots,sulf_sad from this model quite
different than prescribedclimatology
All other comparisons exceptionally good
90
Next Steps• Compute “error bar” for PSS model values of O(3P), O(1D), HOx, NOx/NOy,
ClO/Cly, and BrO/Bry by propagating variance, about the zonal monthly
mean, of inputs to the PSS model that are derived from each CCM model:
i.e., we will formally evaluate how computed variance T, sulfate SA, O3,
H2O, CH4, CO, NOy, Cly, Bry impacts the computed radicals
• Update comparisons as:
additional models submit T3I files
additional files (i.e., sulfate SA, O3P, O1D) are submitted for the nine models that have submitted files
• Run additional cases:
Feb 1996 (NASA STRAT Flights) has been run (included in this PPT)
we are open to suggestion for periods of particular interest
• Submit PSS J values to photocomp !!!
• This is a work in progress … hopefully of some value to CCMVal
We will gladly provide detailed PSS model output, upon request, to groupswho are interested in diagnosing the reasons for differences betweenradical fields computed by their CCM and the PSS model
We’ll develop “grades” based on these comparisons upon request
91
Comparisons for 22N, Feb 1996 to follow
Note: nothing that follows has yet been updated since the Toronto meeting Updates to be provided soon.
We intend to compute g_values for this comparison, whichis essential to search for consistency (or lack thereof)with results found for the 35N, Sept 1993 comparisons
92
WACCM
CCM Tropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
93
CAM 3.5
CCM Tropopause
Model lid at 3.5 hPa might be affectingtransport for species such as N2O
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
94
CMAM
CCM Tropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
95
CNRM-ACM
CCM Tropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
96
GEOS CCM
CCM Tropopause
LowTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
97
MRI
CCM Tropopause
LowTropopause
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
98
UMSLIMCAT
CCM Tropopause
LowTropopause
Model Output Not AvailableFor p > 170 hPa
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Upper Strat Cly Feb 1996
99
WACCM
Woodbridge and Wamsley RelnsScaled to Feb 1996
100
WACCM
Bry differs because Wamsley Reln considers CH2Br2, which is known
to reach the stratosphere
101
CAM 3.5
Error BarsData: 1 std dev, total meas uncertaintyModel: 1 std dev, about zonal mean
Bry low
Model lid at 3.5 hPa might be affectingtransport for species such as N2O
Woodbridge and Wamsley RelnsScaled to Feb 1996
102
CMAM
HTOT slightly low
Bry differs because Wamsley Reln considers CH2Br2, which is known
to reach the stratosphere
103
CNRM-ACM
HTOT high: puzzling due tovery cold tropopause
Woodbridge and Wamsley RelnsScaled to Feb 1996
104
GEOS CCM
Bry slightly low, even considering neglect of CH2Br2
105
MRI
Bry very highCly slightly high
106
UMSLIMCAT
Bry high: a VSL bromocarbon source of ~6 ppt was used in this run
107
WACCM
Some differences in LMS/UT. Unclear why O(1D) would differ.NOx differs by somewhat more than expected, which bears further investigation.
There is ared line
under theblack
There is ared line
under theblack
There is ared line
under theblack
108
CAM 3.5
SuspectBrONO2+O rxn,new for JPL-06,
has not beenadded
Diff in ClO consistentwith diff in NOx
Looks good. Would like to examine O(3P)and determine whether BrONO2+O rxn is included in model
Diff in BrO consistent
with diff in NOx
NOx lowcompared to PSS
109
CMAM
Have confirmedBrONO2+O rxn,new for JPL-06,
has not beenincluded !
Looks great! Would like to examine O(3P) and O(1D)and determine whether BrONO2+O rxn is included in model
NOx lowcompared to PSS
110
CNRM-ACM
NOx in near zerodue to PSC like T
relevant chemistryClOx in PSS activated
due to PSC like TLooks likeBrONO2+O
rxn not added
Looks good; would like to determine whether “PSC chemistry” has been “turned off” for extra-polar regionsand whether BrONO2+O rxn is included in model.
Scale Change
111
GEOS CCM
ClO slightlyhigh
BrO slightlylow
Looks good. However, the chemical mechanism in this model seems to over-estimate ClO and under-estimate BrO.Would like to run with GEOS sulfate surface area.
Looks likeBrONO2+O
rxn not added
BrO not imp.at high alt
Has ClO+OH HCl notbeen added ?!?
112
MRI
O(3P) filenot submitted
O(1D) filenot submitted
ClO high
The chemical mechanism used in this model seems to differ from PSS in important manners.Would like to run with MRI sulfate surface area; however, this will not affect comparisons for p < ~20 hPa.
Scale Change
NOx lowcompared to PSS
113
UMSLIMCAT
O(3P) filenot submitted
O(1D) filenot submitted
BrO file not submitted;have to compare BrO+Br
ClO offset consistent with NOx
HOx offset consistentwith NOx
Difference inconseq.;BrO does not matter
at high altitude
Looks great. Would like to examine O(3P), O(1D), BrO,and run with UMSLIMCAT sulfate surface area.
Scale Change
NOx lowcompared to PSS
Recommended