Understanding systematics in exoplanet observations

Pieter Deroo (1), Mark Swain (1)) & Tom Greene (2) 1: NASA/Jet Propulsion Laboratory, California Institute of Technology; 2: NASA/Ames

National Aeronautics and Space Administration

Abstract: The most successful exoplanet characterization method to date – combined light photometry – relies on extremely accurate spectrophotometric differential precision. The required dynamic range has pushed exoplanet observations into previously unexplored instrument systematics. Mitigating these small but very significant effects is core to exoplanet science and the methodology how to remove the instrument noise is frequently a source of controversy. Here, we model the systematic noise floor induced by the most significant instrument systematic effects. We focus on near-IR spectroscopic characterization of exoplanet atmospheres using HgCdTe detectors. We establish the: 1.  Raw systematics noise floor 2.  Decorrelated/post-processed systematics noise floor and this for a range in instrument parameters (right-hand side). Three specific instruments are considered as well (bottom): HST/NICMOS which allowed for the first spectroscopic comparative exoplanetology; JWST/NIRSpec &FINESSE, which are future instruments that will characterize exoplanet atmospheres from the optical to the near-IR (~ 0.5 – 5 µm). FINESSE is a 76 cm aperture Explorer mission with well-sampled spectra at R ~ 1000.

Methodology: A Monte Carlo simulation in which common instrument errors (pointing & focus errors) are used as forcing functions for photometric variation induced by:

1.  Position dependent gain variations and uncertainties in the focal plane array 2.  Slit-losses

Various  illumina,on  pa.erns  –  PSFs  with  different  width  and  posi,on  (le:)  –  are  propagated  onto  a  detector  with  a  specific  inter  and  intra  pixel  gain  varia,on  (middle)  and  the  oversampled  PSF  is  read  out  at  the  different  detector  pixels  (right).    The  FPA  is  Monte  Carlo  simulated;  the  ensemble  of  MC  simula,ons  together  with  a  sta,s,cal  distribu,on  in  poin,ngs  and  PSF  sizes  (FWHM)  is  used  to  establish  the  systema,c  noise  floor.    

Instrument design vs spectrophotometric differential precision:

Raw  Capability   Decorrelated  

Assuming: state-of-the-art HgCdTe inter and intra pixel response; Gaussian instrument errors. Overall systematics noise floor: combine the different partials appropriately The approximate location for NIRSpec & FINESSE @ 2 µm is annotated. The error distribution for those instruments can significantly differ from the assumed Gaussian. Our instrument specific results are given below.

Instrument specific systematics

Raw instrument noise partials @ 1.7 µm

HST/NICMOS   JWST/NIRSpec   FINESSE  

Poin,ng  induced  gain  changes   480  ppm/orbit   364  ppm/10,000s   17  ppm/orbit  

Systema,c  PSF  varia,ons   932  ppm/orbit   3434  ppm/2wks   60  ppm/orbit  

Poin,ng  induced  slit-­‐loss   NA   1  ppm/10,000s   22  ppm/orbit  

HST/NICMOS   JWST/NIRSpec   FINESSE  

Poin,ng  induced  gain  changes   21ppm/orbit   75  ppm/10,000s   0.3  ppm/orbit  

Systema,c  PSF  varia,ons   145ppm/orbit   49  ppm/2wks   4  ppm/orbit  

Poin,ng  induced  slit-­‐loss   NA   1  ppm/10,000s   2  ppm/orbit  

Decorrelated instrument noise partials @ 1.7 µm

HST/NICMOS   JWST/NIRSpec   FINESSE  

RAW   1048  ppm/eclipse   1052  ppm/eclipse   67  ppm/eclipse  

DECORRELATED   146  ppm/eclipse   217  ppm/eclipse   5  ppm/eclipse  

Overall systematics noise floor per eclipse [8 hr] @ 1.7 µm

Systematics and exoplanet spectroscopy:

Simulated  spectra  (top  panel)  for  the  Super-­‐Earth  GJ  1214b  and  the  warm  Neptune  GJ  436b  as  observed  with  FINESSE  (middle  panel)  and  JWST/NIRSpec  (bo.om  panel)  taking  into  account  all  sources  of  noise,  including  systema,c  noise.  The  noise  for  FINESSE  is  dominated  by  photon  noise;  for  NIRSpec,  systema,c  noise  dominates  spectra  at  the  short  wavelength  range.  To  illustrate  the  fact  that  NIRSpec  requires  mul,ple  instrument  secngs  to  cover  the  en,re  wavelength  range  (and  thus  mul,ple  eclipses),  we  have  shaded  the  spectral  region  according  to  the  minimum  number  of  instrument  secngs  required.    

