Update on NOAA CO 2 Retrievals: Validation and Future Directions - - PowerPoint PPT Presentation

▶
update on noaa co 2 retrievals validation and future
SMART_READER_LITE
LIVE PREVIEW

Update on NOAA CO 2 Retrievals: Validation and Future Directions - - PowerPoint PPT Presentation

Feb 08, 2024 203 likes •490 views

Update on NOAA CO 2 Retrievals: Validation and Future Directions Eric Maddy, Chris Barnet, Mitch Goldberg, Xingpin Liu, Lihang Zhou and Walter Wolf. NOAA/NESDIS/STAR AIRS Science Team Meeting October 10, 2007 E. Maddy (PSGS, Inc.) Update on

slide-1
SLIDE 1

Update on NOAA CO2 Retrievals: Validation and Future Directions

Eric Maddy, Chris Barnet, Mitch Goldberg, Xingpin Liu, Lihang Zhou and Walter Wolf.

NOAA/NESDIS/STAR

AIRS Science Team Meeting

October 10, 2007

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 1 / 22

slide-2
SLIDE 2

Outline

  • Description of NOAA AIRS CO2 retrieval methodologies
  • How well can we do with a simple climatology?

CO2(t) = a0 + a1 ∗ t

  • Development of error estimates.
  • Paper on averaging kernels (related to these error estimates)

accepted (with revisions) to IEEE TGARS.

  • Validation with full resolution data vs. NOAA ESRL/GMD Aircraft

(2005)1and Global Gridded data vs. JAL Matsueda (August 2003 - 2006.)

  • Comparison of AIRS and models – What new information can

AIRS provide to modeling community?

1Submitted to JGR in review

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 2 / 22

slide-3
SLIDE 3

NOAA AIRS CO2 Retrievals

  • Use AIRS Science Team Methodology.
  • Version 4.7.
  • before cloudy regression introduced.
  • NOAA O3 regression on.
  • 70 channels (mostly 15 micron).
  • Derive CO2 in 4 layers in troposphere, 1 stratospheric.
  • Use Optimal Estimation w/ SVD
  • Runs within offline science code (consistent RTA/channel set).
  • Derive 6 - 10 CO2 basis functions.
  • Runs very fast No appreciable difference in run-time compared to

AIRS Science Team methodology

  • Validation with full resolution data vs. NOAA ESRL/GMD Aircraft

(2005) and JAL Matsueda between August 2003 - 2006 (only AIRS Science Team approach). Each retrieval methodology has the ability to calculate averaging kernels and related diagnostics (d.o.f., etc.) and propagate error estimates.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 3 / 22

slide-4
SLIDE 4

NOAA AIRS CO2 Retrievals

  • Use AIRS Science Team Methodology.
  • Version 4.7.
  • before cloudy regression introduced.
  • NOAA O3 regression on.
  • 70 channels (mostly 15 micron).
  • Derive CO2 in 4 layers in troposphere, 1 stratospheric.
  • Use Optimal Estimation w/ SVD
  • Runs within offline science code (consistent RTA/channel set).
  • Derive 6 - 10 CO2 basis functions.
  • Runs very fast No appreciable difference in run-time compared to

AIRS Science Team methodology

  • Validation with full resolution data vs. NOAA ESRL/GMD Aircraft

(2005) and JAL Matsueda between August 2003 - 2006 (only AIRS Science Team approach). Each retrieval methodology has the ability to calculate averaging kernels and related diagnostics (d.o.f., etc.) and propagate error estimates.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 3 / 22

slide-5
SLIDE 5

NOAA AIRS CO2 Retrievals

  • Use AIRS Science Team Methodology.
  • Version 4.7.
  • before cloudy regression introduced.
  • NOAA O3 regression on.
  • 70 channels (mostly 15 micron).
  • Derive CO2 in 4 layers in troposphere, 1 stratospheric.
  • Use Optimal Estimation w/ SVD
  • Runs within offline science code (consistent RTA/channel set).
  • Derive 6 - 10 CO2 basis functions.
  • Runs very fast No appreciable difference in run-time compared to

