Chris BarnetNOAA/NESDIS/STAR

(the office formally known as ORA)AIRS Science Team Member

NPOESS Sounder Operational Algorithm Team MemberGOES-R Algorithm Working Group – Chair of Sounder Team

May 11, 2006

Carbon Fusion, Edinburgh

The Challenges in using Atmospheric Trace Gas Products from Thermal Sounders

Mitch Goldberg: AIRS & IASI Science Team, Climate ProductsWalter Wolf: Near Real Time Processing & Distribution to UsersLihang Zhou: Regression Retrieval & Near Real Time Web PageEric Maddy: CO2 retrieval, “tuning”, vertical averaging functions.

Xiaozhen Xiong: CH4 retrieval

Xingpin Liu: Re-processing, Statistics, Product web-page

2

Topics

• Introduction to our plans to use operational sounders to retrieve carbon products.

• Example of CO Product

• Example of CH4 Product & Validation Efforts

• Example of CO2 Product & Validation Efforts

• Inter-comparison of AIRS trace gas products.

3

Acronyms• Infrared Instruments

– AIRS = Atmospheric Infrared Sounder– IASI = Infrared Atmospheric Sounding Interferometer– CrIS = Cross-track Infrared Sounder– HES = Hyperspectral Environmental Suite

• Microwave Instruments– AMSU = Advanced Microwave Sounding Unit– HSB = Humidity Sounder Brazil– MHS = Microwave Humidity Sensor– ATMS = Advanced Technology Microwave Sounder– AMSR = Advanced Microwave Scanning Radiometer

• Imaging Instruments– MODIS = MODerate resolution Imaging Spectroradiometer– AVHRR = Advanced Very High Resolution Radiometer– VIIRS = Visible/IR Imaging Radiometer Suite– ABI = Advanced Baseline Imager

• Other– CALIPSO = Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observations– EUMETSAT = EUropean organization for exploitation of METeorological SATellites– GOES = Geostationary Environmental Operational Satellite– METOP = METeorological Observing Platform– NESDIS = National Environmental Satellite, Data, and Information Service– NPOESS = National Polar-orbiting Operational Environmental Satellite System– NDE = NPOESS Data Exploitation– NPP = NPOESS Preparatory Project– OCO = Orbiting Carbon Observatory– STAR = office of SaTellite Applications and Research

4

NOAA/NESDIS 20 year StrategyUsing Existing Operational Sounders.

• Now: Develop and core & test trace gas algorithms using the Aqua (May 4, 2002) AIRS/AMSU/MODIS Instruments– Compare products to in-situ (e.g., ESRL/GMD Aircraft, JAL, INTEX, etc.)

to characterize error characteristics.– The A-train complement of instruments (e.g., MODIS, AMSR, CALIPSO)

can be used to study effects of clouds, surface emissivity, etc.

• 2006: Migrate the AIRS/AMSU/MODIS algorithm into operations with METOP (2006,2011,2016) IASI/AVHRR. Study the differences between instruments, e.g., effects of scene and clouds on IASI’s instrument line-shape.

• 2009: Migrate the AIRS/IASI algorithm into operations for NPP (2009?) & NPOESS (2012?,2015?) CrIS/ATMS/VIIRS. These are NOAA Unique Products within the NOAA NDE program.

• 201?: Migrate AIRS/IASI/CrIS algorithm into GOES-R HES/ABI

5

AIRS Was Launched on the EOS Aqua Platform May 4, 2002

AIRS

HSB

AMSU-A1(3-15)

AMSU-A2(1-2)

Delta II 7920

MODIS

Aqua Acquires 325 Gb of data per day

6

AIRS has a Unique Opportunity to Explore & Test New Algorithms for Future Operational Sounder Missions.

5/4/2002

≥ 9/2008

12/18/2004

7/15/2004

Apr. 28, 2006

7

In 2007 we will have IASI & AIRS making global measurements, 4/day

• Initial Joint Polar System: An agreement between NOAA & EUMETSAT to exchange data and products.– NASA/Aqua in 1:30 pm

orbit– EUMETSAT/IASI in 9:30

am orbit (July 2006)– NPOESS/CrIS in 1:30 pm

orbit (2009?)

8

Spectral Coverage of Thermal Sounders (Example Radiances: AIRS, IASI, & CrIS)

AIRS, 2378Channels

CrIS 1305

IASI, 8401Channels

CO2 CO2O3 COCH4

9

Thermal sounder forward modelExample: upwelling radiance term

Each channel samples a finite spectral region

Vertical temperature gradient is critical for thermal sounding.

