anna serdyuchenko, victor gorshelev, wissam chehade, john p. burrows, mark weber

Harmonization of GOME, SCIAMACHY, GOME-2

ozone cross-sections: final results from laboratory

measurementsAnna Serdyuchenko, Victor Gorshelev,

Wissam Chehade, John P. Burrows, Mark WeberUniversity of Bremen, Institute for Environmental Physics

HARMONICS Final Meeting - ESRIN, Frascati

1. HARMONICS project goals

2. Measurements of cross-section in laboratory: setup

3. Results analysis

4. Outlook

Agenda


1. HARMONICS project goals

HARMONICS project goalsNew measurements of cross-section in

laboratoryResults analysis


Cross-section issues in satellite DOAS: direct comparisons with BMD 326.6-334.5 nm

Direct comparison by fitting cubic polynomial (DOAS type), a wavelength shift, and Gaussian slit function to match DBM to other cross-sections: SCIAMACHY FM requires a scaling of +5% and GOME2 FM of +7% wrt to GOME in TO3 retrieval

(Eskes et al., 2005, Lerot et al., 2009, Weber et al., 2011) There are significant shifts between cross-sections (FMs and Bass-Paur),

▪ Note: total ozone change is -6DU/0.01nm shift (or -2%/0.01 nm) !!! advantage of satellite FMs: ILS does not need to be known disadvantage of satellite FM: Measurements were done at external facility under high time pressure

(satellite calibration period)

T[K] Shift [nm]

Scaling [-]

Gaussian FWHM, [nm]

FWHM from solar fits, [nm]

Scaling wrt. GOME FM

Burrows et al. 1999GOME FM

225240

+0.017(2)

+0.017(2)

1.027(2)1.023(2)

0.158(5)0.159(6)

0.165(8) --

Bogumil et al. 2003SCIAMACHY FM

225240

+0.008(2)

+0.009(2)

0.970(3)0.973(3)

0.222(6)0.219(6) 0.202(13) -5.6%

-4.9%

Gür et al., 2005GOME2 FM3

225240

-0.031(3)-0.031(3)

0.960(8)0.949(7)

measured ILS

asymmetric ILS(~0.29 nm)

-6.5%-7.3%

Bass & Paur, 1985(original, MPI

database)225240

+0.020(3)

+0.023(5)

1.015(3)1.000(3)

0.079(10)0.087(10) - +1.2%

+2.3%

Bass & Paur, 1985(ACSO web page,

NASA)225240

+0.024(4)

+0.021(4)

1.004(3)1.004(3)

0.084(12)0.078(11) - +2.3%

+1.9%

Weber et al., 2011


Retrieval issues related to cross-sections

SCIAMACHY total O3 retrieval (using SCIAMACHY FM reference spectra) were 5% higher than GOME (with GOME FM reference spectra) in the range 325-335 nm (5% direct cross-sections scaling wrt. GOME FM).

GOME2 total O3 retrieval (using GOME2 FM3) is 9% higher than calculated with resolution adjusted GOME FM. (7% direct cross-sections scaling wrt. GOME FM).

Harmonisation of O3 and NO2 FM cross-sections from GOME and SCIAMACHY for a consistent retrieval.

Two approaches: reanalysis of data from the CATGAS campaigns;new laboratory measurements.


Work plan for HARMONICS

Re-analyse the satellite FM cross-section data based upon available raw data from the CATGAS campaigns (WP 100) GOME FM: no raw data available will remain unchanged SCIAMACHY FM: raw measurement files are available reanalysis GOME-2 FM3: raw measurement files are available reanalysis NO2: no need for change (stable gas conditions)

Note: FM cross-section data are relative measurements (need to be scaled to reference data, here : BDM/BassPaur, alternative: new IUP)

New laboratory measurements at IUP at high spectral resolution (UV to near IR)

Verification of new results and total ozone retrieval tests HARMONICS Final Meeting - ESRIN, Frascati

Current status of the available ozone cross-sections datasets

Importance of the cross-sections set choice is vigorously discussed last years within the ozone observing community.