AIRS Science Team methodology

  • Validation with full resolution data vs. NOAA ESRL/GMD Aircraft

(2005) and JAL Matsueda between August 2003 - 2006 (only AIRS Science Team approach). Each retrieval methodology has the ability to calculate averaging kernels and related diagnostics (d.o.f., etc.) and propagate error estimates.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 3 / 22

slide-6
SLIDE 6

NOAA AIRS CO2 Retrievals

  • Use AIRS Science Team Methodology.
  • Version 4.7.
  • before cloudy regression introduced.
  • NOAA O3 regression on.
  • 70 channels (mostly 15 micron).
  • Derive CO2 in 4 layers in troposphere, 1 stratospheric.
  • Use Optimal Estimation w/ SVD
  • Runs within offline science code (consistent RTA/channel set).
  • Derive 6 - 10 CO2 basis functions.
  • Runs very fast No appreciable difference in run-time compared to

AIRS Science Team methodology

  • Validation with full resolution data vs. NOAA ESRL/GMD Aircraft

(2005) and JAL Matsueda between August 2003 - 2006 (only AIRS Science Team approach). Each retrieval methodology has the ability to calculate averaging kernels and related diagnostics (d.o.f., etc.) and propagate error estimates.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 3 / 22

slide-7
SLIDE 7

Improvement Over Simple Climatology

  • Theoretical error analysis for our Version 5 Climatology

Polar

2 4 6 8 10 CO2 error, [ppmv] 5 10 15 20 25 Altitude, [km]

Midlatitude

2 4 6 8 10 CO2 error, [ppmv] 5 10 15 20 25

Tropical

2 4 6 8 10 CO2 error, [ppmv] 5 10 15 20 25

H2O T(p) O3 Tsurf Smoothing Total a priori

  • Calculation uses a priori covariance calculated as the dif ference between ESRL aircraft

and our simple Version 5 climatology.

  • Ability to partition error sources and their effect on the retrieval. Effect minimized as we

have assumed a perfect knowledge of the error covariation of intefering species (assumed ad-hoc: S(z, z′) = σ(z)σ(z′) · exp(−|z − z′|/L)).

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 4 / 22

slide-8
SLIDE 8

ESRL/GMD Aircraft Validation Approach

  • Use full resolution AIRS retrievals (previously validated w/ 3◦x

3◦grids)

  • Average AIRS CO2 between 6-10 km (nominally where jacobian

has maximum sensitivity).

  • Use nominal jacobians (wrt. latitude) to weight ESRL aircraft.
  • Enables comparison of scalar measurements
  • Removes variability in lowest 2.5 km
  • Average all retrievals within 200km with temporal matchup window

between 1 day - 1 month.

  • Profile statistics will also be shown. NOAA/ESRL

CarbonTracker2model used to extend profiles above 8 km.

2http://www.cmdl.noaa.gov/ccgg/carbontracker

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 5 / 22

slide-9
SLIDE 9

Example Comparison: Estevan Point, British Columbia

Estevan Point, British Columbia NOAA ESRL/GMD Aircraft and AIRS Retrieval Timeseries

Dec 03 Jul 04 Jan 05 Aug 05 Mar 06 Date, [1] 365 370 375 380 385 390 395 CO2, [ppmv]

AIRS CO2 6km to 8km AIRS CO2 +/- σ(CO2 ) Aircraft above 2.5km

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 6 / 22

slide-10
SLIDE 10

AIRS Science Team Algorithm vs. ESRL/GMD Aircraft

|∆t| < 7.00 days

365 370 375 380 385 390 AIRS CO2, [ppmv] 365 370 375 380 385 390 ESRL CO2, [ppmv]

Bradgate, Iowa Beaver Crossing, Nebraska Briggsdale, Colorado Dahlen, North Dakota Estevan Point, British Columbia Fairchild, Wisconsin Molokai Island, Hawaii Harvard Forest, Massachusetts Homer, Illinois Park Falls, Wisconsin Worcester, Massachusetts Oglesby, Illinois Poker Flat, Alaska Rowley, Iowa Rarotonga, Cook Islands Trinidad Head, California

N MATCHUP : 495 RMS(PPMV) : 2.06 SDV(PPMV) : 1.80 IQR(PPMV) : 2.23 BIAS(PPMV) :-0.99 CORREL : 0.77 RANKCORREL: 0.79

  • Total magnitude of drawdown at LEF not captured possible
  • ver-regularization wrt. characteristic variability.
  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 7 / 22

slide-11
SLIDE 11

AIRS Science Team Algorithm vs. ESRL/GMD Aircraft

|∆t| < 7.00 days

365 370 375 380 385 390 AIRS CO2, [ppmv] 365 370 375 380 385 390 ESRL CO2, [ppmv]

