EarthFinderReport_compressedforarxiv

EarthFinder Mission Concept

2019 Probe Study Report

i

This research was carried out at

George Mason University and

the Jet Propulsion

Laboratory, California Institute of Technology, under a contract with the National

Aeronautics and Space Administration.

Reference herein to any specific commercial product, process, or service by trade

name, trademark, manufacturer, or otherwise, does not constitute or imply its

endorsement by the United States Government or the Jet Pro

pulsion Laboratory,

California Institute of Technology.

EarthFinder Mission Concept

2019 Probe Study Report

ii

Team Members

Peter Plavchan (PI, George Mason University)

Gautam Vasisht (Co

-

I, Instrument team lead,

NASA Jet Propulsion Lab)

Chas Beichman (Instrument team lead, NASA

Exoplanet Science Institute)

Heather Cegla (Stellar Activity team lead, U.

Geneva)

Xavier Dumusque (Stellar Activity team lead,

U Geneva)

Sharon Wang

(Telluric team lead, Carnegie

DTM)

Peter Gao (Ancillary Science team lead, UC

Berkeley)

Courtney Dressing (Ancillary Science Team

Lead, UC Berkeley)

Fabienne Bastien (Penn State University)

Sarbani Basu (Yale)

Thomas Beatty (Penn State)

Andrew Bechter (Not

re Dame)

Eric Bechter (Notre Dame)

Cullen Blake (U Penn)

Vincent Bourrier (University of Geneva)

Bryson Cale (George Mason University)

David Ciardi (NASA Exoplanet Science

Institute)

Jonathan Crass (Notre Dame)

Justin Crepp (Notre Dame)

Katherine de Kleer

(Caltech/MIT)

Scott Diddams (NIST)

Jason Eastman (Harvard)

Debra Fischer (Yale)

Jonathan Gagné (U Montreal)

Scott Gaudi (Ohio State)

Catherine Grier (Penn State)

Richard Hall (University of Cambridge)

Sam Halverson (MIT)

Bahaa Hamze (George Mason Universit

y)

Enrique Herrero Casas (CSIC

-

IEEC)

Andrew Howard (Caltech)

Eliza Kempton (U Maryland)

Natasha Latouf (George Mason University)

Stephanie Leifer (NASA Jet Propulsion Lab)

Paul Lightsey (Ball Aerospace)

Casey Lisse (JHU/APL)

Emily Martin (UCLA)

William Matzko (George Mason University)

Dimitri Mawet (Caltech)

Andrew Mayo (Technical University of

Denmark)

Simon Murphy (University of Sydney)

Patrick Newman (George Mason University)

Scott Papp (NIST)

Benjamin Pope (New York University)

Bill Purcell (

Ball Aerospace)

Sam Quinn (Harvard University)

Ignasi Ribas (CSIC

-

IEEC)

Albert Rosich (CSIC

-

IEEE)

Sophia Sanchez

-

Maes (Yale University)

Angelle Tanner (Mississippi State)

Samantha Thompson (Univ. of Cambridge)

Kerry Vahala (Caltech)

Ji Wang (

Ohio State

)

Peter Williams (Northern Virginia Community

College)

Alex Wise (University of Delaware)

Jason Wright (Penn State

)

EarthFinder Mission Concept

2019 Probe Study Report

iii

Table of Contents

1

SCIENCE

................................

..........................

1

-

1

1.1

Executive Summary

................................

..........................

1

-

1

1.1.1

Study Findings

................................

..................

1

-

1

1.1.2

Study Recommendations

................................

1

-

2

1.2

State of the Field

................................

1

-

4

1.2.1

Scope of

Study

................................

..................

1

-

4

1.2.2

EarthFinder Overview

................................

....

1

-

5

1.2.3

Mission Yield

................................

....................

1

-

6

1.3

EarthFinder El

iminates Telluric Contamination

................................

..........

1

-

9

1.3.1

Methodology

................................

...................

1

-

10

1.3.2

Errors Induced by Tellurics vs. Wavelength

................................

.............................

1

-

12

1.3.3

Tellurics’

Contribution to the RV Error Budget

................................

.......................

1

-

12

1.3.4

Additional Limitations Set by Tellurics

................................

......

1

-

13

1.4

EarthFinder Can Uniquely Mitigate Stellar Jitter

................................

........

1

-

14

1.4.1

Stellar signals

................................

...................

1

-

14

1.4.2

The Advantage of High Cadence in Planet Discovery and Mitigation of Stellar

Activity

................................

.............................

1

-

16

1.4.3

The Radial Velocity Color Advantage of EarthFinder

................................

.............

1

-

18

1.4.4

High SNR, high resolution & line by line analysis

................................

......................

1

-

1

1.4.5

Continuum determ

ination

................................

..............................

1

-

3

1.5

General Astrophysics with EarthFinder?

................................

......................

1

-

5

1.5.1

Instrument Capabilities

................................

...

1

-

5

1.5.2

Stellar Astrophysics

................................

..........

1

-

6

1.5.3

Exoplanet and Disk Science Applications

................................

...

1

-

7

1.5.4

Extragalactic Science

................................

.......

1

-

8

2

ENGINE

ERING, INSTRUMENT, & MISSION DESCRIPTION

................................

....................

2

-

1

2.1

Overview

................................

.............

2

-

1

2.1.1

The Top

-

Level Requirements

................................

........................

2

-

1

2.2

Spacecraft

................................

............

2

-

3

2.2.1

Science Observing Profile

................................

2

-

4

2.2.2

Doppler Spectrograph Requirements

................................

...........

2

-

5

2.2.3

PAYLOAD Details

................................

.........

2

-

6

2.3

Wavelength Calibration

................................

..................

2

-

10

2.3.1

Spec

trograph Calibration Options

................................

..............

2

-

10

2.3.2

Optical Frequency Combs as Spectral Rulers

-

Astrocombs

................................

..

