GeoCarb Mission Update

IWGGMS 16

Berrien Moore III

Sean Crowell

GeoCarb Mission: Overview

The GeoCarb Mission is designed to collect

observations of the column averaged concentrations of

carbon dioxide (CO2), methane (CH4), and carbon

monoxide (CO), and solar induced fluorescence (SIF)

from geostationary orbit (GEO) at a spatial resolution of

5-10 km over the Americas between 50º North and 50º

South Latitudes as a hosted payload on a commercial

communication satellite.

The Goal of the GeoCarb Mission is to provide

observations and demonstrate methods to realize a

transformational advance in our scientific

understanding of the global carbon cycle.

GeoCarb is an international effort!

Science Hypotheses

1. The ratio of CO2 fossil source to biotic sink for CONUS is

~4:1

2. Variation in productivity controls spatial patterns of

terrestrial sinks

3. Amazonian ecosystems are a large (~0.5-1.0 GtC/y) net

sink for CO2

4. Larger cities emit less CO2 emission per capita than

smaller ones

5. Amazonian ecosystems are a large (~50-100 MtC) net

source for CH4

6. The CONUS methane emissions are a factor of 1.6 ± 0.3

larger than in EDGAR and EPA databases

GeoCarb aims to fundamentally advance our understanding of

how the carbon cycle behaves on regional scales.

Baselin

e M

issio

nT

hre

shold

Mis

sio

n

Req ID

PLRA

Project Level 1 Requirements Baseline Threshold Expected

4.1.1.a /

4.1.2.a

Retrieve estimates of the column-averaged dry air

mole fractions XCO2, XCH4, XCO and SIF from

space-based measurements over cloud-free scenes

XCO2,

XCH4,

XCO, SIF

XCO2,

SIF

XCO2,

XCH4,

XCO, SIF

4.1.1.b.a /

4.1.2.b.a

The bias corrected, clear-air, multi-sounding GeoCarb

retrieval estimates for XCO2 will demonstrate multi-

sounding precision better than

<0.3% <0.6 % 0.2%

4.1.1.b.b /

4.1.2.b.b

The bias corrected, clear-air, multi-sounding GeoCarb

retrieval estimates for XCH4 will demonstrate multi-

sounding precision better than

<0.6% N/A 0.4%

4.1.1.b.c/

4.1.2.b.c

The bias corrected, clear-air, multi-sounding GeoCarb

retrieval estimates for XCO will demonstrate multi-

sounding precision better than

< 10% or

12 ppb

(whichever

is greater)

N/A 8%

4.1.1.c

4.1.2.c

Retrieval estimates of solar induced fluorescence

(SIF) with NESR (W/m2/µm/sr)

<0.75 <1.0 <0.5

4.1.4.a Geostationary orbit longitude 85º W ±20°

N/A 103º W

4.1.4.c Space-based measurements shall have spatial

resolution at the sub-satellite point (single sounding)

< 60 km2

<100km2

< 60 km2

Key Level 1 Requirements

At IWGGMS 15, We Discussed the

Instrument Simplification

GeoCarb: Baseline Instrument PDR

Door Assembly

Entrance Baffle

Star Trackers

Survival Heater Electronics Box (SHEB)

Electronics Tray- GeoCarb Electronics Box (GEB)- Focal Plane Electronics Box (FPE)- Autonomous Star Tracker Electronics (ASTE)- Decoder Electronics Box (DEB)- Cryocooler Electronics (CCE)

Main Bench

PZT Strain Gauge Pre-Amp

Cryocooler Compressor & Cold Head

Spectrograph

S/C Nadir Panel

South Radiators

Heat Pipes not shown

North Radiators

X

Y

E

W

N

S

Z (nadir)Inertial Measurement Unit

Simplified Instrument Overview

Entrance Baffle & Door Assy

Nadir Panel Mounted Components- GeoCarb Electronics Box (GEB)- Focal Plane Electronics Box (FPE)- Star Tracker Electronics Unit (STEU)- Decoder Electronics Box (DEB)- Cryocooler Electronics (CCE)

Cryocooler Thermomechanical Unit

Spectrograph

S/C Nadir Panel

SouthRadiators

NorthRadiators

+Z (nadir)

S (+Y)

E (+X)

