dr. benjamin r. lintner
DESCRIPTION
Blowin’ in the wind: Atmospheric circulation, tracer transport, and CO 2 variability. Dr. Benjamin R. Lintner. Department of Atmospheric & Oceanic Sciences and Institute of Geophysics & Planetary Physics University of California Los Angeles. - PowerPoint PPT PresentationTRANSCRIPT
Dr. Benjamin R. Lintner
I.Y. Fung, C.D. Koven (UC Berkeley); W. Buermann (UCLA)
A.B. Gilliland (Air Resources Laboratory)
C.J. Tucker, J.E. Pinzon (Goddard Space Flight Center)
A. Angert (Weizmann Institute of Science); K.P. Bowman (Texas A&M)
Department of Atmospheric & Oceanic Sciences and Institute of Geophysics & Planetary Physics
University of California Los Angeles
Blowin’ in the wind:Blowin’ in the wind: Atmospheric circulation, tracer transport,
and CO2 variability
Conservation of CO2
€
∂χ∂t
+ Γ(χ ) = S(χ )
€
χ =[CO2] Concentration of CO2
€
Γ(χ ) =v
∇ ⋅(v v χ ) Transport of CO2
€
S(χ ) = Sδ(p − ps) + P(χ ) Source of CO2
( emission + production)
0
The big picture…
€
∂χ∂t
+ Γ(χ ) = S(χ )
Q: How can we obtain this term?Bottom-up: Field measurements (direct but costly)Top-down: Inversions (indirect but uncertain)
General inversion methodology: For a set of observations χobsi,construct and
minimize a cost function J assuming a set of unit source/sink basis regions ûk and a forward transport model G.
€
J = σ χ i
2 (χ iobs − ˆ χ i)
2
i
∑
€
ˆ χ ik = Gi( ˆ u k )⇒ ˆ χ i = mk
ˆ χ ik
k
∑
€
S est = m kk
∑ ˆ u k
€
∂J
∂mk mk = m k
€
=0
Q: What kind of information can be extracted?
Q: How much confidence do we have in G?
Lat-lon distribution of fossil fuel (FF) CO2*
Strong interhemispheric (meridional) gradient roughly co-located with the ITCZ; (generally) weak zonal gradients, reflecting rapid east-west mixing, although locally large near source regions.
Lat-height distribution of FF CO2
Meridional gradients are relaxed aloft compared to the surface; a “reversed” vertical gradient is characteristic of low latitudes in the nonsource (Southern) hemisphere.
Lat-Height characteristics in the Tropics are broadly understood in terms of the mean meridional overturning (Hadley) circulation.
2-box model
€
SS (t)
€
SN (t)€
χS (t)
€
χN (t)
€
y0
€
yS
€
yN
€
∂χN (t)
∂t= SN (t) + vχ{ }
y0
€
{A} = AdppT
pS∫
€
∂χS (t)
∂t= SS (t) − vχ{ }
y0
€
AN = A{ }dyy0
yN∫
€
AS = A{ }dyyS
y0∫
€
vχ{ }y0
≈ −κdχ
dyy0
≈ −χ N (t) − χ S (t)
τ IHT (t)
€
τ IHT (t) = 2χ N (t) − χ S (t)
SN (t) − SS (t) −∂tχ N (t) + ∂tχ S (t)Interhemispheric exchange time (τIHT):
Time required for χS to “catch-up” to χN.
Definitions:
Anthropogenic CO2 in a 2-box model
Observational Constraints:
χsfc= 2.5 ppmv (χcolumn 2/3χsfc = 1.7 ppmv)
χG/t = (1/2)(χN +χS)/t = 1.5 ppmv/yr
Consider Fossil Fuel (FF): SNFF = 6 PgC/yr*; SS
FF ~ 0 pgC/yr
For τIHT ~ 1 yr χ(est) = 3.0 ppmv (4.5 ppmv at surface)
STotal = 3.4 PgC/yr SNSink - SS
Sink = 2.6 pgC/yr
(SNTotal + SS
Total) = 3.0 PgC/yr
SNSink = 2.8 PgC/yr; SS
Sink = 0.2 PgC/yr
The implied “sink” partitioning is:
*Note: 1PgC = 0.5 ppmv (mixed over whole troposphere)
Q: How do we obtain this value?