Quantum mechanical approach: a potential of producing a noise-free cross-sections. however, the quality of the ab initio calculations is still inferior to the experimental data

accuracy. a comparison of calculated spectra with experimental high resolution data obtained in a

wide spectral range at different temperatures should play an important role in such studies.

Experimentally obtained ozone absorption cross-sections from pioneer works (1932) until most recent studies: online spectral atlas of gaseous molecules of the Max‑Planck Institute for Chemistry, Mainz.

A large-scale initiative to review and recommend ozone cross-sections for all of the commonly used atmospheric ozone monitoring instruments was started in spring 2009. Absorption Cross-sections of Ozone (ACSO) committee: a joint commission of the

Scientific Advisory Group of the Global Atmosphere Watch of the World Meteorological Organization (WMO-GAW) and the International Ozone Commission (IO3C) of the International Association of Meteorology and Atmospheric Sciences.HARMONICS Final Meeting - ESRIN, Frascati

Current status of the available ozone cross-sections datasets1. The high-resolution broadband data obtained by Bass and Paur (BP) :

cover spectral region of 245 nm to 343 nm only. currently included in the standard ozone total column and profile retrievals using ground-

based (Brewer, Dobson) and satellite (SBUV, TOMS V8) spectrometers. included in the latest version of the high-resolution transmission molecular absorption

database HITRAN 2008; however, a wavelength shift must be applied.

2. The high-resolution broadband data by Brion, Daumont, Malicet (BMD) available for 195 nm - 830 nm for room temperature and for 194.5 nm - 520 nm for lower

temperatures down to 218K. subject of debates on substituting them for Bass and Paur data within the WMO-GAW and

IGACO community and in the next version of the HITRAN dataset. Currently data are used on the OMI and GOME GDP5 retrievals.

3. The high-resolution broadband cross-sections obtained by Bremen team Temperature-dependent absorption cross-sections at230-830 nm were recorded with a

Fourier-transform spectrometer.

4. Measurements at single wavelengths: Axson (2011), Anderson (1993), Enami (2004), El Helou (2005), Peterson (2012)



Recommendation on using the dataset on existing instruments and future missions is a subject for research of many ozone investigating groups and nets around the world.

Nevertheless, few obvious criteria can be formulated:

spectral resolution at least order of magnitude better than the instrumental bandwidths of modern remote sensing spectrometers;

wide spectral region to include spectral channels and windows of as many instruments as possible;

availability of high quality experimental data at several temperatures going as low as 190K (ozone hole conditions).



Despite of the high quality of the already available data experimental works on the ozone absorption spectra continue.

There are few important issues to be addressed:

improvement of the absolute accuracy of the cross-section up to 1-2% at least in the Huggins band to answer the requirement to measure small changes in stratospheric and tropospheric ozone;

improvement of the wavelength calibration;

improvement of the temperature dependence in the different spectral regions, especially in the Huggins and Wulf bands;

measurements of the weak absorption in the bottom of the Huggins and Chapuis bands and in the Wulf band NIR region longer than 900 nm, which were poorly covered so far.


Demands on a modern dataset

wavelength coverage UV/VIS/NIR 200 – 1000 nm;vacuum wavelength accuracy ~ 0.001 nm;spectral resolution ~ 0.02 nm;absolute intensities accuracy: 2%;temperature: 193K-293 K, step 10K


2. Cross-section measurements in laboratory: set-up




Ozone cross sections at 213- 1100 nm

Huggins

Hartley

Chappuis Wulf

Spectral region, nm213 – 290

Top Echelle

290 – 320

UV I FTS

320 – 350*

UV II FTS

350 – 450

Bottom Echelle

450 – 500

Visible FTS

500 – 800*

NIR FTS

800 – 1100

IR FTS

*absolute measurements


Main parameters of experimental set-ups


Main parameters of experimental set-ups

Setup VIS/IR Setup UV/VIS

Spectrometer Fourier Transform Echelle (‘cross dispersion’)

Source Xe and Tungsten lamps Xe and D2 lamp

Detector Si/GaP photodiode ICCDResolution, FWHM 0.02 nm @ 300 nm 0.02 nm @ 300 nmWavelength region 290– 1100 nm 212 nm – 600 nm