Absorption coefficients, , for a any spectrally active molecular species, i, (e.g., water, ozone, CO2, etc.) must be computed.

is a strong function of pressure, temperature, and interaction with other species.

Inversion of this equation is highly non-linear and under-determined.

Full radiative transfer equation includes surface, down-welling, and solar reflection terms.

10

Retrieval is a minimization of cost function, optimized for the instrument

Covariance of observed minus computed radiances: includes instrument noise model and spectral spectroscopic sensitivity to components of X that are held constant (Physics a-priori information).

Derivative of the forward model is required to minimize J.

Covariance of product (e.g., CO, CH4, CO2) can be used to optimize minimization of this underdetermined problem. Currently we are using minimum variance approach (C = I) due to lack of knowledge of correlations.

11

Utilization of thermal product requires knowledge of vertical averaging

• Thermal instruments measure mid-tropospheric column– Peak of vertical weighting is a

function of T profile and water profile and ozone profile.

– Age of air is on the order of weeks or months.

– Significant horizontal and vertical displacements of the trace gases from the sources and sinks.

• Solar/Passive instruments (e.g., SCIA, OCO) & laser approaches measure a total column average.– Mixture of surface and near-

surface atmospheric contribution– Age of air varies vertically.

12

Sounding Strategy in Cloudy Scenes:Co-located Thermal & Microwave (& Imager)

• Sounding is performed on 50 km a field of regard (FOR).

• FOR is currently defined by the size of the microwave sounder footprint.

• IASI/AMSU has 4 IR FOV’s per FOR

• AIRS/AMSU & CrIS/ATMS have 9 IR FOV’s per FOR.

• ATMS is spatially over-sampled can emulate an AMSU FOV.

AIRS, IASI, and CrIS all acquire 324,000 FOR’s per day

13

Thermal Sounder “Core” Products(on 45 km footprint, unless indicated)Radiance Products RMS Requirement Current Estimate

AIRS IR Radiance (13.5 km) 3% < 0.2 %

AIRS VIS/NIR Radiance 20% 10-15%

AMSU Radiance 0.25-1.2 K 1-2 K

HSB Radiance (13.5 km) 1.0-1.2 K (failed 2/2003)

Geophysical Products RMS Requirement Current Estimate

Cloud Cleared IR Radiances 1.0K < 1 K

Sea Surface Temperature 0.5 K 0.8 K

Land Surface Temperature 1.0K TBD

Temperature Profile 1K/1-km layer 1K/1-km

Moisture Profile 15%/2-km layer 15%/2-km

Total Precipitable Water 5% 5%

Fractional Cloud Cover (13.5 km) 5% TBD

Cloud Top Pressures 0.5 km TBD

Cloud Top Temperatures 1.0 K TBD

14

Trace Gas Product Potential from Operational Thermal Sounders

gas Range (cm-1) Precision Interference

UTH 1400-1500 15% T(p)

O3 1025-1050 10% H2O,emissivity

CO 2080-2200 15% H2O,N2O

CH4 1250-1370 20 ppb H2O,HNO3

CO2680-795

2375-23952 ppm2 ppm

H2O,O3

SO2 1340-1380 1000% H2O,HNO3

HNO3860-920

1320-133040%25%

emissivityH2O,CH4

N2O 1250-13152180-2250

10%10%

H2OH2O,CO

CFCl3 (F11) 830-860 20% emissivity

CF2Cl (F12) 900-940 20% emissivity

CCl4 790-805 50% emissivityHaskins, R.D. and L.D. Kaplan 1993

ProductAvailable

Now

In Work

Held Fixed

15

Retrieval of Atmospheric Trace GasesRequires Unprecedented Instrument Specifications

• Need Large Spectral Coverage (multiple bands) & High Sampling (currently, we use 1680 AIRS and 14 AMSU channels in our algorithm)– Increases the number of unique pieces of information

• Ability to remove cloud and aerosol effects.• Allow simultaneous retrievals of T(p), q(p), O3(p).

• Need High Spectral Resolution & Spectral Purity– Ability to isolate spectral features → vertical resolution– Ability to minimize sensitivity to interference signals..

• Need Excellent Instrument Noise & Instrument Stability– Low NEΔT is required.– Minimal systematic effects (scan angle polarization, day/night

orbital effects, etc.)