W

NDoor Assembly

Entrance Baffle

Main Bench

Scan Box Assy

Spectrograph Assy+Z (nadir)

S (+Y)

E (+X)N

W

Telescope Assy

Optics PackageStruts, 6X

Star TrackerOptical Heads

Single slit, 4-Channel IR

Scanning Littrow Spectrometer Bands: 0.76𝝁m, 1.61𝝁m, 2.06𝝁m

and 2.32𝝁m

Simplified Instrument

• Optical redesign: reduction of SNR from smaller aperture

area - doubled stare time to meet SNR reqts– SNR has margin in all channels, despite the presence of increased

noise from thermal glow

– Areal coverage rate retained by removal of double sampling

• Removal of Secondary Solar Calibration Drum: redundancy

to assess changes in primary diffuser – already planned

lunar observations will be sufficient– Lunar observations will be used to assess changes in the calibration

over time. This methodology has been demonstrated by OCO-2.

Working with lunar cal experts to adapt the OCO-2 approach for

GeoCarb sampling

• Removal of IMU: expected pointing knowledge is reduced by

~0.1km, but it still meets MDRA requirements

We meet all Level 1 requirements and satisfy the mission hypotheses

with the Replanned Project, though with some reduced margin

Mexico City: OCO-2 and GeoCarb

A Day in the Life of GeoCarb

Host Spacecraft

• Working with SES to ensure

maximum compatibility of

GeoCarb with candidate host

spacecraft – Enveloping candidate spacecraft

environments

• Mechanical, thermal,

EMI/EMC, and contamination

– Accommodate 100V spacecraft

busses

– Simplifying instrument to

spacecraft interfaces

– Simplifying satellite integration

and test operations

– Streamlining the instrument

concept of operations

GeoCarb Instrument

Nadir Deck of Comm Satellite

E, +X

W

NS, +Y

Z (nadir)

System Data Flow Block Diagram

Ground ICDs are preliminary at PDR, baselined at CDR

Organization and External Interface View

Also, At IWGGMS 15, We Discussed

the Move to a Firm Fixed Price

Contract with Lockheed Martin for the

Instrument

This was to occur 17 July 2019 at the

It Did Not Happen

Bob Dylan: 2016 Nobel Prize in Literature

• "for having created new poetic

expressions within the great

American song tradition”

• The Times They Are a Changin’

• So too for GeoCarb

• Confirmed for Phase C on 27

December 2019 with Associated

Necessary Changes in Program

Management

NEW Team Organization – Phase C-D*

*GeoCarb Phase E-F team organization will emphasize OU management and is under development.

GeoCarb Project

Science Mission Directorate

PI OfficeProject Management Office

Code 422

Systems

Engineering

Safety & Mission

Assurance

Instrument Developmentt S/C

Host

Earth System Science Pathfinder

Program Office

Business

Management

Goddard Space Flight Center

Code 100

Flight Projects Directorate

Code 400

Earth Science Projects Division

Code 420

Earth Science Division

Flight Programs

Assoc. Dir. Flight

Program Executive

Science Research

Assoc. Dir. Research

Program Scientist

Public CommContracts

Ground Systems and Mission

Operations

NASA Administrator

Management Direction

Funding and Programmatic Direction Institutional Direction

Financial & Business Ops

Focal Plane

I&TDiscipline Engineering

Project Science

Science Team

Instrument MA

Instrument ScienceInstrument Ops

Center

Science Ops Center

S/C Ops and Communications

GeoCarb Data Ops Center

Distributed Active Archive Center

Contracts

Lockheed Martin Advanced

Technology

Center

University of Oklahoma

Langley Research Center

Colorado State University

Ames Research Center

Goddard Space Flight Center

NASA HQ

SES-GS

Science Algorithms

L2 Validation

L0/L1

L2 Development

Launch Services: In FLUX

• Launch services are provided by SES– Procured and managed by SES separately from the Spacecraft contract

• SES has recent experience with a variety of commercial launch vehicles:– SpaceX (Falcon 9)– Ariane 5 ECA– Soyuz– Proton Breeze M

– Launch services include:• All applicable licensing and permitting• TIMs and applicable reviews• Launch site integration and testing• Launch and early orbit support

Falcon 9Ariane V

ECA SoyuzProton

Breeze M

Here At IWGGMS 16, We Can Report

that We Believe that GeoCarb Is

FINALLY in a Good and Realistic

Position to Move Forward.