Acquisition time Slow (tens of minutes) Fast (minutes)

Wavelength calibration

Excellent(0.0005 nm in UV)

Excellent (agrees with NIST Hg line at 253 nm better than 0.001 nm )

Absorption path 135 and 270 cm 5 cm, 135 cm – 30 mCooling Double jacket quartz cell, pre-cooler, cryogenic cooling


LNeII 0)(

AOD )()(

Accuracy and precision of the new data

I / I0 transmitted intensity with /without ozone

L absorption path length, (1 – 20 m)N / p ozone density / pressure (1- 50

mbar)k Boltzmann constantT Temperature (200 – 300 K)

kTpN ,LNA Art of uncertainty SourceStatistical uncertainty: 0.1-0.5%, depends on spectral regions

Reproducibility of spectra intensities I, I0:- Source intensity drift;- White noise;- Fluctuations of temperature;- Fluctuations of pressure.

Systematic uncertainty: better than 3%, independent on spectral regions

Absolute calibration parameters p, T, L, A:- Purity of oxygen and ozone (leaks, ozone decay);- Accuracy of temperature sensors;- Temperature sensors off-set;- Absolute accuracy of the cell length.- Calibration procedure: Least squares fit coefficients

accuracy


Accuracy and precision of the published data

Data Scaling method

Uncertainty, %

Statistical Systematic Total

Hearn, 1961 Absolute , pure ozone - - 2.1

BP, 1980 – 1985 Using Hearn 1 ? 2.3

Bogumil (SCIAMACHY), 2003

Using BP - - 3.1*

Burrows (GOME), 1999

Absolute, titration

Lamp drift < 2 2.6 2.6 – 4.6

Voigt, 2001 Integrated GOME 3 – 6

BMD, 1992 - 1998

Absolute (pure ozone) 0.9 –2

1.5 (420 –830 nm) 1.5–4 (350 –420 nm)

1.8 –2.51.8 –4.5

*excluding regions with the cross-section below 10-23 cm2/molecule (365 – 410 nm and longer than 950 nm) and 305-320 nm


3. Cross-section measurements in laboratory: results analysis




Huggins (DOAS)

Hartley

Minimum

Chappuis

NIR

IR


Wulf

Results analysis

Polynomial dependence on temperature:• Bass-Paur parameterization (2nd order polynomial) for selected wavelengths in Huggins band

Band-integrated cross-sections as functions of temperature (insensitive to resolution)• Hartley, Huggins, Chappuis bands

Comparison with existing datasets• High resolution datasets: Bass – Paur , Brion et al : wavelength shift, absolute scaling factor

• Low resolution satellite datasets: resolution matching, wavelength shift, absolute scaling factor

Total ozone retrieval (Wissam Chehade)

Please note: comparison between cross-sections is not the same

thing as comparison between retrievals results!!!!


Results analysis

Nowadays, spectral channels of the satellites-based (GOME, GOME-2, SCIAMACHY, SAGE, SBUV, TOMS, OMI, OSIRIS) and ground based-instruments (Brewer, Dobson) measuring ozone and other trace gases cover a wide spectral range from near UV to the visible and IR.

Ozone absorption spectrum in near UV – near IR affects channels for detection of other traces gases, aerosols and clouds.


Band-integrated cross-sections

T, K

245 – 340 nm Hartley band, x 10-16

325 – 340 nmHuggins band, x 10-20

410 – 690 nmChappuis band, x 10-19

663 – 1000 nmWulf band, x 10-19

Mean * New** Mean New Mean New New203

3.53 ± 1.2% 3.52 5.47 ± 3.3% 5.46 6.48 ±

1.6% 6.30 0.991223

3.53 ± 1.1% 3.53 5.72 ± 2.6% 5.74 6.35 ±

2.4% 6.31 0.999243 3.54 ± 1% 3.53 6.23 ±1.9% 6.30 6.35 ±

1.7% 6.33 1.006273 3.55 ± 1% 3.54 7.26 ± 1.3% 7.35 6.44 ±

2.3% 6.38 1.028293

3.55 ± 0.6% 3.53 8.23 ± 1% 8.30 6.38 ±

1.6% 6.36 1.04HARMONICS Final Meeting - ESRIN, Frascati

Results verification

Huggins (DOAS)

