Page 1 1 of 25, AGU Fall Meeting Dec 17 2008

Vijay Natraj (Caltech), Hartmut Bösch (Leicester), Rob Spurr (RT Solutions),

Yuk Yung (Caltech)

AGU Fall MeetingAGU Fall Meeting

December 17, 2008December 17, 2008

Glint and Target Mode Simulations for the Orbiting Carbon Observatory

Page 2 2 of 25, AGU Fall Meeting Dec 17 2008

Project and Mission Overview

Salient Features:• High-resolution, three-channel grating spectrometer• Partnership with HS (Instrument) and OSC (Spacecraft)• High heritage spacecraft, flies in formation with the A-Train• Launch date: 15 January 2009 on Taurus XL from VAFB• Operational life: 2 years• Principal Investigator: Dr. David Crisp, Deputy: Dr. Charles Miller• Project Manager: Thomas Livermore, Deputy: Dr. Ralph Basilio• Earth Science Flight Projects Office Manager: Dr. Steven Bard, JPL• ESSP Program Manager : Edward Grigsby, LaRC• Program Scientist: Dr. William Emanuel, NASA HQ• ESSP Program Executive: Eric Ianson, NASA HQ

The Orbiting Carbon Observatory (OCO)Watching The Earth Breathe…Mapping CO2 From Space

Science:• Collect the first space-based measurements of atmospheric CO2 with the precision, resolution, and coverage

needed to characterize its sources and sinks on regional scales and quantify their variability over the seasonal cycle.

• Use independent data validation approaches to ensure high accuracy (1-2 ppm, 0.3% - 0.5%)• Reliable climate predictions require an improved understanding of CO2 sinks

• What human and natural processes are controlling atmospheric CO2?

• What are the relative roles of the oceans and land ecosystems in absorbing CO2?

Page 3 3 of 25, AGU Fall Meeting Dec 17 2008

Mission System Description

Mission Ops (OSC) NASA GN (GSFC) and SN (TDRSS)

3-channel Spectrometer

(JPL/HS)

Data Processing

Center (JPL)

Ground Validation Sites

Taurus XL 3110 (KSC)

Please visit http://oco.jpl.nasa.gov for more information

Dedicated Spacecraft (OSC)

Page 4 4 of 25, AGU Fall Meeting Dec 17 2008

OCO Glint Mode

Glint Spot

Ground Track

R I

• Glint Observations: views “glint” spot

• Angle of reflection equals angle of incidence of sunlight at surface: R = I

• Improves SNR over oceans

• 70% time spent over oceans

Spacecraft Coordinates

Azimuth Orientation

Page 5 5 of 25, AGU Fall Meeting Dec 17 2008

OCO Target Mode

• Tracks a stationary surface target (calibration site) to collect large numbers of soundings

• Uplooking ground-based FTS data acquired simultaneously through same slant column

• Acquire Target data over 1 surface validation site each day

447-

m W

LE

F T

ow

er

Geolocation Accuracy

Scan Direction Spa

tial D

irect

ion

Alo

ng S

lit

Page 6 6 of 25, AGU Fall Meeting Dec 17 2008

Polarization Characteristics of OCO Spectrometers

• Transmits light with polarization parallel to slit

• I: intensity; Q, U: components of linear polarization; : angle between slit axis and principal plane (polarization angle)

• Nadir and glint modes:

• Target mode: measurement not restricted to principal plane

UQII 2sin2cos2

1//

o90

Page 7 7 of 25, AGU Fall Meeting Dec 17 2008

2OS Model Schematic

Scenario 1 Scenario 2

scatterer

Scenario 3

scatterer

Scenario 4

scatterer 1

scatterer 2

Natraj and Spurr, JQSRT, 107, 263–293, 2007

Page 8 8 of 25, AGU Fall Meeting Dec 17 2008

Glint Mode: Scenarios

• Solar Zenith Angle (SZA): 15°, 45°, 60°, 65°, 70°, 75°

• Aerosol Optical Thickness (AOT): 0 (Rayleigh), 0.01, 0.05, 0.1, 0.3

• Dusty maritime aerosol (Kahn et al., JGR, 2001)

• Background stratospheric aerosol

• Ocean surface reflectance simulated using Cox-Munk model

• Wind Speed: 4 m/s, 8 m/s, 12 m/s

Page 9 9 of 25, AGU Fall Meeting Dec 17 2008

Spectral Residuals (Glint): Scalar Model

Wind speed = 4 m/s

Residual = Model-Exact(VLIDORT)

