Global Monitoring of Tropospheric Pollution from Geostationary OrbitKelly Chance1, Thomas P. Kurosu1, Xiong Liu2,3, Alfonso Saiz-Lopez1, Doreen O. Neil2, James J. Szykman2,4,

Jack Fishman2, R. Bradley Pierce5, James H. Crawford2, David Edwards6, Gary Foley4, and Rich Scheffe4

1CfA 2NASA 3UMBC 4EPA 5NOAA 6NCAROutline

• Introduction and motivation• Descriptions of current satellite instruments• Measurement requirements

- Geophysical, spatial, temporal requirements- UV/visible gas concentrations

• Scalable strawman• Orbital considerations (not part of the strawman)• Future work – The two outstanding requirements

Acknowledgement: Development at the Harvard-Smithsonian Center for Astrophysics has been supported by the Smithsonian Institution.

SO2 in Kilauea activity, source of the VOGevent in Honolulu, 9 November, 2004 (OMI)

1.5 to 3 days40×40(40×80 wide-swath, 40×10

zoom)

0.24 – 0.53240-790Linear ArraysGOME-2a(2006)

Daily50×50 250×250(depending on

product)

0.42 – 1.0250-3802-D CCDsOMPS-1(2010?)

Daily15×30 – 42×162(depending on swath position)

0.42 – 0.63270-5002-D CCDsOMI(2004)

6 days30×30 30×60 30×90 30×120 30×240 (depending on

product)

0.2 – 1.5239-2380Linear ArraysSCIAMACHY(2002)

3 days40×320(40×80 zoom)

0.2 – 0.4240-790Linear ArraysGOME(1995)

Global Coverage

Ground Pixel Size [km2]

Spectral Resolution [nm]

SpectralCoverage [nm]

DetectorsInstrument

GOME-1/ SCIAMACHY/ OMI/GOME-2/OMPS nadir

Previous Experience (since 1985 at SAO): Scientific and operational measurements of O3, NO2, SO2, HCHO, and CHOCHO (and BrO, OClO, IO, H2O).

Fitting UV/Visible Trace SpeciesFitting UV/Visible Trace Species

•• Requires precise (Requires precise (dynamicdynamic) wavelength (and often slit ) wavelength (and often slit function) calibration, Ring effect correction, undersampling function) calibration, Ring effect correction, undersampling correction, and proper choices of reference spectra (correction, and proper choices of reference spectra (HITRAN!HITRAN!))

•• Best trace gas column fitting results (NOBest trace gas column fitting results (NO22, HCHO, CHOCHO) , HCHO, CHOCHO) come from come from directlydirectly fittingfitting L1b radiancesL1b radiances

•• Best tropospheric OBest tropospheric O33 and SOand SO22 from direct profile retrievals from direct profile retrievals using optimal estimationusing optimal estimation

•• Remaining developments:Remaining developments:1.1. Tuning PBL OTuning PBL O33 from UV/IR combination (demonstrated for from UV/IR combination (demonstrated for

the OMI/TES combination by SAO + JPL)the OMI/TES combination by SAO + JPL)2.2. Tuning direct PBL SOTuning direct PBL SO22 from optimal estimation (underway)from optimal estimation (underway)

Tracking of most urban diurnal variation

4.0×1014CHOCHO

Distinguish clean from moderately polluted scenes

1.0×1016HCHO

Distinguish structures for anthropogenic sources

1.0×1016SO2

Distinguish clean from moderately polluted scenes

3.0×1015NO2

~10 ppbv in PBL; reality (profiling) more complicated

2.4×1016O3

Sensitivity DriverVertical column (cm-2)

Molecule

Required Concentrations*

*In PBL. One of two issues needing the most work(traceability from AQ requirements and modeling)

Introduction and Motivation

The target tropospheric gases are O3, NO2, SO2, HCHO,CHO-CHO (plus CO, O3 and CH4 in the IR). Plus aerosols.