Conclusions: Our simulations quantify the expected systematic noise floor for exoplanet spectroscopic observations and the improvements using the current-best decorrelation/post-processing [right, bottom]. Our simulations for HST/NICMOS are in agreement with the observed pre- and post-processed values, validating the simulations [bottom]. Simple trades in the instrument design can significantly optimize it for exoplanet spectrosopy [right]. As such, the well-sampled spectra of a modest aperture mission like FINESSE could have better limiting signal-to-noise than the poorly sampled JWST/NIRSpec at short wavelengths [bottom].

!"#$%&'&()*++( !"#$%%$(,-(../++01,**)-( 2'345%6(78235'63'(9:#$':'64;45%6(

75<=&'(7>7-?+(@2#;3'3&;A4(7%$B%=4CD(E;<'(7?F(;6B(7?G(HI'(%&(B5I3$%I=&'(%A(56A%&:;45%6(3%64;56'B(%6(4J5I(IJ''4(5I(I=KL'34(4%(4J'(&'I4&5345%6(%6(4J'(M'I4&5345N'(.%453'(#;<'(%A(4J5I(#&%#%I;$>(

(

!"#$%&'%()*+

,-$)#&."##)*&/00)12*3

,-$)#&'4*"#&/##"3

5)6*43"2*)&'-7&'%"+)

877)#&'4*"#&/##"3

9/:

'6);$#46%4$41)$)#

!<);$"2*)&=4>)#

?(7@*)$!"#$%&'(')*+,- '6";);#"A$&4>)#>()B&"7+&1"#@(70C

'./01%&'(')*+#- :8D!''!&A($0&(7&,6$(47&/E&FG″ '$"$(;&!7>)*46)C

!"#$%&'(')*+#- /16*)&64B)#&@)7)#"$(47&$%#4-@%&"**&'H=&46)#"$(470C

'4*"#&/##"3&0$#-;$-#)&";$0&"0&

'-7&0%()*+

ICJK&1

LCFK&1

M)7($%&/7$)77"

N46&5);O

'4*"#&/##"3&9"7)*

'$"#&N#";O)#

P"+("$4#&9"7)*BH=44*(7@&9(6)

=#34;44*)#D"+(#&

/7$)77"

5)6*43"2*)'-7&'%"+)&Q0$4B)+R

P)";$(47&?%))*QI&6*";)0R

870$#-1)7$&!*);$#47(;0

."$$)#3QS($%(-1T847R

=#34;44*)#

N4#U-)&P4+0QI&6*";)0R

P)";$(47&?%))*!*);$#47(;0

8/V

D"+(#&/7$)77"

=#34;44*)#!*);$#47(;0

'T."7+N#"70647+)#W"@7)$41)$)# W8WV

!"#$%&'(')*+"- S"-7;%&1"00&1"#@(7&(0&*"#@)C

'./01%&'(')*+"- '30$)1&2*4;O&+("@#"10%4B0&:8D!''!&%"0&%(@%&%)#($"@)C

20#3435%6

2,.%7,%)839:;

2"<%=>?%9:; @>66%753

=X5Y Z[ Z[N)*);411-7(;"$(470 LZ I\N%)#1"*&=47$#4* Z\ Z\/5=' JL IG9"3*4"+ LIZ [!*);$#(;"*&94B)#&Q*400)0R LZ J P)0(0$(>)&%"#7)00&X&2"$$)#3&*400)0

)#3%1A"5>14&!>5"$ B+C +DE

2>$"1&F11"4&92F;&G>H%1&I%J0.1%6%75& EDK LEM

'/&64B)#&@)7)#"$(47&#)U-(#)1)7$&Q2"0)+&47&1"](1-1&);*(60)&+-#"$(47L&"7+&I[&+)@#))&64(7$(7@&4AA0)$&A#41&'-7R

'/&="6"2(*($3&T&.,S JJL JJL ."0)+&47&'/&0(̂ )&"7+&)AA(;()7;3'/&="6"2(*($3&T&!,S JLG JLG V0)+&$4&+)$)#1(7)&64B)#&1"#@(7

G>H%1&="1/.7 BLN +EON

_"*-)0&(7;*-+)&;47$(7@)7;3&#"7@(7@&A#41&G` &$4&I[` &2"0)+&47&;41647)7$&%)#($"@)

L&W"](1-1&);*(60)&+-#"$(47&QGa&1(7-$)0R&#)0-*$0&(7&KK` &#)+-;$(47&(7&">)#"@)&&&&4#2($"*&64B)#

@011%75&P%35&Q35.6"5%&="33&

9R/;

="33&S1>H5T&

@>75&9N;

="U&QU8%,5%?&="33&9R/;

