ASL Basics Global Lower-Tropospheric Measurements Calibration of - - PowerPoint PPT Presentation

▶

Sep 07, 2023 195 likes •495 views

ASL Basics Global Lower-Tropospheric Measurements Calibration of CO 2 with AIRS Results Comparison Conclusion Breno Imbiriba, Larrabee Strow, Sergio de Souza-Machado, and Scott Hannon. Atmospheric Spectroscopy Laboratory (ASL) University

SLIDE 1

Basics Calibration Results Comparison Conclusion

ASL

Global Lower-Tropospheric Measurements

f CO2 with AIRS

Breno Imbiriba, Larrabee Strow, Sergio de Souza-Machado, and Scott Hannon.

Atmospheric Spectroscopy Laboratory (ASL) University of Maryland Baltimore County Physics Department and the Joint Center for Earth Systems Technology

Airs Science Team Meeting - Pasadena - CA May 5, 2009

1 / 28

SLIDE 2

Basics Calibration Results Comparison Conclusion

ASL

Overview

Try to sense as low in the atmosphere as possible. Complements Chahine’s 250 mbar retrievals.

−0.025 −0.02 −0.015 −0.01 −0.005 100 200 300 400 500 600 700 800 900 1000

AIRS 791.7 wn Jacobian (dBT/dCO2) dK/dppm mbar

Must handle surface carefully. Clear only. May try cloud-cleared radiances in the future. Ocean zonal CO2 derived using this algorithm extensively validated in our 2007 JGR paper. This work: Validate land CO2 measurements. Nominal reporting grid is 1-2 months, 5 degree grid boxes.

2 / 28

SLIDE 3

Basics Calibration Results Comparison Conclusion

ASL

Data

FOV Selection

Used AIRS ACDS clear FOVs Removed about 7% of FOVS due to cirrus ECMWF (with adjustments) used for atmospheric state.

Atmospheric State

Atmospheric state from ECMWF adjusted for Tsfc and total column water. Some FOVs removed due to poor water vapor. Sea surface emissivity - Masuda. Land surface emissivity: UW MODIS-based model. Further adjustments to the ǫTs product done simultaneously with CO2 retrieval.

3 / 28

SLIDE 4

Basics Calibration Results Comparison Conclusion

ASL

How Good is ECMWF?

ECMWF strongly ties temperature to sondes, dynamic bias adjustment procedure applied to satellite data Difference of Std of bias between AIRS and ECMWF and AIRS NEDT is ∼0.03 to 0.05K, equivalent to ∼ 1-2 ppm of CO2.

700 720 740 760 780 800 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 Wavenumber (cm−1) Std, Noise in K ECMWF Std AIRS NEDT 4 / 28

SLIDE 5

Basics Calibration Results Comparison Conclusion

ASL

CO2 Retrieval

790cm−1 (surface channel, no CO2 sensitivity) 791cm−1 (temperature insensitive CO2 channel) B790

bs − B790

calc

= J790

Ts δTs

B791

bs − B791

calc

= J791

Ts δTs + J791 CO2δCO2

Assume emissivity constant between 790 and 791 cm−1. Jacobians J computed for each FOV CO2 also retrieved similarly using SW channels (2395 cm−1 region). These are much more temperature sensitive and provide a diagnostic on errors in ECMWF T(z).

5 / 28

SLIDE 6

Basics Calibration Results Comparison Conclusion

ASL

Bias Adjustment Needed for LW and SW CO2 Retrieval

Spectroscopy plus radiometric errors could easily reach 5-10 ppm Used NOAA’s GlobalView data set 400-500 mbar sensitivity limited validation to 11 aircraft sites (all US). Hope to find more validation data sets in Russia, Amazonia.

6 / 28

SLIDE 7

Basics Calibration Results Comparison Conclusion

ASL

NOAA’s GlobalView Aircraft Sites

Limited CO2 profile information even with aircraft sites. Simple approach; use the highest altitude flight only (usually 5-8 km). GlobalView smooths the raw data. Form time series → and linearly interpolate to AIRS measurement times. Coincidence criteria: 4 degrees lat/lon and 4 days.