2

-

14

2.3.3

Options for Flight Astrocombs

................................

...................

2

-

15

2.3.4

Flight Comb Requirements and Architecture

................................

...........................

2

-

1

7

2.3.5

Path Forward

–

Technology Roadmap

................................

......

2

-

18

3

SPECTROGRAPH THERMAL EVALUATION

................................

....................

3

-

1

4

COST, RISK, HERITAGE ASSESSMENT

................................

...............................

4

-

1

4.1

Cost Assessment

................................

4

-

1

5

References

................................

..........................

5

-

1

A.

Acronyms

................................

...........

A

-

1

EarthFinder Mission Concept

2019 Probe Study Report

iv

Table of Tables

Table 1

-

1

: The PRV method remains the best technique for measuring planet masses, and the second best

for planet discove

ry after the transit method (only a small fraction of planets, 1

-

10%, transit their host

star).

................................

............................

1

-

4

Table 1

-

2

: HabEx Target

Catalog used for EarthFinder survey simulations

................................

.............

1

-

6

Table 1

-

3

: Summary of RV rms precision lower limits for different instrument using d

ifferent RV extraction

methods. EFE stands for EarthFinder Equivalent on the ground.

................................

.

1

-

13

Table 1

-

4

: Known sources of stellar signal, th

at EarthFinder can model and mitigate, with their typical

timescales and amplitudes for main sequence stars.

................................

.................

1

-

14

Table 1

-

5

: List of the

planetary parameters used in the System

-

2 model.

................................

...............

1

-

17

Table 2

-

1:

The Science Traceability Matrix is used to derive the EarthFinder Mission and Instrument

functional requirements.

................................

..............................

2

-

2

Table 2

-

2

: EarthFinder NIRS PRV Error Budget.

................................

...................

2

-

12

Table 2

-

3

: EarthFinder OPS PRV Error Budget

................................

......................

2

-

13

Table 2

-

4:

Power consumption of a compact fiber laser comb package, housing

consisting of one 18 cm x

20 cm x 2.5 cm box (Sinclair et al, 2015). Many of the elements in this system would also be used in a

soliton microcomb

-

based flight comb system.

................................

.........................

2

-

17

Table 2

-

5:

Advanced microcomb calibration system power estimates and components for 4 comb

architecture concepts.

................................

...............................

2

-

18

Table 2

-

6

: Payload Mass and Power

................................

........

2

-

19

Table 4

-

1

: Cost Estimate

................................

............................

4

-

1

Table of Figures

Figure 1

-

1

: PRV

-

discovered exoplanets less than 10 M

Earth

as a function of stellar mass and planet mass

modulo the unknown inclination. Black circles are data from the NASA Exoplanet Archive. The blue

-

green orb corresponds to the Earth. The blue curve corresponds to the approximate current detection limit

of the

PRV method, the green curve corresponds to the NEID spectrometer (or similarly, EXPRES, or

ESPRESSO), and the black curve corresponds to EarthFinder and its unprecedented 1 cms

-

1

sensitivity.

.

1

-

6

Figure 1

-

2

: HabEx targets are on average more active than the typical quiescent PRV survey target. We

compare a histogram of S

-

index activity indicators for stars in the HabEx catalog (red) comp

ared with the

S

-

index distribution of the California Planet Search survey sample (blue; from Wright et al. 2004).

......

1

-

8

Figure 1

-

3:

A dedi

cated EarthFinder mission survey of 61 direct imaging targets with a 1.45m aperture

(blue)

outperforms

a 25%

-

time survey on a 8

-

m ground

-

based facility of 53 targets (red) in the number

of observations per star, with a median of 349 vs. 124 epochs respecti

vely.

................................

.............

1

-

8

Figure 1

-

4: The cadence advantage of EarthFinder enables it to detect Earth

-

mass HZ analogs with

three times better accuracy on average for the recovery of the velocity semi

-

amplitude relative to an

equivalent ground

-

based survey. Once factoring in the telluric and

stellar activity correction error

terms present from the ground, the space advantage grows.

Top: Log of the absolute relative error in

the recovered vs. injected velocity semi

-

amplitude as a function of the semi

-

amplitude for the simulated

planets in the

EarthFinder survey. A linear fit is shown as a solid line. Bottom: The same for the ground

-

based survey. The dashed lines are the solid lines from the other panel, overlaid for a direct comparison to

highlight the space advantage in performance.

................................

..........................

1

-

9

EarthFinder Mission Concept

2019 Probe Study Report

v

Figure 1

-

5

:

EarthFinder detects most (>90%) of our simulated exoplanets, including most of the 23

Earth

-

mass HZ analogs with semi

-

major axes 0

.95<a<1.67 and masses 0.5<m<4 m

Earth

(following

Kopparapu et al. 2013). We do not test the recovery of all planets with orbital periods >5 years and

K(m/s)/P(day) < 5.476*10^

-

3 (m/s/day) shown as the blue line for computational efficiency. Detecting

more E

arth

-

mass planets requires an intrinsically higher value of

η

Earth

or surveying more stars with a

larger aperture or longer primary mission. By comparison, GAIA will only be able to detect the Jovian

planets greater than 1 Jupiter mass in the upper right c

orner of this plot (K >10 m/s, approximately).

....

1

-

9

Figure 1

-

6

: Convolved telluric lines “pull” the centroid of stellar absorption lines

off from their true

Doppler

-

shifted stellar absorption lines, as shown in this illustration (Wright et al. 2013).

....................

1

-

10

Figure 1

-

7

:

Comparison of the line profiles for a CO

2

line near 1602 nm for two observatories: Kitt Peak

(used in simulated spectra) and Mauna Kea (used as input to fit the simulated spectra). Both spectra are at

airmass = 1. The Mauna Kea Scaled spectrum is the best

-

fit version when fitting the Kitt Peak profile by

scaling the Mauna Kea line by a power law. The RMS of the residual of this fit is 1.2%, which is around

the typical value for all lines and smaller than the typical values reported by Ulmer

-

Moll et al. (201

9).