Bradgate, Iowa Beaver Crossing, Nebraska Briggsdale, Colorado Dahlen, North Dakota Estevan Point, British Columbia Fairchild, Wisconsin Molokai Island, Hawaii Harvard Forest, Massachusetts Homer, Illinois Park Falls, Wisconsin Worcester, Massachusetts Oglesby, Illinois Poker Flat, Alaska Rowley, Iowa Rarotonga, Cook Islands Trinidad Head, California

N MATCHUP : 495 RMS(PPMV) : 2.06 SDV(PPMV) : 1.80 IQR(PPMV) : 2.23 BIAS(PPMV) :-0.99 CORREL : 0.77 RANKCORREL: 0.79

  • 0.5% uncertainty from space!
  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 8 / 22

slide-12
SLIDE 12

Calculated a priori and Retrieval Error Covariances

Tropical Mid-Latitude/Polar

A Priori Error Correlation Matrix

2 4 6 8 10 Relative Altitude, [km]

A Priori Error Patterns

19.83 (90.4%) 5.76 (7.6%) 2.04 (1.0%) 1.41 (0.5%) σ( xa ) σ( x

^ )

Retrieval Error Correlation Matrix

2 4 6 8 10 Relative Altitude, [km] 2 4 6 8 10 Relative Altitude, [km]

Retrieval Error Patterns

  • 10
  • 5

5 10 Scaled Eigenvector, [ppmv] 16.49 (83.1%) 6.66 (13.6%) 2.56 (2.0%) 1.40 (0.6%)

A Priori Error Correlation Matrix

2 4 6 8 10 Relative Altitude, [km]

A Priori Error Patterns

30.96 (76.0%) 15.27 (18.5%) 5.78 (2.6%) 3.90 (1.2%) σ( xa ) σ( x

^ )

Retrieval Error Correlation Matrix

2 4 6 8 10 Relative Altitude, [km] 2 4 6 8 10 Relative Altitude, [km]

Retrieval Error Patterns

  • 10
  • 5

5 10 Scaled Eigenvector, [ppmv] 23.49 (63.3%) 15.67 (28.1%) 6.20 (4.4%) 3.93 (1.8%)

  • 1.0
  • 0.5

0.0 0.5 1.0

  • Total variance of the retrieval is less than the a priori indicating a gain in information.
  • First eigenfunction variance (and percent of total variance) of the retrieval is less than a

priori.

  • Retrieval tends to redistribute variance among higher order eigenfunctions, which are

similar in shape to the a priori, indicating we have only 1 piece of information, albeit well constrained, in the vertical. Vertical resolution ≈6-8 km.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 9 / 22

slide-13
SLIDE 13

Validation of Error Propagation

  • In general, predicted errors and actual errors compare very well.

Polar

2 4 6 8 10 CO2 error, [ppmv] 2 4 6 8 10 Altitude, [km]

Midlatitude

2 4 6 8 10 CO2 error, [ppmv] 2 4 6 8 10

Tropical

1 2 3 4 5 CO2 error, [ppmv] 2 4 6 8 10 Predicted Actual a priori

  • Largest discrepancy is above 8 km where the NOAA CarbonTracker model was used to

extend the aircraft profiles.

  • Uncertainties in the profile extension procedure, the model profiles, AIRS retrievals and/or

error analysis are possible explanations to the disagreement.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 10 / 22

slide-14
SLIDE 14

Validation of Error Propagation

  • In general, predicted errors and actual errors compare very well.

Polar

2 4 6 8 10 CO2 error, [ppmv] 2 4 6 8 10 Altitude, [km]

Midlatitude

2 4 6 8 10 CO2 error, [ppmv] 2 4 6 8 10

Tropical

1 2 3 4 5 CO2 error, [ppmv] 2 4 6 8 10 Predicted Actual a priori

  • Largest discrepancy is above 8 km where the NOAA CarbonTracker model was used to

extend the aircraft profiles.

  • Uncertainties in the profile extension procedure, the model profiles, AIRS retrievals and/or

error analysis are possible explanations to the disagreement.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 10 / 22

slide-15
SLIDE 15

Comparison of OE and SVD approaches: independent validation

  • 10
  • 5

5 10 BIAS, [ppmv] 1000 800 600 400 200 Pressure, [hPa] 1 2 3 4 5 6 SDV, [ppmv] SVD Retrieval OE Retrieval A priori

  • Two retrievals with completely different methods of regularization yield almost the same

results.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 11 / 22

slide-16
SLIDE 16

JAL Aircraft Validation Approach

  • NOAA 3◦x 3◦gridded subset
  • Average AIRS CO2 between 6-10 km (nominally where jacobian

has maximum sensitivity).