2002 2003 2004 2005 2006 2007 2008 365 370 375 380 385 390 Date CO2 ppm Beaver Crossing, Nebraska, United States (BNE) (40.80N,97.18W) 1000m 3000m 5000m 7000m 2002 2003 2004 2005 2006 2007 2008 366 368 370 372 374 376 378 380 382 384 386 USA sites − Highest altitude airplanes Date CO2 ppm

NWR 3526m LEF 4000m NHA 5500m BNE 7000m DND 7000m HIL 7000m RIA 7000m HAA 7500m HFM 7500m TGC 7500m CAR 8000m

7 / 28

SLIDE 8

Basics Calibration Results Comparison Conclusion

ASL

Sample Histograms of Obs-Calc CO2, Day

Std due to AIRS Noise should be 7-9 ppm CO2

−30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 1600 1800 2000 ΔCO2 ppm Count HAA − Day − Mean offset = 8.11 ppm − Standard Deviation 7.03 ppm −30 −20 −10 10 20 30 40 50 500 1000 1500 2000 2500 3000 ΔCO2 ppm Count BNE − Day − Mean offset = 7.68 ppm − Standard Deviation 7.44 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 ΔCO2 ppm Count TGC − Day − Mean offset = 7.57 ppm − Standard Deviation 7.18 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 1600 1800 2000 ΔCO2 ppm Count RIA − Day − Mean offset = 7.89 ppm − Standard Deviation 7.67 ppm −30 −20 −10 10 20 30 40 50 500 1000 1500 ΔCO2 ppm Count HIL − Day − Mean offset = 7.64 ppm − Standard Deviation 7.70 ppm −30 −20 −10 10 20 30 40 50 50 100 150 200 250 300 350 400 450 500 ΔCO2 ppm Count HFM − Day − Mean offset = 8.35 ppm − Standard Deviation 8.63 ppm

8 / 28

SLIDE 9

Basics Calibration Results Comparison Conclusion

ASL

Sample Histograms of Obs-Calc CO2, Night

−30 −20 −10 10 20 30 40 50 500 1000 1500 2000 ΔCO2 ppm Count HAA − Night − Mean offset = 8.19 ppm − Standard Deviation 6.86 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 1600 1800 ΔCO2 ppm Count BNE − Night − Mean offset = 7.05 ppm − Standard Deviation 7.72 ppm −30 −20 −10 10 20 30 40 50 500 1000 1500 2000 2500 3000 ΔCO2 ppm Count TGC − Night − Mean offset = 6.72 ppm − Standard Deviation 7.30 ppm −30 −20 −10 10 20 30 40 50 500 1000 1500 ΔCO2 ppm Count RIA − Night − Mean offset = 5.81 ppm − Standard Deviation 7.59 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 1600 1800 ΔCO2 ppm Count HIL − Night − Mean offset = 5.29 ppm − Standard Deviation 7.53 ppm −30 −20 −10 10 20 30 40 50 100 200 300 400 500 600 700 800 ΔCO2 ppm Count HFM − Night − Mean offset = 4.73 ppm − Standard Deviation 8.10 ppm

9 / 28

SLIDE 10

Basics Calibration Results Comparison Conclusion

ASL

Bias Calibration

−30 −20 −10 10 20 30 40 50 2000 4000 6000 8000 10000 12000 14000 ΔCO2 ppm Count USA − Day − Mean offset = 7.70 ppm − Standard deviation = 7.62 ppm −30 −20 −10 10 20 30 40 50 2000 4000 6000 8000 10000 12000 ΔCO2 ppm Count USA − Night − Mean offset = 6.28 ppm − Standard deviation = 7.76 ppm

Errors appear to be relatively gaussian Mean bias derived from ∼200-500 AIRS FOVs per site Daytime (Nighttime) Bias: 7.70 (6.28) ppm Individual site Std: ∼6 ppm. Uncertainty = (mean over 11 sites)/ √ 11 ≈ 0.4ppm. Roughly the same as single site statistical uncertainty. Too low; US only sites too homogeneous.

10 / 28

SLIDE 11

Basics Calibration Results Comparison Conclusion

ASL

Time series

Hard to examine AIRS versus aircraft CO2 time series since single FOV noise high. So, fit AIRS data with the a simle function: f (t) = A + Rt + C1 sin(ωyt + φ1) + C2 sin(2ωyt + φ2), Two examples: HAA (7500 m) and BNE (7000 m)

2002 2003 2004 2005 2006 2007 2008 2009 350 355 360 365 370 375 380 385 390 395 400 longwave − daytime − HAA AIRS − obs GV − obs AIRS − fit 2002 2003 2004 2005 2006 2007 2008 2009 350 355 360 365 370 375 380 385 390 395 400 longwave − daytime − BNE AIRS − obs GV − obs AIRS − fit

11 / 28

SLIDE 12

Basics Calibration Results Comparison Conclusion

ASL

Southern Hemisphere Independent Data Set

Rarotonga, Cook Islands (RTA) - Cape Grim, Tasmania, Australia (AIA)