...

1

-

11

Figure 1

-

8

: RV errors added by tellurics as a function of wavelength for three different methods used in

this work. Each point

plotted is the average RV error for 7 neighboring chunks centered at each

wavelength. The RV error of each chunk is the RMS of RVs of this chunk over the simulated time span of

365 days. The spectrum plotted in red at the bottom is an illustration of tell

uric absorption.

.................

1

-

12

Figure 1

-

9

:

RV signals vs. time, as measured by the visible arm (upper panel) and NIR arm (lower

panel) of an Ear

thFinder equivalent (EFE) spectrograph from the ground, extracted with two different

telluric mitigation methods, assuming SNR=100 per pixel for R=120,000. The RV signal of an Earth

analog is plotted as the green line (semi

-

amplitude = 9 cms

-

1

).

................................

...............................

1

-

13

Figure 1

-

10:

Telluric absorption spectrum (blue) plotted as a function of wavelength, and compared with

G, K and M dwarf spectra in yellow,

orange and red respectively, showing the overlap of spectral regions

rich in stellar RV information content blocked by deep telluric absorption. The horizontal arrows along

the top and the shaded regions in the plot indicate the wavelength coverage of some

representative

operational spectrographs.

................................

........................

1

-

14

Figure 1

-

11

: The wide field of regard of EarthFinder is superior to

ground

-

based facilities which are

additionally limited by weather.

................................

...............

1

-

16

Figure 1

-

12

: The violin plots of the posterior distributions for

the parameters of the two Earth

-

like planets

in System

-

2 from Hall et al. (2018). Results for six different observation schedules are plotted here. The

distributions colored in blue are the results for data including only Gaussian noise; the red results a

re

when stellar “noise” is also added to the dataset, however corrected for to a level of 75%. The horizontal

line in each plot shows the correct parameter value. A narrow, single peaked, Gaussian

-

like distribution

around the correct parameter is the desir

ed result from the analysis; the space

-

based sampling is optimum

for obtaining well constrained solutions around the correct parameters.

................................

.................

1

-

18

Figure 1

-

13

:

Top

: A portion of our simulated stellar activity 5 yr time

-

series based upon the active Sun

for the visible (blue) and NIR (red) arms of EarthFinder, with peak

-

to

-

peak changes of ~20 m/s.

Bottom

:

The visible minus NIR RV color, showing t

hat the amplitude of the RV color obtained with EarthFinder

is large compared to our measurement precision

. To first order, the RV color is proportional to the RV

stellar activity in the visible band and completely free of any planetary signals.

Thus, we c

an exploit

this fact to correct the stellar activity in the visible arm. The correspondence is not perfect, due to

wavelength

-

dependent differences in limb

-

darkening and convective blueshift. Although not accounted

for here, such effects can be modeled.

................................

......

1

-

19

EarthFinder Mission Concept

2019 Probe Study Report

vi

Figure 1

-

14

: (

Left

) Our simulated visible EarthFinder RVs for HIP 61317 with perfectly corrected stellar

activity. From top to botto

m are the full RVs, followed by the best

-

fitting RV time

-

series phased for each

planet isolated from the other four from RADVEL. With stellar activity perfectly corrected, EarthFinder

recovers the detection and orbits of all but the Mercury

-

like planet on

the bottom.

(Center

) Our simulated

EarthFinder RVs for HIP 61317 with stellar activity corrected with a simple linear model from the RV

color (62% reduction in rms) and a GP model for the residuals. (

Right

) Our simulated ground

-

based RVs

with stellar activity modeled with a GP, but with no initial activity correction.

From space, the orbital

information of the HZ super

-

Earth is recovered, but from the ground it is not and the eccentricity

blows up.

In both

cases, the Venus analog is recovered, but the eccentricity & phase are not (although

EarthFinder better recovers the orbital phase). This is not insurmountable however, as our activity

correction was the simplest available to EarthFinder, and we have not y

et developed a more sophisticated

analysis.

................................

....................

1

-

20

Figure 1

-

15

: Bootstrap periodograms of the EarthFinder and ground

-

based RV time

-

series

simulated for

HIP 61317, with the massive Jovian

-

analog removed. The Mercury, Venus, and HZ super

-

Earth analog

orbital periods are indicated with the vertical dashed lines.

................................

.......

1

-

2

Figure 1

-

16

: Line bisector of the average model line profile from Cegla et al. (2019) before (iblack) and

after it was convolved with an in instrumental profile corresponding to three modes: Medium Resolution

(R

= 70,000), High Res (R = 140,000), and Ultra

-

High Res (R = 190,000).

................................

.............

1

-

2

Figure 1

-

17

: Kendall’s Tau correlation coefficient between lin

e core flux and S

-

index (Vaughn et al.

1978), versus energy level of the electron configuration required before each absorption can occur. Each

point corresponds to a given spectral line color coded according to its wavelength. This previously

undiscovered

correlation will be exploitable by EarthFinder at higher SNR for mitigating stellar activity.

................................

....

1

-

3

Figure 1

-

18

: RV of two spectral line

s in the 2010 RV data set of Alpha Centauri B. Each plot is divided

in two subplots, on the left, the RV of the line as a function of time, on the right the correlation between

the RV of the line and the RV of the star, which for those data is a good proxy

for stellar activity. The RV

of the line presented on the left is not correlated with stellar activity, while the one on the right is strongly

correlated (Dumusque 2018).

................................

.....................

1

-

3

Figure 1

-

19

: RV data of Alpha Cen B strongly affected by stellar activity. We show here the RV

measured using all the spectral line (top) only the very affected ones (middle) and the less affected ones

(bottom). By optimally selecting the lines to measure RV, the bottom selection shows it is possible to

mitigate stellar activity by 38% (from Dumusque 2018).

................................

..........