Hartley

Minimum

Chappuis

NIRIRMean values are taken from

Orphal, J. (2003) A critical review of the absorption cross-sections of O3 and NO2 in the ultraviolet and visible. J. Photochem. Photobiol. A 157, 185-209;

Estimated experimental uncertainty is ~2%

Band-integrated cross-sections: temperature dependence


Hartley

Huggins

Chappuis

Wulf

Comparison on wavelengths of remote sensing and ground based instruments


Instrument

FWHM Channels

Dobson 1 - 3 306/325, 312/332, 318/340Brewer 0.6 306, 310, 314, 317, 320 TOMS V8 1 312, 318, 331, 360OMI 0.42 -0.63 264 - 504OSIRIS 1 274 – 810 SBUV 1.13 256, 274, 283, 288, 292, 298, 302, 306, 313, 318,

331, 340SAGE III 1.2 – 2.5 433-450, 53-622, 759-771, 933-960, 385, 676, 758,

869, 1020

The spectral resolution of diverse ozone observing instruments changes:―with wavelength from the UV to the near IR―from channel to channel for each instrument.


BMD and BP – interpolated at 293K;All datasets – resolution downgraded 1 nm

Comparison on wavelengths of remote sensing and ground based instruments

Comparison at Hg lamp and He-Ne laser lines, 293K

[nm]Mean Value, x 10-20

J. Orphal, 2003

New data*, x 10-

20

253.65 1141 ±0.9% 1126 ± 1.09%289.36 149 ±2.0% 154 ± 0.67%296.73 60.3 ±1.6% 61.60 ± 0.54%302.15 29.2 ±1.8% 30.40 ± 0.48%543.52 0.0314 ±1.3% 0.0316 ± 0.25%576.96 0.0477 ±0.8% 0.0485 ± 0.20%

594.10 0.0470 ±1.2% 0.0474 ± 0.18%604.61 0.0522 ±1.0% 0.0526 ± 0.2%611.97 0.0466 ±0.7% 0.0468 ± 0.21%632.82 0.0346 ±1.2% 0.0347 ± 0.18%

* Uncertainty of the experimental OD


Temperature dependence in UV region at 253.65 nm (Hartley band) Very weak temperature dependence (within 1%) Very good agreement with BMD, BP, GOME, SCIAMACHY FM revised, GOME-

2 revised


Huggins band, DOAS window


Temperature dependence on selected wavelengths in Huggins band

No resolution matching‘Bass-Paur ’parametrisation

328 nm “hill”, 330 nm “valley”

Black– New Data, Blue – BMD, Magenta– SCIAMACHY


New data vs BMD and SCIAMACHY (293K)


BP – Bass Paur, BMD – Brion et al

2nd order polynomial inter/extrapolation 193-293K with step of 10K

inter/extrapolation on the common grid with step of 0.01 nm no resolution matching needed mean absolute deviation:

Green – comparison BP versus new experiment (325-340 nm);

Red – comparison BMD versus new experiment (325-345 nm);

Black – comparison BP versus BMD

Comparison in Huggins band region: new data versus high resolution datasets, relative difference and shifts

BP – BMD

BMD– New Data

BP –New Data

Shift, nm 0.02 0.005 0.02Mean absolute deviation

3% 1-2% 2-3%


Comparison in Huggins band region: new data versus high resolution datasets, relative difference and shifts


1st dataset

2nd dataset

Shift, nm

Mean integral difference, %

2nd dataset

Shift, nm


2nd dataset

Shift, nm


this work

BPexp +0.019 2.1/0.959 BMDexp -0.012 1.6/0.985 Bogumil 0.003 1.5

this work

BPcalc +0.017 1.5/0.992 BMDcalc -0.015 1.4/0.997 Voigt -0.006 1.04/0.979

this work

BPHITRAN +0.002 1.5/0.993 - - - - - -

BPHITRAN BPcalc 0.015 0.01/1 BMDcalc -0.014 1.3/0.998 Voigt 0.007 1.1/1.013BPHITRAN BPexp 0.017 0.4/0.998 BMDexp -0.010 1.3/0.998 - - -