−30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 ΔCO2 ppm Count RTA − Night − Mean offset = 9.20 ppm − Standard Deviation 6.91 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 1200 1400 1600 ΔCO2 ppm Count RTA − Day − Mean offset = 9.96 ppm − Standard Deviation 7.12 ppm −30 −20 −10 10 20 30 40 50 200 400 600 800 1000 ΔCO2 ppm Count AIA − Night − Mean offset = 8.65 ppm − Standard Deviation 7.48 ppm −30 −20 −10 10 20 30 40 50 100 200 300 400 500 600 ΔCO2 ppm Count AIA − Day − Mean offset = 12.60 ppm − Standard Deviation 8.23 ppm

RTA: 4500 m, ocean, good agreement AIA: 6500 m, daytime bias implies we are a few ppm low

12 / 28

SLIDE 13

Basics Calibration Results Comparison Conclusion

ASL

AIRS Trends

Northern Hemisphere (30-50 deg) zonal avg

2003 2004 2005 2006 2007 2008 −6 −4 −2 2 4 6 8 10 12 Time CO2 −370 (ppm) Night Day

13 / 28

SLIDE 14

Basics Calibration Results Comparison Conclusion

ASL

Jacobians - Day and Night differences

Weighed mean of the pressure field - using the calculated Jacobians as the weighing function. Overall, Daytime sees lower over continental areas. Fill in blancks with surrounding averaged data (Sahara/Poles). For now we use night only climatological Jacobians for CT comparisons

14 / 28

SLIDE 15

Basics Calibration Results Comparison Conclusion

ASL

Yearly mean (Fall to Fall) - 2002 to 2006

CO2 mean over all 5 years

15 / 28

SLIDE 16

Basics Calibration Results Comparison Conclusion

ASL

5-Year Seasonal Mean

16 / 28

SLIDE 17

Basics Calibration Results Comparison Conclusion

ASL

AIRS Growth Rate

Very rough estimate, just raw differences

Mean is around 2.5ppm/year Will fit each grid point to rate equation in future Higher rates for high-latitude land? Southern Africa anomaly is Kalahari Desert - will investigate.

17 / 28

SLIDE 18

Basics Calibration Results Comparison Conclusion

ASL

AIRS versus NOAA’s CarbonTracker

Carbontracker - NOAA’s asimilated CO2 transport model. Uses GobalView data as ingest. Data is in 4D form - We average in time and interpolate to AIRS pressure levels before applying our measurement weighting function.

18 / 28

SLIDE 19

Basics Calibration Results Comparison Conclusion

ASL

Error in Using Zonal Jacobian Climatology

Left: Zonal climatology, Right: Actual Jacobians Climatology for Jacobians introduces 1-2 ppm errors. Will fix.

19 / 28

SLIDE 20

Basics Calibration Results Comparison Conclusion

ASL

5-Year seasonal mean - Spring - Summer

20 / 28

SLIDE 21

Basics Calibration Results Comparison Conclusion

ASL

5-Year seasonal mean - Fall - Winter

21 / 28

SLIDE 22

Basics Calibration Results Comparison Conclusion

ASL

Seasonal Cycle of Year 2006 - Spring - Summer

22 / 28

SLIDE 23

Basics Calibration Results Comparison Conclusion

ASL

Seasonal Cycle of Year 2006 - Fall - Winter

23 / 28

SLIDE 24

Basics Calibration Results Comparison Conclusion

ASL

AIRS and Schiamachy

Scimachy - near IR - daytime only.

24 / 28

SLIDE 25

Basics Calibration Results Comparison Conclusion

ASL

5-Year seasonal mean - Spring/Summer

25 / 28

SLIDE 26

Basics Calibration Results Comparison Conclusion

ASL

5-Year seasonal mean - Fall/Winter

26 / 28

SLIDE 27

Basics Calibration Results Comparison Conclusion

ASL

Outlook

Very encouraging results Not discussed: AIRS SW versus LW differences suggest that ECMWF errors are equivalent to ∼1 ppm. AIRS and the assimilated model CarbonTracker agree to some degree. AIRS indicates CarbonTracker transport is too “strong”. Of concern, our low SH ocean CO2. That is also where our day/night differences are largest. Some agreement with preliminary SCIAMACHY data. SCIAMACHY unreasonably low at times??? (Will discuss with Bremen.) Need to generate, and save, gridded Jacobians for proper comparison to CarbonTracker (or other models). Like to improve clear yield in NH winter, or move to cloud-cleared radiances??

27 / 28

SLIDE 28

Basics Calibration Results Comparison Conclusion

ASL

250 mbar (Chahine) vs 450 mbar (UMBC) CO2

250 mbar 450 mbar

28 / 28