1

-

3

Figure 1

-

20

: A brightness proxy versus RV, where the proxy is determined by the integrated area

underneath the disc

-

integrated model line profiles. This area has been normalized by its maximum value,

mean

-

subtracted, and convert

ed to parts per million (Cegla et al. 2019).

................................

..................

1

-

5

Figure 1

-

21

: We plot the integration times required for R=150,000 to reach the labeled

SNR at the

indicated wavelengths as a function of apparent magnitude for a Sun

-

like star (left) and M dwarf (right).

Fainter targets can be reached in tractable integration times at degraded spectral resolution from spectral

pixel binning, which is more c

ost effective than hardware for lower resolution modes without a significant

detector noise penalty.

................................

1

-

5

Figure 1

-

22

: Sensitivity kerne

ls for (a) sound

-

speed and (b) density for a few radial modes. Note that the

lower frequency model (n=4) has the largest sensitivity at the core. Such low

-

frequency modes are lost in

granulation noise in brightness intensity measurements.

................................

...........

1

-

7

Figure 2

-

1:

Thermal management and control of the spacecraft derives much of its heritage from features

implemented previously in NASA's

Spitzer mission.

................................

................

2

-

4

EarthFinder Mission Concept

2019 Probe Study Report

vii

Figure 2

-

2

: Schematic design of the EarthFinder instrument suite shows the telescope primary (inverted

and at the

bottom) and the three spectrometers UVS, OPS, NIRS along with the fine guidance camera.

.

2

-

6

Figure 2

-

3:

Payload block diagram showing th

e feeds three spectrographs and the fast guiding camera.

.

2

-

7

Figure 2

-

4:

Schematic layout of a NIRS

-

like spectrometer based on an instrument concept for the Keck

HISPEC instrument. Light enters via a bundle of single mode fibers (short and long wavelengths, sky, optical

fiber combs) and is collimated into a ~25 mm diameter beam

and projected onto an R4 echelle grating (~10

lines/mm) using a Three Mirror Anastigmat (TMA) . The echelle is slighted tilted to send the outgoing beam

to a beamsplitter which sends the light to long and short arms with crossed

-

dispersing gratings (or pr

isms)

and cameras illuminating Teledyne H4RG/H2RG infrared detectors. The overall scale of the optics footprint

is 0.4 m

Ð

0.7 m. Credit: J. Fucik, Caltech Optical Observatories.

................................

..............................

2

-

9

Figure 2

-

5

: The Keck HISPEC Echellogram layout using an H2RG and an H4RG to cover 960

-

2500 nm

range. EarthFinder will use two H4RGs. Credit: J. Fucik, Caltech Optical Observatories.

.....................

2

-

10

Figure 2

-

6

: A generic error budget for a high

-

resolution RV spectrometer has many terms which the

EarthFinder eliminates or mitiga

tes through operation in space (red), use of a diffraction limited

spectrometer (orange), and a single

-

mode fiber (yellow), or via calibration with a laser frequency comb

(green). A representative EarthFinder error budget is shown in

Table 2

-

2

and

Table 2

-

3

.

....................

2

-

10

Figure 2

-

7:

System throughputs of the OPS and NIRS spectrographs compared with the throughput of the

new, next generation NN

-

EXP

LORE NEID spectrograph. The high optical throughput of space

-

based

systems allows for a robust trade with the telescope aperture

-

diameter.

................................

.................

2

-

11

Figure 2

-

8

: The EarthFinder photon limited RV precision per spectral order for V=5 and 11 targets

(top/bottom panels) with three different effective temperatures, in an hour of integration. The colored points

show the achievable precision within

in

dividual OPS and NIRS diffraction orders. With EarthFinder's

effective area, high measurement precision (<50 cm/s) can be achieved within individual echelle orders on

bright targets. Most RV information content for FGK stars is contained in OPS data, while

the NIRS channels

provide ~ 10 cm/s precision, but crucially chromatic diagnostics for evaluating stellar noise. NIRS provides

as much or more information content as OPS when observing the cooler stars.

................................

..........

2

-

14

Figure 2

-

9:

Silica microastrocomb package demonstrated at the Keck Observatory (Suh et al, 2018).

.

2

-

16

Figure 2

-

10:

Image of soliton comb projected onto the NIRSPEC Echelle spectrometer at the Keck

Observatory in orders 44 to 51 with the corresponding wavelength ranges of each order indicated. Th

e white

dashed box indicates soliton emission and has been heavily filtered to prevent potential damage to the

spectrograph detector (Suh et al, 2018). ADU: Analog

-

to

-

Digital Units

................................

......................

2

-

16

Figure 2

-

11:

The Chip Scale Atomic Clock (CSAC): A RF Frequency Reference for the comb repetition

rate stabilization will be needed where no GPS signal is available.

................................

.........................

2

-

17

EarthFinder Mission Concept

2019 Probe Study Report

1

-

1

SCIENCE

1.1

EXECUTIVE SUMMARY

EarthFinder is a NASA Astrophysics Probe

mission concept selected for study as input

to the 2020 Astrophysics National

Academies Decadal Survey.

The EarthFinder concept is based on a

dramatic shift in our understanding of how

PRV measurements should be ma

de. We

propose a new paradigm which brings the high

precision, high cadence domain of transit

photometry as demonstrated by Kepler and

TESS to the challenges of PRV measurements

at the cm/s level. This new paradigm takes

advantage of: 1) broad wavelength c

overage

from the UV to NIR which is only possible

from space to minimize the effects of stellar

activity

; 2) extremely compact, highly stable,

highly efficient spectrometers (R>150,000)

which require the diffraction

-

limited imaging

possible only from space

over a broad

wavelength range; 3) the revolution in laser

-

based wavelength standards to ensure cm/s

precision over many years; 4) a high cadence

observing program which minimizes sampling

-

induced period aliases;

5)

exploiting the absolute

flux stability f

rom space for continuum

normalization for unprecedented line

-

by

-

line

analysis

not possible from the ground

;

and

6

)

focusing on the bright stars which will be the

targets of future imaging missions so that

EarthFinder can use a

~

1.5 m telescope

.