The aims are:1. To retrieve tropospheric gases from geostationary orbit

at high spatial and temporal resolution.2. To integrate the results into air quality prediction,

monitoring, modeling, and climatological studies.This follows from our successful developments (since

1985, with SAO as U.S. investigator) of SCIAMACHY, GOME-1 and GOME-2; participation in OMI, and collaboration on OMPS design. *

Successful retrievals have involved development of algorithm physics coupled with chemistry and transport modeling* and multiple-scattering radiative transfer calculations. With several minor exceptions (below) this development has been done and, in most cases, made operational

15o - 50o N, 60o - 130o W (parked at 0o N, 95o W)Measure solar zenith angles from 0o – 70o

Effective solar zenith angles (ESZAs) = 17.6o – 76.0o

Geostationary Minimal Case:Geostationary Minimal Case:Scalable Strawman Scalable Strawman -- 11

OMI Tropospheric NO2 (July 2005) GOME-1 HCHO (Fu et al., 2007)Monthly mean HCHO columns over Asia as observed by GOME from 1996 to 2001 (left panels) and as simulated by GEOS-Chem for 2001 (right panels).

An alternative(not in baseline): Inclined 24 hour orbits!

Better viewing zenith angles at high latitudes

Possibility to measure same location at different VZAs profile information(Thanx, RVM!)

Radiative Transfer Modeling and Fitting Studies

Note cloud windows:

Use of Raman scatteringand of the oxygen collision complex

O2 A band @ 762 nm not in baseline design, to keep it small and simple

Little Chappuis band coverage: Potential implications for PBL O3(NB NASA GEO O3 study)

423-451325-357

315-325423-451

315-335

Fitting window (nm)

1.5×10144.0×1014CHOCHO2.3×10151.0×1016HCHO

1.5×10151.0×1016SO2

1.1×10153.0×1015NO2

5.0×10152.4×1016O3

Slant column (cm-2)

Vertical column (cm-2)

MoleculeMeasurement Requirements

The slant column measurement requirements come from full multiple scattering calculations, including gas loading, aerosols, and the GOME-derived (Koelemeijer et al., 2003) albedo database, and assume a 1 km boundary layer height.

Scalable Strawman Scalable Strawman -- 22

Scalable Strawman Scalable Strawman -- 33

Lat/lon limits are ~3892 km N/S and 7815-5003 km E/W (6565 average), or about 390×657 10×10 km2 footprints.

– Measure 400 spectra N/S in two 200-spectrum integrations (each on two 10242 detector arrays – 1 UV and 1 visible).

– 2.5 seconds per longitude (2×1 s integration, 0.5 s step and flyback) total sampling every < ½ hour (27 min).

Detectors: Rockwell HyViSi TCM8050A CMOS/Si PIN– 3×106 e- well depth; will need several rows (or readouts)

per spectrum to reach the necessary statistical noise levels.

– Complicated by brightness issues; can’t always have full wells.

• 200 spectra on each of two 10242 arrays; each spectrum uses 4 detector rows (800 total out of 1024).

– Channel 1: 280-370 nm @ 0.09 nm sample, 0.36 nm resolution (FWHM).

– Channel 2: 390-490 nm @ 0.1 nm sample, 0.4 nm resolution (FWHM); includes O2-O2 @ 477 nm.

– 4 samples per FWHM virtually eliminates undersampling for a symmetric instrument transfer (slit) function [Chance et al., 2005].

• Pointing to 1 km = 1/35,800 = 6 arcsecond.• Size optics to fill sufficiently in 1 second (≈ 1 cm2 (GOME

size) × √1.5 (GOME integration time) × 35,800 km / 800 km = 55 cm “telescope” optics). More realistically ….