Page 10 10 of 25, AGU Fall Meeting Dec 17 2008

Spectral Residuals (Glint): R-2OS Model

Wind speed = 4 m/s

Residuals from R-2OS model are smaller by 1–2 orders of magnitude

Page 11 11 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 Errors

Scalar Model R-2OS Model

XCO2 errors from R-2OS model < 1 ppm; scalar model errors as high as 5 ppm

AOT ↑

Wind speed = 4 m/s

Page 12 12 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 and Surface Pressure Errors

Retrieval error dominated by incorrect estimation of surface pressure; other effects become more important for large AOTs

Page 13 13 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 Errors

Scalar Model R-2OS Model

XCO2 errors larger when only O2 A band contributes to forward model error => CO2 and O2 errors cancel out in the ratio

Page 14 14 of 25, AGU Fall Meeting Dec 17 2008

Target Mode: Scenarios

• Location: Bremen (OCO validation site)

• Solar Zenith Angle (SZA): 50.4°

• Polarization angle: 122.68°, 177.385°, 118.961°

• Scatterer scenarios: 0.05 AOT, 0.05 AOT+0.25 Cirrus OT, 0.3 AOT

• Surface: Lambertian

Page 15 15 of 25, AGU Fall Meeting Dec 17 2008

Target XCO2 Errors

Scalar Model R-2OS Model

0.05 AOT

0.05 AOT+0.25 Cirrus OT

0.3 AOT

Page 16 16 of 25, AGU Fall Meeting Dec 17 2008

Target XCO2 Errors

Scalar Model R-2OS Model

XCO2 errors from R-2OS model < 1 ppm; scalar model errors as high as 30 ppm

Page 17 17 of 25, AGU Fall Meeting Dec 17 2008

Summary

• Ignoring polarization could lead to significant (as high as 40 ppm) errors (that are much larger than the measurement noise) in XCO2 retrievals

• 2OS approach to account for polarization works very well (in and out of principal plane), giving XCO2 errors that are typically smaller than 1 ppm, and smaller or comparable to measurement noise

• Errors dominated by errors in retrieved surface pressure

• R-2OS model two orders of magnitude faster than a full vector calculation

• Model needs to be tested for glint over land

Page 18 18 of 25, AGU Fall Meeting Dec 17 2008

Backup SlidesBackup Slides

Page 19 19 of 25, AGU Fall Meeting Dec 17 2008

RT Model

• Scalar multiple scattering model: Radiant (R)– Discrete ordinate solution for layer reflection and transmission matrices

– Adding method to obtain combined matrices for different layers

– Linearized: derivatives of intensity w.r.t. optical depth and single scattering albedo obtained analytically

• Polarization: 2OS– Polarization approximated by two orders of scattering

– Analytic integration over optical depth

– Fast invariant imbedding approach to add individual layers

– Linearized

Page 20 20 of 25, AGU Fall Meeting Dec 17 2008

Radiance Results: Clear Sky

SZA ↑

Glint reflectance ~ 10 times larger at 75° than at 15°

Wind speed = 4 m/s

Page 21 21 of 25, AGU Fall Meeting Dec 17 2008

Radiance Results: Cloudy Sky (AOT = 0.3)

Wind speed = 4 m/s Q decreases!

For large SZA, slant-path attenuation of solar beam very large; large fraction of light comes from atmospheric scattering

Page 22 22 of 25, AGU Fall Meeting Dec 17 2008

Linear Error Analysis

• Forward model errors systematic

• Bias in retrieved parameters x

• Bias can be expressed as follows:

• G: gain matrix– Describes mapping of measurement variations into retrieved vector

variations

• ΔF: forward model error

FGx

Page 23 23 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 Errors

Scalar Model R-2OS Model

XCO2 errors from R-2OS model < 1 ppm; scalar model errors as high as 5 ppm

AOT ↑

Wind speed = 4 m/s

Wind speed = 8 m/s

Wind speed = 12 m/s

Page 24 24 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 and Surface Pressure Errors

Retrieval error dominated by incorrect estimation of surface pressure; other effects become more important for large AOTs

Wind speed = 4 m/s

Wind speed = 8 m/s

Wind speed = 12 m/s

Page 25 25 of 25, AGU Fall Meeting Dec 17 2008

Glint XCO2 Errors

Scalar Model R-2OS Model

Wind speed = 4 m/s

Wind speed = 8 m/s

Wind speed = 12 m/s

XCO2 errors larger when only O2 A band contributes to forward model error => CO2 and O2 errors cancel out in the ratio