In this sum

mary we

summarize the key

findings

and recommendations

of the report

with more detail presented in subsequent

sections

.

1.1.1

STUDY FINDINGS

“Measurements from space might be a final

option if the telluric contamination problem

cannot be solved.”

-

National Academies

Exoplanet Science Strategy report, 2018

1.

The Earth’s atmosphere will limit precise

radial velocity (PRV) measurements to

~10 cm/s

at wavelengths longer than

~

700 nm and greater than 30 cm/s

at

>

900 nm (

see

Section

1.3

), making it

challenging to mitigate the effects of

stellar activity without a measurement of

the color dependence due to stellar

activity in the PRV time series.

EarthFind

er can greatly reduce the effects

of stellar jitter through its great spectral

grasp, from the UV to the near

-

IR.

2.

Simultaneous visible minus

near

-

infrared (

NIR

)

PRV measurements

(“PRV color”)

perfectly

subtracts off

the planet signal(s), uniquely isolating

the chromatic stellar activity signal from

the planet signal(s) in the PRV time

-

series (Section

1.4.3

). EarthFinder’s

broad spectral grasp offers the highest

SNR measu

rement of this chromatic

activity because the lack of the Earth’s

atmosphere permits PRV measurements

at sufficient precision

at wavelengths

greater than

~700 nm. This unique

space advantage will permit

disentangling exoplanet and stellar

activity signals.

3.

“Line

-

by

-

line” analysis with high SNR

and

high

-

resolution

data (R>100,000)

can mitigate stellar jitter. In

a

few cases

from the ground

,

this technique has

resulted

in

a reduction in stellar activity

PRV

RMS

of 33

-

50%

(Dumusque 2018,

Lanza et al.

2018, Wise et al. 201

8

)

but

g

reater mitigation (>75%) is needed to

detect Earth

-

mass analogs (Hall et al.

2018). Cegla et al. (2019) demonstrate

that with better continuum

normalization

enabled by a space

platform

, the ability to distinguish

between PRVs a

nd stellar activity from

convection and granulation strengthens

dramatically

(Section

1.4.5

)

.

EarthFinder Mission Concept

2019 Probe Study Report

1

-

2

4.

The UV channel of our space platform

permits the simultan

eous observations

in the near

-

UV of the Magnesium II

lines at 280 nm in addition to the

Calcium II H&K absorption lines, the

latter of which routinely observed from

the ground for PRV activity correlation

analysis. These Mg II and Ca II activity

sensitive

spectroscopic features are

produced at different scale heights in the

chromosphere of main

-

sequence Sun

-

like stars.

5.

Diurnal and seasonal limitations of the

ground introduce aliasing which draws

power away from the planet signal

frequencies and puts them i

nto

frequencies that are aliases of one day

and one year. EarthFinder

provides

a

large

field of regard

(FOR)

and, for stars

outside the

FOR

, two long visibility

windows per year

which

completely

eliminates the diurnal alias and greatly

reduces the annual a

lias

(

Section

1.4.2

)

.

M

ultiple longitudinally

-

spaced ground

-

based telescopes

and PRV

spectrometers

will only

partially mitigate

daily aliases due to airmass optimization,

weather losses and time

-

varying zero

-

point velocity offsets between them.

6.

EarthFinder’s near continuous

observing capability and the efficiency

of its diffraction

-

limited spectrographs

give EarthFinder’s 1.45 m t

elescope an

effective light gathering power of a

much larger

ground

-

based facility.

7.

EarthFinder

is

perfectly suited to find

and characteriz

e

the masses and orbits

of the planets orbiting

~50

bright

main

sequence

stars (3<V<10 mag) which

will be the targets

for future NASA

flagship missions to image and obtain

spectra of nearby Earth

-

analogs.

8.

High resolving power

spectrograph

s

(R~150,000)

with simultaneous UV,

visible and NIR coverage offers exciting

new capabilities for general astrophysics

(

Section

1.5

).

9.

A preliminary TRL and cost estimate for

EarthFinder establishes this mission

concept as a Probe

-

class

($1B)

mission

with a Kepler

-

sized telescope using a

Kepler

-

deriv

ed spacecraft

.

1.1.2

STUDY RECOMMENDATIONS

We describe a roadmap for future

science and technology work to enable and

further refine and evaluate this mission concept

over the next decade:

“NASA and NSF should establish a strategic

initiative in extremely precis

e radial velocities

(EPRVs) to develop methods and facilities

for measuring the masses of temperate

terrestrial planets orbiting Sun

-

like stars.”

-

National Academies Exoplanet Science

Strategy

(ESS)

report top

-

level

recommendation, 2018

1.

Aligned with the t

op

-

level ESS

recommendation, we recommend the

immediate development of a testbed

(e.g. upgrade

-

able) diffraction limited

spectrograph facility with a target single

measurement precision and long

-

term

stability of 3 cm/s velocities to

investigate the mitiga

tion of stellar

and/or solar activity and

instrumentation development, to be

directly followed by a space PRV

mission. It is time now to commence

the development of the next generation

of PRV spectrometers, testing them on

the ground

first

but also with an

application for space. We envision a

testbed analogous to NASA JPL’s high

-

contrast imaging testbed facility which

combines detailed analysis of error

budgets with steady improvements in

performance. The facility would require

the necessary personnel and s

cience,

engineering and technical staff to

EarthFinder Mission Concept

2019 Probe Study Report

1

-

3

support its development.

This testbed

would initially support disk

-

integrated

Solar observations akin to the HARPS

Solar telescope feed, so as to correlate

and refine the analysis of the high

-

resolution spectrosco

pic data with the

wealth of information available from

heliophysics space and ground assets.