16

Radiances versus Products

Assimilate Radiance Assimilate CO2 ProductProduct volume is large:

In practice, a spectral subset (10%) and spatial subset (5%) of the observations is made

Product volume is small: all instrument channels can be used to minimize all parameters (T,q,O3,CO,CH4,CO2,clouds,etc.)

Instrument error covariance is usually assumed to be diagonal, for apodized interferometers (e.g. IASI) this is not accurate.

CO2 product error covariance has vertical, spatial, and temporal off-diagonal terms.

Require fast forward model, and derivative of forward model.

Most accurate forward model is used with a model of detailed instrument characteristics.

Small biases in T(p), q(p), O3(p),due to representation error, have large impact on derived CO2.

A-priori used in retrieval may be different than assimilation model. We used very simple a-priori to assess ret. skill.

Tendency to weight the instrument radiances lower (due to representation error) to stabilize the model. Need correlation lengths to stabilize model horizontally, vertically, and temporally.

Retrieval weights the radiances as high as possible, since determined state is on instrument sampling “grid.”

17

Example of Carbon Monoxide Products from AIRS

in collaboration with

W. Wallace McMillin & Michele McCourt

University of Maryland, Baltimore County

UMBC

W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

18

The CO (& O3) Product Improves Utility of the CH4 and CO2 Products

• CO can be used to help us distinguish combustion sources (fossil or biomass) from other sources/sinks in our methane and carbon dioxide products.

• CO can be used to estimate horizontal and vertical transport during combustion events.

• CO (and Ozone) may help us improve atmospheric vertical transport models in the mid-troposphere.

• Inter-comparison w/ aircraft (e.g., ESRL/GMD, INTEX-A, INTEX-B), and MOPITT products is in-work.– McMillan et al. 2005. GRL v.32 doi:10.1029/2004GL0218– McCourt, PhD dissertation, UMBC, 2006

19

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and TrajectoriesJuly 2004 Fires in Alaska

20

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

21

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

22

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

23

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

24

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

25

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

26

500 mb700 mb850 mb

UMBC W. McMillan (mcmillan@umbc.edu) AIRS Science Team Meeting, Cal Tech, 3/8/06

AIRS CO and Trajectories

CO from northern Alaskan fires was transported to the lower atmosphere in SE of US

CO from southern Alaska Fires was transported to Europe at high altitudes (5 km)

27

Example of Methane Products from AIRS

28

AIRS CH4 Kernel Functions are Sensitive to H2O(p) & T(p)

Polar Mid-Latitude Tropical

Isothermal vertical structure weakens sensitivity.

moisture optical depth pushes peak sensitivity upwards.

29

Also providing the vertical information content to understand CH4 product

AIRS mid-trop measurement column

CH4 total column f/ transport model (Sander Houweling, SRON)

Fraction Determined from AIRS Radiances

Peak Pressure of AIRS Sensitivity

30

Comparison of CH4 product &

ESRL/GMD Continuous Ground Site

Barrow Alaska

3deg. x 3deg. gridded retrieval averaged over 60-70 lat, & -165 to -90 lng

31

Comparisons of AIRS product to ESRL/GMD Aircraft Observations

ESRL/GMD aircraft profiles are the best validation for thermal sounders since they measure a thick atmospheric layer.

32

ESRL/GMD Flask Data from Poker Flats, Alaska: Seasonal cycle is a function of altitude

Surface Flasks (Barrow)

1.5 km 850 mb

5.5 km 500 mb

7.5 km 385 mb

33

We need to determine how much of our CH4 signal is from stratospheric air

Dobson-Brewer circulation

Can depress high latitude, high altitude methane signals in winter/spring time-frame.

UT/LS region in high latitudes has “older” air.

We will explore tracer correlations to unravel surface

vs stratospheric

sources. (working w/ L. Pan, NCAR)

34

Example of Carbon Dioxide Products from AIRS

35

LW Thermal CO2 Kernel Functions are also Sensitive to H2O, T(p), & O3(p).

Mid-Latitude

TPW = 1.4 cm

Polar

TPW = 0.5 cm

Tropical

TPW = 2.5 cm

Isothermal vertical structure weakens

sensitivity

moisture optical depth pushes peak sensitivity upwards

36

Also providing the vertical information content & comparing CO2 product with models

AIRS mid-trop measurement column

CO2 Transport ModelRandy Kawa (GSFC)

Fraction Determined from AIRS Radiances

Averaging Function Peak Pressure

37

Preliminary Comparisons to ESRL/GMD aircraft

• Comparison of AIRS & ESRL/GMD observations..• Usually 5 hour time

difference• Limit retrievals within 200

km of aircraft.• Spot vs. regional sampling

• Retrieval is average of “good” retrievals• 3 – 50 ret’s are used in

each dot. = ± 3.1 ppmv,

correlation = 0.83• Investigation of outliers is

in-work.