Estimates of annual-mean τIHT
Air
mas
s
CC
l 3F
CH
3CC
l 3
CH
4
14C
O2
85K
r
SF
6
All
Ob
s.
1yr
Observations (see Lintner 2003)
Mean : 1.280.33 years
AN
U
GF
DL
-GC
TM
GIS
S
MU
TM
CC
C
CS
U
-SK
YH
I
GIS
S-U
VIC
/UC
B
NIR
E
TM
2
All
Mod
.
1yr
TransCom Models (Denning et al., 1999)
Mean : 0.880.22 years
IHT versus dynamics
0.5
0.7
0.9
1.1
1.3
18 20 22 24 5 6 7 8 9
SF6 CH4 CFC11
τ IHT (
year
s)
Peak Hadley Strength (x 1010 kg/s) JJA Land Precip. at 18N (mm/day)
From GISS simulations of Rind et al., 2007
Changes to model vertical (+ horizontal) resolution alter dynamics, which in turn alters τIHT. Generally, stronger Hadley circulation (land region convection) favors faster IHT.
τIHT seasonal cycle
From Lintner et al., 2004
0.86 years
Seasonal cycle: ±20-30% of annual mean
“fast” IHT/small τIHT in winter/summer and “slow” IHT/large τIHT in spring/autumn
0.2
0.6
1.0
1.4
1.8
NCEP/MATCH
GISS-UCB 1.27 years
J MF A JM J SA O DN
year
s
Q: What is the source of this seasonality?
CFC-11 (1979-1988)
0.73 years
Transport partitioning
€
vv χ
Total{
=v v χ
Mean1 2 3
+v v * χ *
Stationary Eddy1 2 3
+ ′ v v ′ χ
Transient Eddy1 2 3
€
A =1
x f − x i
Adxxi
x f∫
€
A* = A − A
€
A =1
t f − ti
Adtt i
t f∫
€
′ A = A − A
Defining:
Meridional transport in the GISS-UCB model
Month
x 10
7 kg/
mon
th
€
vχ{ }
€
v χ{ }
€
v* χ *{ }
€
′ v ′ χ { }
1.50
1.25
1.00
years
€
τ IHT
DJF Streamfunction
MAM StreamfunctionMean meridional, stationary-eddy, and transient eddy: ~1/3 of annual-mean total transport (Denning et al., 1999)
From seasonal cycleRind et al., 2007 sensitivity
τ IHT (
year
s)
8 12 16 20
1.0
1.4
1.8
Max. Hadley Strength (x 1010 kg/s)
Seasonality dominated by mean meridional (Hadley) circulation, with fast (slow) IHT occurring when Hadley circulation is strongly asymmetric (symmetric) w.r.t. equator.
Regional transport characteristics
6
0
-16€
vχJanuary Julyx 107 kg/month
-30
-10 10
30
€
′ v ′ χ x 105 kg/monthJanuary July
Outflow associated with deep convection
MOPITT CO for January 20-27, 2001 at 500 mb
From Edwards et al., 2003
Small scale vertical transport: Biomass burning plumes are uplifted near the ITCZ and diverged aloft.
τIHT interannual variability (IAV)
El Niño
La Niña
GISS-UCB
NCEP/MATCH
Interannual variations: ±5-10% of annual mean
“fast” IHT/small τIHT during La Niña events and “slow” IHT/large τIHT El Niño events(?)
Sources of GISS-UCB IAV
From Lintner et al., 2004
10E-40E
60E-100E 110E-150E
140W-110W
40W-10W
160E-160W
75W-45W
Often, more intense convection favors faster IHT (e.g., Indian Ocean/South Asia during JJA; South America during ASO), but not always (e.g., Eastern Pacific).
Conditional- averaged STRENGTH of convection for fast IHT and slow IHT.
Sources of GISS-UCB IAV
From Lintner et al., 2004
10E-40E
60E-100E 110E-150E
140W-110W
40W-10W
160E-160W
75W-45W
Conditional- averaged LOCATION of convection (ITCZ) for fast IHT and slow IHT.