Near Term Challenge is the “Slit

Homogenizer” and the Longer-Term

Challenge is Schedule

PROGRESS

SLIT HOMOGENIZER

Initial Modeling Studies

• Initial work showed that a slit homogenizer was

necessary to meet baseline science requirements

Percentage of obs passing

quality filter

Scatter before filtering

Scatter after filtering

• The brightness variation across the slit “skews” the ISRF center of mass away

from the uniform illumination ISRF, which causes unresolved forward model

errors in the simulated radiances -> biased retrievals

Illumination Patterns

(Partially Obscured Slit)

ISRF Errors

(O2 A-band)

Without Slit Homogenizer With Slit Homogenizer• Errors grow with scene

albedo variation

• Number of good

observations decrease

with scene albedo

variation

• Both more quickly without

the slit homogenizer

• Atmospheric modeling

showed mission

objectives are attainable

w/o homogenizer

Slit

Entr

ance

Slit Exit

Slit homogenizer

performance model

was developed in

conjunction with and

validated by testing

of the CarbonSat

prototype

homogenizers

Testing of Prototype Devices at ITO

• Two vendors were selected to make three depths

(0.5mm, 1mm, 1.5mm) of homogenizer (each), yielding

six different options for an eventual flight model

• Each of the prototypes was shipped to ITO for

performance testing – knife edge moved across the slit

to provide high resolution imagery of response of SH to

partially obscured slit

• Main results:– Performance mostly independent of manufacturer

– Performance dependent on band and depth (as predicted)

– Significant polarization dependence in the response to

illumination

– Slit image width is a strong function of the polarization,

particularly in the 1.6um band

– Throughput also a strong function of the input illumination

polarization (and band)

20.01.2016Universität Stuttgart – Institut für Technische Optik 5

Test set-up - Alignment

Alignment laser

Fully Illuminated 50% Obscured

Perf

orm

ance R

esults f

or

Weak C

O2 B

and

H pol

V pol

Slit homogenizer model was updated to incorporate the

effects of the coatings in the two polarizations – the gross

effects were reproduced

H pol

V pol

Simulation Strategy

• We decided to employ the existing end-to-end simulator to assess the tradeoffs for the entire GeoCarb FoR

• Divide the slit into 12 sub-slits using– High resolution (500m) surface albedo– High resolution clouds from GOES– High resolution surface elevation data

• Create a simulated radiance for each sub-slit with an ILS that includes the effects of the slit homogenizer (or lack thereof)

• Perform retrievals on the scene total radiances with – ILS that accounts for polarization response of instrument

with/without the slit homogenizer– Polarization model that represents slit homogenizer (or none)– Scene average surface elevation– Imperfect meteorology/clouds/etc– Noise from instrument model (now with polarization dependence

and throughput dependence from SH)– Polarized surface

Overall Results

L1 Reqt No Scene

Inhomogeneity

With Slit

Homogenizer

Without Slit

Homogenizer

Total land soundings 32681 32681 32681

Total # Converged 13931 13574 11724

% Good Soundings

(rel to total)*

23% 21% 15%

% Soundings Lost

due to SI

N/A 2% 32%

CO2 Error Std

(ppm)

1.3 0.9 1.1 1.3

CH4 Error Std (ppb) 12 8 8.3 7.8

CO Error Std (ppb) 10 2.9 3.3 3

• Though we “meet” the uncertainty requirement without the homogenizer, there

would be no margin for other error sources

• With a roughly 25% “pass” rate through the quality filter, and about 4M soundings

per day, the homogenizer would yield about 60K more soundings per day.