38

Comparisons to JAL aircraft observations & ESRL/GMD MBL model

Matsueda et al. 2002 aircraft – single level measurement

ESRL/GMD Marine Boundary Layer Model (surface measurement)

AIRS CO2 retrieval from GRIDDED dataset – mid-trop thick layer measurement

JAL Aircraft data provided by H. Matsueda

39

Same as before, but in-situ CO2 adjusted by a-priori in retrieval

JAL Aircraft data provided by H. Matsueda

In-situ data is adjusted by the % of a-priori in our regularized retrieval. This depresses the seasonal amplitude of the in-situ data and is a gauge of our retrieval performance.

Differences should exist between single level (surface or altitude) observations and AIRS thick layer observations.

40

Again, to what extent does stratospheric age of air play a role?

Model runs of Brewer-Dobson circulation effect are courtesy of Run-Lie Shia, Mao-Chang Liang, Charles E. Miller, andYuk L. Yung, California Inst. Of Technology & NASA Jet Propulsion Lab.

Surface measurements (dashed line) and model of 500 mb concentration (solid line) can differ by 5 ppm

(NOTE: Vert. Scale is 350-385 ppm)

Brewer-Dobson effect has altitude, latitude, and seasonal variation with a maximum in northern winter/spring.20022000 2004

41

NOAA AIRS CO2 Product is Still in Development

• Measuring a product to 0.5% is inherently difficult– Empirical bias correction (a.k.a. tuning) for AIRS is at the 1 K level and can

affect the CO2 product.– Errors in moisture of ±10% is equivalent to ±0.7 ppmv errors in CO2.– Errors in surface pressure of ±5 mb induce ±1.8 ppmv errors in CO2.– AMSU side-lobe errors minimize the impact of the 57 GHZ O2 band as a

T(p) reference point.

– Bottom Line: Core product retrieval problems must be solved first.

• Currently, we can characterize seasonal and latitudinal mid-tropospheric variability to test product reasonableness.

• The real questions is whether thermal sounders can contribute to the source/sink questions.– Requires accurate vertical & horizontal transport models – Having simultaneous O3, CO, CH4, and CO2 products is a unique contribution

that thermal sounders can make to improve the understanding of transport.

42

… And many groups are working on AIRS CO2 algorithms

P.I. Methodology Type of Scenes

Temperature/CO2

Separation

Alain Chédin & Cyril Crevoisier

Neural Network Clear 57 GHz O2 (Aqua/AMSU)

Richard Engelen 4DVAR Clear 57 GHz O2 (all AMSU’s)

& radiosonde

Moustafa Chahine Partial Vanishing Derivatives

(unconstrained LSQ)

Clear & Cloudy

Multi-spectral (15 vs 4 µm) and 57 GHz (Aqua/AMSU)

Larrabee Strow Unconstrained LSQ Clear T(p) = ECMWF

Chris Barnet Regularized LSQ

(optimal estimation)

Clear & Cloudy

Multi-spectral (15 vs 4 µm) and 57 GHz (Aqua/AMSU)

43

CH4

CO

CO2

O3

CH4 & Tsurf

AIRS measures multiple gases (and temperature, moisture and cloud products) simultaneously

• 29 month time-series of AIRS trace gas products: Alaska & Canada Zone (60 lat 70 & -165 lon -90)

• July 2004 Alaskan fires are evident in CO signal (and CH4 ??)

• Seasonal methane is correlated to surface temperature (wetlands emission?)

• We have begun to investigate correlations that should exist between species (e.g., O3, CO, CH4 interaction)

Aug.2003 Mar.2006

44

29 month time-series of AIRS productsEastern China Zone (20 lat 45, 110 lon 130)

Correlation of CH4 with skin temperature is significantly smaller

45

29 month time-series of AIRS productsSouth America Zone (-25 lat EQ , -70 lon -

40)

Biomass burning

Correlation of CH4 with skin temperature is non-existent

46

For more information:http://www.orbit.nesdis.noaa.gov/smcd/spb/airs/index.html

• Trace GAS web-pages allow a quick look at the trace gas products as a function of geography, time, and comparisons with in-situ datasets.

USERID & PASSWORD

Request via e-mail:

chris.barnet@noaa.gov