Sizing for 10×10 km2 Footprint,1 Second Integration Time

40.71.98×1063.56×10-44.85×1046.22×1012CHOCHO20.88.23×1055.51×10-43.97×1045.65×1012HCHO0.234.76×1037.25×10-32.06×1042.94×1012SO2

0.0633.09×1038.99×10-34.87×1046.25×1012NO2

5.091.28×1051.40×10-32.51×1043.57×1012O3

a×Effnϕ/4RMSϕ cm-2 px-1⟨Rad⟩Mol

• ⟨Rad⟩: Minimum clear-sky radiance, cross-section weighted (photons s-1 nm-1 sr-1 cm-2)

• ϕ cm-2 px-1: # photons cm-2 pixel-1 @ instrument in 1 second; 10×10 km2 7.80 ×10-8 sr solid angle

• RMS: Fitting RMS required for the minimum detectable amount = 1 / required S/N

• ϕ px-1: # photons pixel-1 needed in 1 second to meet RMS-S/N requirements; includes factor of 4 for 4 detectors rows per spectrum

• a×Eff: Telescope collecting area (cm2) × overall optical efficiency

Formaldehyde (HCHO) is the driver for almost any conceivable choice of requirements! (Unless VOCs are considered unimportant, in which case O3 would be the driver, with the above as a low estimate).20.76 cm2 is a16-cm diameter telescope @ 10% optical efficiency (GOME, a much simpler instrument, is 15 – 20% efficient in this wavelength range).Also, IR needs (CO, O3, CH4) must be addressed.

Major Tradeoffs and Questions

Outstanding Needs

Tradeoffs: # samples (footprint) vs. sensitivity (S/N) vs. integration time vs. geographical coverage vs. max SZA:

– 5×5 km2 footprints in 1/2 hour with a 32 cm diameter telescope, if the instrument is 10% efficient. (Spatial resolution: rows vs. # readouts; could do 5×5 km2 on 1 chip with multiple readouts.)

– Spatial Nyquist sampling must be carefully addressed.Questions: Are lat and lon sampling necessarily the same? Is constant sampling necessary?IR: Priorities ≈ 2.4 µm CO > 9.6 µm O3 > 4.7 µm CO > CH4.

– Scanning Fabry-Perot instruments may provide a compact IR solutionOptions and extensions (near-term U.S. GEO-CAPE attention!):MODIS channels for aerosols? (TOMS absorbing aerosol index is automatic, but little else is operational.)

– OMI aerosol products should be reviewed.– Should include polarization-resolved measurements;– Several such UV channels will improve PBL O3 [Hasekamp and Landgraf,

2002a,b; Jiang et al., 2003]. Visible (Chappuis) band to further improve PBL O3? Discrete detectors? Everything is debatable; this is why it is a strawman, but we must show why alternatives are better.

1. Science Requirements (S/N, geophysical, spatial, temporal) from sensitivity and modeling studies (OSSEs), providing traceability for AQ forecast improvement and other uses.– Unless things change a lot, HCHO will be the driver for instrument

requirements. Then address trade space.2. Instrument Design. Reducing “smile”, enabling multiple readouts, increasing

efficiency, optimizing ITF shape ….– GEO instrument is not just a super-OMI/OMPS with CMOS/Si detectors

instead of CCDs. Minimal geostationary requirements imply scanning instead of a pushbroom and they imply getting many more spectra onto a rectangular detector than OMI and OMPS have obtained.

– Instrument optical and spectrograph design, including fully-informed choice of detector type, is the single most important outstanding issue in demonstrating the feasibility of geostationary pollution measurements. NB PBL O3 instrument drivers!

O3 profiles/tropospheric ozone fromGOME-1: Biomass burning andozone hole orbit, October 1997

OMI glyoxal (CHO-CHO)

OMI measurements of troposphericNO2 over the United States

OMI formaldehyde (HCHO)

VE

TAS

R I

Best fitting:2.0×10-4 full-scale radiance

(Instrument vs. algorithm telescope optics size

floor)