ADDITIONAL PROGRESS

“Day in the Life” Test of Algorithms

• Successfully ran a day’s worth of GeoCarb

simulated radiances through L2, resulting in ~318K

good quality retrievals over the western

hemisphere

• Average processing time: ~8 minutes per sounding

– RT speedups are in progress

Solar-induced chlorophyll fluorescence (SIF)

• New retrieval algorithm, GASBAG,

written from scratch – based on

established OCO/GOSAT concepts– [Frankenberg et al. 2011, Köhler et al. 2015]

• Can utilize both physics- and data

driven-based approaches for the

retrieval– Produce equivalent results for clear-sky

scenes

• Algorithm successfully

employed for early OCO-3

assessment of SIF (and

other single-band retrievals)

(JPL press release, 2019/7/12)

OCO-3 SIF over Azerbaijan

Solar-induced chlorophyll fluorescence (SIF)

• Physical vs. data-driven retrievals

– Physics-based retrieval interprets cloudy

scenes as having (unphysical) negative

SIF

– Data-driven retrieval ”sees” clouds as

having zero SIF

– Both are wrong, but consistent with

the assumptions of the algorithms

• Cloud screening is part of L2 SIF

data production, thus relies on tested

and tried cloud screening algorithms

• SIF algorithms are ready!

A Great Job Opportunity

We seek a teamwork-oriented individual who will serve as OU’s local expert on the

GeoCarb instrument. Duties will include:

• Supporting the development and testing efforts of the Lockheed Martin instrument

team;

• Supporting pre-flight calibration activities;

• Supporting the verification of instrument performance and knowledge requirements;

• Supporting the project science by helping to ensure that the instrument

performance meets the high-level science requirements;

• Developing instrument operations plans at OU for implementation after launch; and

• Evaluating and monitoring instrument performance and leading anomaly

investigations during mission operations.

The position will involve regular travel to work with the instrument team on site in Palo

Alto, California, USA.

The successful candidate will possess a minimum of a Master’s degree in earth

science, physics, engineering, astronomy, or a related field with experience in

characterizing optical systems as well as a deep understanding of remote sensing.

Overall Summary

• GeoCarb will make daily observations of XCO2, XCH4,

and XCO over the western hemisphere in order to

constrain local to continental surface fluxes

• After numerous challenges, GeoCarb is on a good path

programmatically and technically

• Hosted payloads are tough!

• Science algorithms and the data processing pipeline are

maturing

• Our concept of operations is baselined using odd/even

day strategy

• We are looking for a great instrument scientist to support

the development and operations effort at the University

of Oklahoma!

See You at IWGGMS 17 and at Launch!!!!!

VIELEN DANK

THANK YOU

34–31CDR 34 – L2 Validation– January 29-31, 2020

Backup

LM Instrument Org Chart – May 2020

5.3 Mechanical Lead

Brad DeLima 5.8 Integration and Test

Phil Tung

5.3 Instrument Lead

Cathy Chou

5.5 Software Lead

Noelle Nguyen-Phuc

5.1 Finance and Business Ops

Barbara Richardson– Contracts

Carol Brosious – Finance

Svetlana Teplitsky – Planner

Emilie Schott - Planner

Terry Wicks – SC manager

Travis Burdette– Sub

John Gamble – Sub

David Schiff – IT Systems

Heidi Shaylor – CM

Instrument Manager

5.1 D L Hoffmann

Claude Kam - CAM

5.4 Thermal Lead

Rose Navarro

-Dave Akin

-Gwen Gradwohl

-Craig Hom

-Buck Holmes

-Deborah Hudson

-Jessica Sun

-Roel Vanbezooijen

-Paul Zhu – M&P

-Eric Dixon

-Joe Mobilia

-Diana Pang

-Alex Petrovsky

-Riza Bravo

-Ross Yamamoto

-Rodrigo Zeledon

-Craig Farless

-Willie Huang

-Bruce Imai

-Ray Yu

-Lomita Rubio

-Mengie DelaCruz

5.1 Instrument SE

Brett Allard

-Alyssa Ruiz-Brandon Chan

-Roger Chevalier

-Donovan Kelly

-David Mansir

-Phil Tung

-Constantine Orogo

-Dennis Wong

5.6 Electrical Lead

Chris Edwards5.6 Focal Plane

Tony Magoncelli

5.2 Instrument MA

Armando Cortez

Swathy Karuppaian – SQA

Donald Cooper – Safety

Commercial Civil Space

Director

Adrian Cuadra

OU Program

Berrien Moore

Shelly Finley

5.1 Instrument Scientist

Eric Burgh

-Margaret Shaw-Lecerf

-Paul Schweiger

-Torben Andersen

-Test Leads TBD

NASA GSFC

Todd King

Adam Matuszeski

5.1 SW Systems

Oversight

Noelle Nguyen-Phuc

Consulting Support

Gary Kushner

What are our objectives and criteria?