This work will be placed into context

of the vast wealth of information

currently being obtained from visible

wavelength seeing

-

limited

spectrometers that are now

operating

with instrument stability of 10

-

30 cm/s

(e.g. ESPRESSO, EXPRES, NEID),

and lay the groundwork for the follow

-

on space mission EarthFinder. This

demonstration must include addressing

questions of thermo

-

mechanical

stability under

realistic operating

conditions for a spacecraft operating at

L2. Design and experimental work

must be carried out so that each entry

in a detailed PRV error budget can be

determined with sufficient accuracy so

that the overall PRV precision can be

predicte

d.

2.

NASA should convene a workshop to

be held by PRV instrument designers,

Laser Frequency Comb (LFC) experts,

and space electronics engineers to lay

out a roadmap for future innovation and

TRL maturation. NASA should invest in

the development program

reco

mmended by these experts.

Wavelength standards such as laser

frequency combs can reduce the

requirement on absolute instrument

stability by turning many sources of

instrument instability into a common

-

mode error which can be reduced by

reference to a dense

, ultra

-

stable comb

of spectral lines. As discussed in

Section

2.3

, there remains significant work to

develop space qualified frequency

standards such as laser frequenc

y combs

or etalons capable of providing 1 cm/s

long term stability over 3

-

5 years. These

frequency standards must provide a

dense comb of lines in the visible (0.4

-

1.0 m) and NIR (1.0

-

2.

5

m) with few

G

H

z spacing so as to be resolvable with

EarthFinder’s sp

ectrometers.

3.

NASA should invest in a national data

analysis center or coordinated funding

activity to address the signal processing

required to model and mitigate the

effects of stellar activity. This effort

should comprehensively span the variety

of curr

ent and future approaches being

explored to mitigate stellar activity,

including line

-

by

-

line analysis, RV color,

time

-

dependent and physically

motivated modeling, extreme spectral

resolution, etc. to build comprehensive

and specialized processing tools an

d

statistical analyses. The scale of the

effort required

most likely

necessitates

the specialization of different teams, as

opposed to individual PI

-

led

teams

attempting to cover all aspects of stellar

activity mitigation.

4.

NASA should bridge the NASA

Astro

physics division with the extensive

expertise in Doppler spectroscopy of

the Sun from NASA Heliophysics. In

addition to theory and modeling efforts,

this includes experiments to extend

single

-

wavelength Solar Doppler

observations to space

-

based and/or

ball

oon

-

based, multi

-

wavelength

spanning Doppler measurements, and in

the NIR free of telluric contamination,

with the goal of both understanding our

Sun and building better models of stellar

activity for mitigating the PRVs of

nearby stars with EarthFinder.

EarthFinder Mission Concept

2019 Probe Study Report

1

-

4

1.2

STATE OF THE FIELD

The astronomical community is on the

cusp of fulfilling the NASA strategic goal to

“search for planetary bodies and Earth

-

like

planets in

orbit around other stars.” (U.S.

National Space Policy, June 28, 2010). The 2018

E

SS

report

recommends that “NASA should

lead a large strategic direct imaging mission

capable of measuring the reflected

-

light spectra

of temperate terrestrial planets orbitin

g Sun

-

like

stars.”

(

Charbonneau

et al. 2018)

“The radial velocity method will continue to

provide essential mass, orbit, and census

information to support both transiting and

directly imaged exoplanet science for the

foreseeable future.”

-

ESS

report top

-

l

evel

finding, 2018

Without precise radial velocity data,

some of NASA’s largest planned observatories

will fall short of the ultimate goal to determine

whether exoplanets can support life. PRVs will

provide several critical contributions to the

scientific

yield and optimization of a future

direct imaging mission such as HabEx or

LUVOIR which will survey the nearest fifty to

several hundred FGKM stars. First, the masses

of these planets as determined from PRVs

(prior or contemporaneous or otherwise) will be

needed for constraining the atmospheric

models. Second, the orbits of these planets as

determined by PRVs will be necessary to assess

habitability. Third, the target selection

optimization, observation timing, and required

number of direct imaging revisits

depend on

whether or not we will know a priori from

PRVs which nearby stars host Earth

-

mass

planets in Habitable Zone (HZ) orbits.

Currently, the PRV method achieves ~1

m/s single measurement precision with ground

-

based telescopes searching nearby stars

that are

known to be quiescent (magnetically inactive) to

minimize the negative impact of stellar activity

and thus to maximize planet mass sensitivity

(see

Section

1.4

). A new generation of visible

wavelength PRV instruments (ESPRESSO,

EXPRES, NEID) are on sky now. The best

stability demonstrated to date is ~30 cm/s on a

time

-

scale of hours within a single night on a

single quiescent target (ESPRESSO SPIE

conference pre

sentation, 2018), despite an

instrument precision requirement of 20 cm/s.

In order to push the sensitivity of the PRV

met

hod to 1

-

10 cm/s on all nearby stars

,

a

space

-

based PRV would provide a unique

platform to overcome many of the factors that

challenge

the current PRV performance from

the ground.

1.2.1

SCOPE OF STUDY

“Radial velocity measurements are currently

limited by variations in the stellar

photosphere, instrumental stability and

calibration, and spectral contamination from

telluric lines. Progress will

require new

instruments installed on large telescopes,

substantial allocations of observing time,

advanced statistical methods for data

analysis informed by theoretical modeling,

and collaboration between observers,

instrument builders, stellar astrophysic

ists,

heliophysicists, and statisticians.”

-

ESS top

-

level finding, 2018

T

able

1

-

1

:

The PRV method remains the best

technique for measuring planet masses, and the

second best for planet discovery after the transit

method (only a small fraction of planets, 1

-

10%, transit

their host s

tar).