BPcalc BPexp 0.002 0.4/0.998 BMDcalc -0.029 1.3/0.998 Voigt -0.022 1.1/1.013BPexp - - - BMDexp -0.025 1.4/0.976 Voigt -0.025 1.8/1.023

BMDcalc - - - BMDexp -0.005 0.02/1 Voigt 0.007 0.9/1.018BMDexp - - - - - - Voigt 0.002 1.2/0.993

Comparison in DOAS region: new data versus SCIAMACHY (old and revised)

Data convolved with Gaussian slit function

Disagreement ~ 1 – 5 % for 241K, 273K and 293K, increasing at lower temperatures.

Wavelength shift of -0.005 to 0.015 nm.

Deviation: within the 2% accuracy limits for 243K, 273K and 293K, increase up to 5% for 203K, 223K.


Chappuis and Wulf bands


Direct comparison in the ‘minimum’


SCIA FM revised

Chappuis band

GOME-2 FM3 revised



Wulf Band

Wulf Band

SCIA FM revised


Anderson, Mauersberger, Journal of geophysical research 100(D2),

3033-3048, 1995

Weak absorption, measurements are very sensitive to the baseline and S/N, low accuracy and precision -> FTS upgrade is needed;

Temperature dependence in NIR; Region is interesting for future

missions (SAGEII and SAGE III)


Wulf Band

Advantages and limitations of the new dataset

High resolution: 0.02 nm@300 nm – 0.1 nm @ 1000 nm

11 temperatures, down to 193K Spectral coverage UV- vis – NIR for all temperatures Accurate wavelength calibration, better than 0.001

nm

Absolute accuracy: 2 – 3% in UV/ visible, 30% for 350 – 450 and 900 – 1050 nm


Current O3 datasets in useData set Temperatures, K Resolution,

nmWavelength, nm

GOME FM Burrows et al., 1999

202 221 241 273 293

0.17 @ 330 nm 230 – 800

SCIAMACHY FM Bogumil et al., 2003

203 223 243 273 293

0.20 @ 330 nm 230 – 1070

GOME2 FM Spietz et al., 2005

203 223 243 273 293

0.29 @ 330nm 240 – 790

Paur and Bass, 1985

203 218 228 243 273 298

<0.025 nm(?) 245 – 345

Malicet et al. 1995 Brion et al., 1993Daumont et al., 1992

218 228 243 273 295

0.01-0.02 nm

195 – 345

UV-FTSVoigt et al., 2001

203 223 246 280 293 0.03 @ 230

nm 230 – 850

IUP 2011Serdyuchenko, Gorshelev 2011

193 203 213 223 233 243 253 263 273 283 293

0.02 @330 nm

213 – 1050

http:/

http://www.iup.uni-bremen.de/gruppen/molspec/databases/index.html


http://www.iup.uni-bremen.de/gruppen/molspec/databases/index.html

Current status

Summer 2009: Start of serial measurements

Summer 2010-Winter 2011: Absolute calibration

Spring 2011: Retrievals

Autumn 2011 -Winter2011/2012:Publications preparation

Spring 2012: data release

Starting from summer 2012:1. Feedback from ozone investigating groups and organizations

2. Contact with HITRAN database


Live after HARMONICS


―CH4 line parameters investigation (winter 2011-2012) (ESA

ADVANSE);

―Lab upgrade: dual-channel FT spectrometer (summer 2012)

(University of Bremen) ;

―Possible new measurements (starting from autumn 2012) :

Ozone: NIR and 10 mkm – staff support needed;

SO2 for UV sensors (320 nm) – staff support needed.

Special thanks to former IUP members: Peter Spietz and Juan Carlos Gomez-Martin

Thank you for attention !


anna serdyuchenko, victor gorshelev, wissam chehade, john p. burrows, mark weber

Documents

gome fm reference spectra

gome fm225240

gome fmburrows

no2 fm crosssections

unchangedsciamachy fm

reference data

ozone crosssections

sciamachy fm reference