  • Average all retrievals within 1000km with temporal matchup 1

month.

  • Compare to monthly averaged JAL Matsueda over latitude range

(27 months total between August 2003-2006).

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 12 / 22

slide-17
SLIDE 17

AIRS Science Team Algorithm vs. JAL Matsueda

30S-10S

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

10S-10N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

10N-25N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

25N-40N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

JAL Matsueda AIRS CO2

  • SDVE < 1.5 ppmv for all latitude ranges
  • Variability in the accuracy wrt. latitude on the
  • rder of retrieval precision
  • related to sensitivity of jacobians to

H2O displacement.

  • zonal variability of information content.
  • Averaged over all latitudes, AIRS retrievals

compare very well:

  • 0.62 ± 0.87 ppmv

Latitude SDVE BIAS Range [ppmv] [ppmv] 30S - 10S 1.32

  • 1.08

10S - 10N 1.04

  • 0.06

10N - 25N 1.45

  • 0.42

25N - 40N 1.45

  • 1.43

30S - 40N 0.87

  • 0.62
  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 13 / 22

slide-18
SLIDE 18

AIRS Science Team Algorithm vs. JAL Matsueda

30S-10S

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

10S-10N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

10N-25N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

25N-40N

12/ 03 04/ 04 08/ 04 12/ 04 04/ 05 08/ 05 12/ 05 365 370 375 380 385 CO2, [ppmv]

JAL Matsueda AIRS CO2

  • SDVE < 1.5 ppmv for all latitude ranges
  • Variability in the accuracy wrt. latitude on the
  • rder of retrieval precision
  • related to sensitivity of jacobians to

H2O displacement.

  • zonal variability of information content.
  • Averaged over all latitudes, AIRS retrievals

compare very well:

  • 0.62 ± 0.87 ppmv

Latitude SDVE BIAS Range [ppmv] [ppmv] 30S - 10S 1.32

  • 1.08

10S - 10N 1.04

  • 0.06

10N - 25N 1.45

  • 0.42

25N - 40N 1.45

  • 1.43

30S - 40N 0.87

  • 0.62
  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 13 / 22

slide-19
SLIDE 19

NOAA ESRL/GMD CarbonTracker vs ESRL/GMD Aircraft

  • NOAA ESRL/GMD CarbonTracker weighted using AIRS jacobians.

|∆t| < 0.50 days

365 370 375 380 385 390 ESRL CarbonTracker CO2, [ppmv] 365 370 375 380 385 390 ESRL CO2, [ppmv] Bradgate, Iowa Beaver Crossing, Nebraska Briggsdale, Colorado Dahlen, North Dakota Estevan Point, British Columbia Fairchild, Wisconsin Molokai Island, Hawaii Harvard Forest, Massachusetts Homer, Illinois Park Falls, Wisconsin Worcester, Massachusetts Oglesby, Illinois Poker Flat, Alaska Rowley, Iowa Rarotonga, Cook Islands Trinidad Head, California N MATCHUP : 378 RMS(PPMV) : 1.25 SDV(PPMV) : 1.21 IQR(PPMV) : 1.19 BIAS(PPMV) : 0.32 CORREL : 0.92 RANKCORREL: 0.93

  • 0.5 ppmv better precision than AIRS baseline, however CarbonTracker has been optimized

for N. America.

  • From our eigenvector analysis of our a priori, the 1st eigenfunction, a total column

perturbation, explains 80-90% of the variance.

  • We would expect good agreement near ESRL aircraft sites because constraint of having

surface / tower measurements in the assimilation.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 14 / 22

slide-20
SLIDE 20

Approach to Estimate AIRS Impact

  • Determine scales of variability in CarbonTracker calculated as the

gradient in a given direction over a defined time scale.

  • Compare to see if AIRS captures the same sort of gradients.
  • 3◦x 3◦grids used for comparison.