Technique /

Quantity

Current

Performance

Earth

-

analog

Signal Amplitude

Transit /

Radius

~10 ppm in flux

(Kepler, TESS)

~100 ppm

PRV / Mass

~1 m/s (HARPS)

9 cm/s

Astrometry /

Mass

~5

-

16 μas (GAIA,

V<12 mag)

0.03

-

0.3 μas

(@10

-

100 pc)

Direct Imaging

/ Radius

10

-

6

flux contrast

10

-

10

flux contrast

EarthFinder Mission Concept

2019 Probe Study Report

1

-

5

This final report is the outcome of a

partial selection of our proposal submitted in

response to the NASA solicitation for the

Astrophysics Probes (APROBES) element of

NASA’s ROSES 20

16 (Appendix D.12;

NNH16ZDA001N). We were selected to

“establish the science case for going to space

with a precise radial velocity mission

with no

funding provided to develop a notional mission

architecture or provide mission design lab

sessions.” (NASA Headquarters selection letter).

Consequently, we evaluate the scientific

rationale for obtaining PRV measurements in

space, which is a two

-

part inqu

iry:

•

What can be gained from going to

space?

This is addressed in

Section

1.4

:

Evaluate the unique advantages that a space

-

based platform provides to enable the

identification and mitigation of stellar activity

for multi

-

planet signal recovery in PRV time

series.

•

What ca

n’t be done from the ground?

Section

1.3

:

Identify the PRV limit, if any,

introduced from micro

-

and macro

-

telluric

absorption in the Earth’s atmosphere.

Many unique

additional science cases

would also be possible with EarthFinder. We

highlight some of these science cases in

Section

1.5

,

including direct exoplanet spectroscopy for

characterization, stellar dynamos and

asteroseismology, fundamental atomic

transitions in the Sun and other stars, following

the water in the local

Universe obscured by

telluric water, and brown dwarf atmospheres.

To assess the technical and programmatic

feasibility of EarthFinder, we conducted a short

JPL TeamX study to develop an illustrative

mission concept to confirm the feasibility and

scope of t

he mission class

(

Section

4

).

The

TeamX study considered an earlier iteration of

the mission concept with a 1.1m primary, but the

science case presented herein is for

a 1.45m

primary (the same as the NASA Kepler probe

-

class mission). While we were not

funded to do

an integrated trade between the science yield and

mission architecture, the TeamX study establishes

the cost for EarthFinder at the top end of a

Probe

-

class m

ission.

1.2.2

EARTHFINDER OVERVIEW

The primary science goals of EarthFinder are

the precise radial velocity (PRV) detection,

precise mass measurement, and orbit

characterization of Earth

-

mass planets in

Habitable Zone orbits around the nearest

FGKM stars.

These

goals correspond to a PRV semi

-

amplitude

accuracy

of 1 cm/s on time

-

scales of

several years for a ~10% mass uncertainty

(

Figure

1

-

1

)

, which can be

achieved with 5

cm/s individual measurement precision and

taking advantage of binning down the

uncertainties from hundreds of measurements

.

The nominal spacecraft design is discussed in

more detail in

Section

4

,

but herein we provide

a brief summary.

EarthFinder is based upon the heritage

of Kepler spacecraft by Ball Aerospace, with a

1.45

-

m primary (diffraction limited to ~400

nm). The diffraction

-

limited beams of star

light

are

coupled into single

-

mode fibers illuminating

three high

-

resolution, compact and diffraction

-

limited spectrometer “arms”, one covering the

near

-

UV

(

280

-

380 nm), visible (VIS; 380

-

9

5

0

nm) and near

-

infrared (

NIR; 9

5

0

-

2500 nm)

respectively with a sp

ectral resolution of greater

than 150,000 in the visible and near

-

infrared

arms. The observatory is optimized for the

bright (V~

5

-

6

mag) nearby main sequence stars.

A small Solar telescope near the solar panels

would also be included to obtain Solar spectr

a.

EarthFinder Mission Concept

2019 Probe Study Report

1

-

6

Figure

1

-

1

: PRV

-

discovered exoplanets less than 10

M

Earth

as a function of stellar mass and planet mass

modulo the unknown inclination. Black circles are data

from the NASA Exoplanet Archive. The blue

-

green orb

corresponds to the Earth. The blue curve corresponds

to the approximate current detection limit of the PRV

method, the green curve corresponds to the NEID

spectrometer (or similarly, EXPRES, or ESPRESSO),

and the black curve corresponds to EarthF

inder and its

unprecedented

1 cms

-

1

sensitivity.

EarthFinder will be launched into an

Earth

-

trailing (similar to Kepler and Spitzer) or

Earth

-

Sun Lagrange

-

point orbit.

It

will have a

n

instantaneous

field of regard

(FOR)

of 70.7% of

the celestial sphere

, with a continuous viewing

zone covering 29% of the

sky

greater than 45

o

out of the Ecliptic plane, with 3

-

6 months of

visibility twice per year (e.g. 6

-

12 months total

per year across two “seasons” or visibility

windows) for targets within 45

o

of the Ecliptic

plane.

1.2.3

MISSION YIELD

EarthFinder will be

able to detect

Earth

-

mass planets and their planetary

companions around the nearest Sun

-

like

stars.

We carr

y

out a detailed simulation of the

yield of a five

-

year prime mission survey

focused on 61 nearby bright stars which would

be the likely targets of a

future flagship mission

to directly image and obtain spectra of Earth

-

analogs. Stars later than a spectral type of F2 are

listed in

Table

1

-

2

.

We start

with the HabEx mission study

target list (Gaudi et al 2018), and optimize the

target list and exposure times given the

EarthFinder spectrograph spectral grasp, and

known stellar properties including coordinates,

effective temperature, surface gravity, rota

tional

velocity, apparent magnitude, and metallicity.

We use the prescription in Beatty & Gaudi

(2015) to estimate the RV precision and the

exposure plus readout time necessary to reach a

photon detector noise of 3 cm/s across the

entire visible and NIR ar

ms.

The noise in each

arm is >3 cm/s but the velocity uncertainties

from both arms add in quadrature to 3 cm/s

precision, with the precision higher in the visible

arm given the relative RV information content,

but the NIR arm providing critical constraints

on

the stellar activity.