Generally, more extreme ITCZ displacements favor faster IHT (broadly consistent with chaotic advective mixing noted by Bowman and Cohen, 1997)
But there is significant region-to-region variation…..
Summary of IHT
• 2-box model mean exchange times (τIHT) of 0.8-1.2 years, with pronounced seasonality (20-30% of mean) and IAV (5-10%)
From the perspective of carbon cycle, these variations impact
estimates of the meridional distribution of CO2 sources/sinks.
• τIHT reflects contributions from both zonal-mean and regional circulations
Results are model-dependentObtaining a detailed picture of 3D interhemispheric transport
pathways, i.e., how (and where) transport across the “interhemispheric transport barrier” occurs, is desirable.
What are the temporal characteristics (subseasonal, seasonal, interannual) of these pathways?
Mauna Loa Observatory (MLO), Hawai´i
Hurricane Flossie
155ºW, 19ºN 3397 m a.s.l.
CO2 seasonality at MLO
2002 2003 2004 2005 2006 2007
380
384
376
372
Respiration
Photosynthesis
CO2 to atmos.
CO2 from atmos.
Detrended Seasonal Cycle
Amplitude
ppmv
Changing MLO CO2 amplitude
2000
Temp (K)
-0.8
-0.4
0.0
0.4
0.8
0.9
1.0
1.1
1.2
Relative Amp
Land Surface Temp (30N-80N)
1960 1970 1980 1990
Q: What accounts for this change of behavior?
Mauna Loa Amplitude
Keeling (1996): increasing MLO amplitude from 1960s-early 1990s consistent with high latitude land surface warming and northward “greening”
Since the early 90s, MLO amplitude has decreased but NH land surface temperatures have continued to warm.
MLO and atmospheric circulation
From Lintner et al., 2006
JFM AMJ
JAS OND
NCEP Reanalysis 700 mb wind (vectors) and streamflow (contours)
6-10 day Lagrangian back trajectories
Eurasia
N. America NCEP/MATCH FF SF6 HCFC22
Role of atmospheric transport
1960 1970 1980 1990 2000
0.9
1.0
1.1
1.2
-2
-1
0
1
2
Comparison of observed MLO Amp (black) to MATCH-simulated (blue) with no source/sink variability: Decrease in MATCH over the 1990s suggests nonnegligible transport signature.
Observed MLO Amp
MATCH simulated Amp
Q: What is the source of the simulated trend?
Simulated FF (AMJ)222Rn (AMJ)
From Buermann et al., 2007
Data from simulation of Higuchi et al., 2003: Some indication of a trend over the 1990s?
Backtrajectory cluster analysis*
“Long range” cluster“Short range” cluster
1988 1992 1996 20001988 1992 1996 2000
-2
-1
0
1
2
-2
-1
0
1
2Yearly membership in
short range cluster AMJ FF Yearly membership in
long range cluster AMJ FF
From Lintner et al., 2006*Results shown are for April-May-June (AMJ)
Changing temperature and moisture influence
Warm-season temperatures: persistent +ve correlations with MLO Amp from 1960s-mid 1970s but little thereafter….
Warm-season hydrology: recent development of strong correlations for North Americaeffect of North American drought during 1998-2003 reduced carbon uptake, resulting in decreased amplitude
From Buermann et al., 2007
Schematic of MLO amplitude controls
Eurasia
North America
Cold season
Warm season
Cold season
Warm season
Temperature: 1yr-lag
Temperature Drought/rain cycles
Temperature: 1yr-lag 1 and 2 yr-lag 2 yr-lag
PhotosynthesisPhotosynthesis
Respiration
Respiration
Photosynthesis
Transport
Summary of MLO
• Because of the relationship of the observing site relative to large-scale circulation, MLO is most sensitive to Eurasian influence during boreal cold season and North America influence during boreal warm season.
Provides some basis for the geographic distribution of correlationsNonstationarity/changing influence through time
• Some of the 1990s downward trend may be directly attributable to changes in transport as seen at the observing site.
Less Eurasian-originating transport in boreal spring reduces amplitude (support from backtrajectories/222Rn)