Thanks to Wouter Peters (NOAA ESRL) for suggesting using CarbonTracker for this approach.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 15 / 22

slide-21
SLIDE 21

ESRL/GMD CarbonTracker and AIRS Retrieval CO2 Gradients

Estevan Point, British Columbia Sinton, Texas

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 16 / 22

slide-22
SLIDE 22

ESRL/GMD CarbonTracker and AIRS Retrieval CO2 Gradients

  • 4
  • 2

2 4 January CO2 gradient, [ppmv] Distance from ESP toward TGC, [10-3*km] February March

CT σ(CT) AIRS CO2

  • 4
  • 2

2 4 April May June

  • 4
  • 2

2 4 July August September 0.0 0.5 1.0 1.5 2.0 2.5 3.0

  • 4
  • 2

2 4 October 0.0 0.5 1.0 1.5 2.0 2.5 3.0 November 0.0 0.5 1.0 1.5 2.0 2.5 3.0 December

  • 1-σ monthly variability of FT

gradients shows that in general we need to resolve 1 ppmv signals (larger variability in summer months due to rectifier) on short timescales.

  • Monthly averaged free

tropospheric (FT) gradients are within our expected error budget in terms of matching seasonality and horizontal placement.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 17 / 22

slide-23
SLIDE 23

ESRL/GMD CarbonTracker and AIRS Retrieval CO2 Gradients

Estevan Point, British Columbia Harvard Forest, Mass.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 18 / 22

slide-24
SLIDE 24

ESRL/GMD CarbonTracker and AIRS Retrieval CO2 Gradients

  • 4
  • 2

2 4 January CO2 gradient, [ppmv] Distance from ESP toward HFM, [10-3*km] February March

CT σ(CT) AIRS CO2

  • 4
  • 2

2 4 April May June

  • 4
  • 2

2 4 July August September 1 2 3 4

  • 4
  • 2

2 4 October 1 2 3 4 November 1 2 3 4 December

  • CarbonTracker shows lack of FT

gradient due to rapid advection/mixing of surface fluxes.

  • East-to-west 1-σ variability largest

in the summer months due to frontal passages and hence strong mixing (weekly differences in gradients ≈ ±3 ppm).

  • Considering retrieval error budget

(wrt. aircraft) we may be able to resolve these features on weekly timescales; however, more study is required.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 19 / 22

slide-25
SLIDE 25

Summary

  • Able to provide global retrievals of CO2 on 1-2 weekly timescales

at 1 - 2ppmv precision with a globally fixed a priori.

  • Modeling groups at NASA/GSFC, UC/Berkeley, and University of

Leicester, UK have just begun looking at the product.

  • Theoretical error estimates enable quick calculation of the AIRS

data impact. These require accurate large scale correlations in a priori due to the broad width of the kernel functions.

  • Require more high altitude profile validation data to gain

confidence in product error correlation.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 20 / 22

slide-26
SLIDE 26

Summary: Future Plans

  • True test of product skill is the ability to discern CO2 gradients.
  • Model gradients W-E are generally small due to rapid advection of

surface fluxes – we may be able to capture weekly differences.

  • As expected N-S gradients are larger with monthly variability on

the order of our precision.

  • Monthly comparisons to CarbonTracker show similar features;

more analysis required

1

Determine our ability to match gradients over shorter timescales.

2

Retest AIRS in regions poorly constrained. Model/retrieval comparisons underway for gradient appropriateness.

3

Understand (inter)product error correlations f(time,space) that introduce anomalous gradients in AIRS CO2.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 21 / 22

slide-27
SLIDE 27

Theoretical Gain Using a CarbonTracker a priori

  • Use error propagation to estimate

gain in information content by adding AIRS CO2 sensitive measurements initially with CarbonTracker errors, SCTracker

a

. ˆ S = (A − I)SCTracker

a

(A − I)T +DKbSb(DKb)T +...

  • We plot the error reduction defined

as: diag(ˆ S)/diag(SCTracker

a

)

  • Improvement outside of region

where Jacobian (dotted line) is sensitive is largely due to error correlation assumed in SCTracker

a

.

Polar

0.0 0.2 0.4 0.6 0.8 1.0 σ (x

^) / σ(xa), [1]

5 10 15 20 25 Altitude, [km]

Midlatitude

0.0 0.2 0.4 0.6 0.8 1.0 σ (x

^) / σ(xa), [1]

5 10 15 20 25

Tropical

0.0 0.2 0.4 0.6 0.8 1.0 σ (x

^) / σ(xa), [1]

5 10 15 20 25

Improvement somewhat marginal; however, CarbonTracker is highly constrained by surface measurements hence SCTracker

a

is small.

  • E. Maddy (PSGS, Inc.)

Update on NOAA CO2 retrievals: October 10, 2007 22 / 22