The median target dwell time to reach

this precision is 79 minutes. We then add in

quadrature a 3 cm/s instrumental error to both

arms. We take into account the visibility

windows of the spacecraft given an Earth

-

trailing orbit

(similar to Kepler) and known

target locations. Each successive target is

chosen randomly from the targets visible at

a

given

time to generate a uniform random

cadence for minimizing cadence aliases. We

include a target slew overhead and assume a

dedicated

(100% time) survey.

T

he absolute,

long

-

term stability of the

wavelength

calibration

is <1 cm/s by use of self

-

referenced laser frequency combs

(Section

2.3

)

.

Table

1

-

2

: HabEx Target Catalog used for EarthFinder

survey simulations

Common

Name

HIP

Number

Spec

Type

V mag

Included in

Ground

Survey?

GJ

15 A

1475

M2V

8.13

Yes

GJ 15 B

1475B

M3.5V

11.04

Yes

Zet Tuc

1599

F9.5V

4.23

No

Bet Hyi

2021

G0V

2.79

No

54 Psc

3093

K0.5V

5.88

Yes

HD 4628

3765

K2.5V

5.74

Yes

eta Cas

3821

F9V

3.44

Yes

107 Psc

7981

K1V

5.24

Yes

tau Cet

8102

G8V

3.50

Yes

EarthFinder Mission Concept

2019 Probe Study Report

1

-

7

Table

1

-

2

: HabEx Target Catalog used for EarthFinder

survey simulations

Common

Name

HIP

Number

Spec

Type

V mag

Included in

Ground

Survey?

HD

10780

8362

K0V

5.63

Yes

HD 16160

12114

K3V

5.83

Yes

HD 17925

13402

K1V

6.05

Yes

iot Per

14632

G0V

4.05

Yes

zet01 Ret

15330

G2.5V

5.54

No

zet02 Ret

15371

G1V

5.24

No

kap01 Cet

15457

G5V

4.85

Yes

e Eri

15510

G6V

4.27

No

eps Eri

16537

K2V

3.73

Yes

del Eri

17378

K0IV

3.54

Yes

omi02 Eri

19849

K0V

4.43

Yes

HD 32147

23311

K3V

6.21

Yes

GJ 191

24186

M1V

8.853

No

lam Aur

24813

G1.5IV

4.71

Yes

HD 36395

25878

M1.5V

7.968

Yes

alf Men

29271

G7V

5.09

No

HD 42581

29295

M1V

8.125

Yes

HD 50281

32984

K3.5V

6.57

Yes

alf CMi

37279

F5IV

0.37

Yes

11 LMi

47080

G8V

5.34

Yes

HD 88230

49908

K6V

6.61

Yes

36 UMa

51459

F8V

4.82

Yes

HD 95735

54035

M2V

7.52

Yes

61 UMa

56997

G8V

5.34

Yes

HD 102365

57443

G2V

4.88

No

bet Vir

57757

F9V

3.60

Yes

bet CVn

61317

G0V

4.25

Yes

bet Com

64394

F9.5V

4.25

Yes

61 Vir

64924

G6.5V

4.74

Yes

eta Boo

67927

G0IV

2.68

Yes

HD 122064

68184

K3V

6.52

Yes

V645 Cen

70890

M5.5V

11.13

No

DE Boo

72848

K0.5V

6.01

Yes

HD 131977

73184

K4V

5.72

Yes

lam Ser

77257

G0V

4.42

Yes

zet TrA

80686

F9V

4.91

No

12 Oph

81300

K1V

5.77

Yes

V2215 Oph

84478

K5V

6.34

Yes

41 Ara

84720

G8V

5.48

No

HD 166620

88972

K2V

6.40

Yes

sig Dra

96100

K0V

4.68

Yes

bet Aql

98036

G8IV

3.71

Yes

del Pav

99240

G8IV

3.56

No

HD 191408

99461

K2.5V

5.32

No

Table

1

-

2

: HabEx Target Catalog used for EarthFinder

survey simulations

Common

Name

HIP

Number

Spec

Type

V mag

Included in

Ground

Survey?

HD 192310

99825

K2V

5.723

Yes

61 Cyg A

104214

K5V

5.21

Yes

61 Cyg B

104217

K7V

6.03

Yes

AX Mic

105090

M1V

6.68

No

eps Ind

108870

K5V

4.69

No

TW PsA

113283

K4V

6.48

No

HD 217987

114046

M2V

7.34

No

HD 219134

114622

K3V

5.57

Yes

Targets not included in space mission simulation:

ups And

7513

F9V

4.10

Yes

del Tri

10644

G0.5V

4.87

Yes

tet Per

12777

F8V

4.11

Yes

pi.03 Ori

22449

F6V

3.19

Yes

gam Lep

27072

F6V

3.60

Yes

HD 103095

57939

K1V

6.45

Yes

gam Ser

78072

F6V

3.84

Yes

We compare our EarthFinder simulated

survey to a ground

-

based survey yield for a

subset of 44 of the 61 space targets, plus 7

additional targets (51 total) accessible from a

Northern Hemisphere facility. We assume a

“super”

-

NEID, also capable of achieving

3

cm/s instrument stability, but with the

wavelength coverage, efficiency and resolution

of NEID. We place the super

-

NEID on the

LBT, a

n

8

-

m class telescope. We account

for

the properties of the spectrograph including

spectral resolution, spectral grasp an

d

throughput efficiency, detector noise and read

out times, as well as the slew

-

rate and pointing

limits of the telescope. We account

for

target

airmass and hour angle to optimize sequential

target selection to minimize airmass and to

minimize the time sin

ce a given target was last

observed, based upon the MINERVA survey

dispatch scheduler (Newman et al. in prep.). We

also account

for the

realistic weather losses,

sunrise and sunset times for Arizona. We

assume a PRV survey using 25% of the total

nights ava

ilable for five years.