• Main goal: go/no go on retaining the slit homogenizer in

the baseline

• Considerations– Level 1 Requirements

• Baseline: Multisounding Precision of 1.3ppm (CO2), 12ppb

(CH4), 10ppb or 10% (CO)

• Threshold: Multisounding Precision of 2.5ppm (CO2) (no

requirements on CH4, CO)

– What about sampling?

• Goal was to get a good sample in each 2x2 degree lat/long

box over a relevant time period – hopefully every day, but at

least once per week after cloud screening

– We have to weigh simulated performance against

implementation risk: alignment tolerances, scattered light, other

unknown unknowns

Synthetic

Retrievals

Atmospheric

inversion model

Inferred fluxes

“End to End” Simulator

Level 2 Req’ts

Instrument Performance Model

Level 1 Req’tsMission Objectives

Simulated

Spectra

Level 3/4 Req’ts

Radiance:

� = � ∗ � �⁄

Solar

Zenith Angle(SZA)

Satellite

Zenith Angle(ZA)

�

� = � cos � � � ∗ �

� = 1

cos � � �+

1

cos � �� = aerosol optical depth

ℎ�

�

W/m2/sr/µm

Photons/sec/m2/sr/µm

AW etendue

Photons/sec/µm/pixel

Instrument Line Shape

Photons/sec/pixel

Throughput à

component efficiencies

Individual exposure time

12 co-added exposures (Nf)

à Gain in SNR

Photons/pixel

Warm optics à

thermal emission

� � � =�

� + � ∗ �

N0 à read noise

and thermal background

N1 à instrument response

12

Simulation Strategy (cont)

• Retrieval results are then– Quality filtered using the same techniques that are applied for OCO-2/3

operational products– Bias corrected using the same techniques that are applied for OCO-2/3

operational products (assume known truth for all available soundings -optimistic)

• The quality filtered, bias corrected products are the best representation of the data that 90% of the science community will use (and hence the right data set to look at for this purpose)

• Perform this for 1) no scene inhomogeneity, and scene inhomogeneity included and 2) including a slit homogenizer or 3) not including a slit homogenizer

• Metrics– Number of good quality soundings– Error scatter of good quality soundings– Observational coverage – Regional scale biases – these have been shown to be the main driver of

accuracy of flux estimates

• Not included: instrument calibration uncertainties, spectroscopy errors, systematic errors from scattered light and other instrument sources beyond the scene inhomogeneity effects

Summary: Slit Homogenizer

• Global statistics tell us the throughput and errors are better

with the slit homogenizer

• These differences occur in the regions that are tied to key

mission hypotheses – Amazonia

• Excluding concerns about schedule, implementation,

calibration, and characterization, we recommend keeping the

slit homogenizer in the design

• Further potential research items– Would including an ILS fit parameter in the L2 retrieval alleviate

some of the trace gas retrieval errors in the absence of a slit

homogenizer?

– How does the instrumental polarization knowledge requirement

play into our understanding of these errors?

– How do these errors average down spatially?

• Cloud-free limited area experiments near Lamont and Manaus

indicate significantly better results with spatial averaging.

Solar-induced chlorophyll fluorescence (SIF)

• Retrieval validation (physical only)

– For clear-sky scenes on land (no

aerosols, no clouds), the true bottom-

of-atmosphere (BOA) SIF is fully

recovered by the algorithm

– Retrieval errors in simulations due to

clouds and aerosols are well-

understood

– After cloud-screening, SIF

errors are small up to

moderate aerosol loadings

(AOD ⪅ 0.2), consistent

with literature

Instrument Simulator Development

OCO-3 Observations per 5x5deg Area - 201406

OCO-2 Observations per 5x5deg Area - 201406 GeoCarb Observations per 5x5deg Area - 201406

Advances in instrument simulators allow for more direct comparisons of OCO and GeoCarb data. Here, we show observation counts aggregated over a single month for OCO-2, OCO-3 and GeoCarb. Note the difference in color scale between the GeoCarb and OCO plots. As a result of its ability to scan more than once per day, GeoCarb obtains more than 60x more observations than OCO over the Americas