LIICJ GK` LJFC\KC\ Ga` FC\

LKCZ La` G[C[JC[ a` JCIJCJ L[` KCG

LFC\ Ga` GZC\GIC\ a` GaC[LJCJ aa` GaCK

GILCJ Ga` G\\CJLZKCG LJ` LK[CLIK\C\ GL` Za\CK

\L[C[KK`

Y"#7)00&N4$"*&W"00'6";);#"A$&T&N4$"*&W"00

W);%"7(010&N4$"*&W"00!95'&N4$"*&W"00

'$#-;$-#)0&N4$"*&W"00

=X5Y&N4$"*&W"00

/5='&N4$"*&W"00

20#3435%6VW5%6

9"3*4"+&T&N4$"*&W"00,20)#>"$4#3&T&N4$"*&W"00

N)*);41&N4$"*&W"00N%)#1"*&N4$"*&W"00

S"-7;%&_)%(;*)&="6"2(*($3W"00&W"#@(7

,6)#"$(47"*&S(A)$(1),#2($&9"#"1)$)#0S_&=416"$(2(*($3/5='&/#;%($);$-#)P)+-7+"7;3

!*);$#(;"*&94B)#

N%)#1"*&=47$#4*

N)*);411-7(;"$(470/>(47(;0&/#;%($);$-#)W);%"7(010:"-*$&9#4$);$(47

GZ&14&Q0;()7;)&460R&b&G&14&Q47T4#2($&;%);O4-$R

:9/&"7+&06);$#41)$)#&";$(>)*3&;44*)+c&$)*)0;46)&"7+&'H=&6"00(>)*3&;44*)+'T."7+&+(#);$&$4&@#4-7+&-0(7@&D!Dc&7)"#&417(T+(#);$(47"*&"7$)77"0.#4"+&P)";%&87$)@#"$)+&/>(47(;0&V7($&B($%&P/5Ka[&6#4;)004#

)#3%1A"5>14&&X%4&@T"1",5%1.35.,3

FKCJ&+)@c&aK[TO1&'-7T037;%&4#2($&B($%&J&/W&"0;)7+(7@&74+)DS'T88&,6$(47&/c&N"-#-0&IGL[&4#&/$%)7"&88&FG&(7C&:"(#(7@c&I\d&9/:IT"](0&0$"2(*(̂ )+c&0$)**"#&(7)#$("*c&0-66*)1)7$)+&B($%&6"3*4"+&:e''(7@*)&0$#(7@&B($%&0"A)&14+)&X&0)*);$&2";O-60&Q;*"00&=&1(00(47R

!<);$"2*)&;4>)#&"7+&+)6*43)+&0-70%"+)D4&0-7*(@%$&(7&6"3*4"+&T&'-7&">4(+"7;)&:'?&$)0$)+&47T4#2($&?8'!

5(#);$&!7)#@3&N#"70A)#c&GCF&1G&24+3TA(])+&"##"3&T&(77)#H4-$)#&6"7)*&"00)12*3

I%J0.1%6%753&"7?&="1/.73 I%J0.1%6%75 @"8"#.$.54 ="1/.7S"-7;%&W"00 Za\CK&O@ f\L[&O@ KK`';()7;)&5"$"&'$4#"@)&="6";($3 LCG&e. Z&e. GII`94B)#&e)7)#"$(47&BH&I[&+)@&4AA0)$ ZJa&? JLG&?&Q!,SR Ia`5"$"&_4*-1) Z[[&e.H+"3 J[[&e.H+"3 a[`S4'&g74B*)+@)&QD4#1"*&94(7$R&QL R f&\CI&"#;0);& f&JCG&"#;0); IZ`S4'&=47$#4*&QD4#1"*&94(7$R&QL R f&\CI&"#;0); f&JCJ&"#;0); GJ`S4'&'$"2(*($3&Q:(7)&94(7$R&QL O(4>)#&Z[&1(7R f&[C[Za&"#;0); f&[C[IG&"#;0); Z[`W"](1-1&/$$($-+)&'*)B&P"$) J&+)@H1(7 L\&+)@H1(7 G[[`W"](1-1&')$$*(7@&N(1) a&1(7 f&L&1(7 Z[[`

9"00(>)&;44*(7@L ;44*&6"3*4"+&$4&LI[&g

LLJ&g&B4#0$&;"0)cL[I&g&">)#"@)

JI` &B4#0$&;"0)cLaZ` &">)"@)

/;$(>)&;#34;44*)#&64B)#G IZ&? Z\&?&Q1"]&(76-$R ZL`'6);$#41)$)#&$)16)#"$-#)&0$"2(*($3c&L[&%4-#0 L&g [Ca&g L[[`:9/&$)16)#"$-#)&0$"2(*($3c&L[&%4-#0 a[&1g Ga&1g L[[`:9/&$)16)#"$-#)&0$"2(*($3c&G&3#0 [Ca&g [CLg Z[[`L&9"00(>)&;44*(7@&1"#@(7&(0&2"0)+&47&#"+("$(>)&;"6"2(*($3c&B%(;%&>"#()0&"0&NZ

G&N%)&;#3444*)#&;"7&)];))+&$%)&%)"$&*(A$&#)U-(#)1)7$&23&I[[` c&2-$&"$&@#)"$)#&(76-$&64B)#&$%"7&"**4;"$)+

FINESSE

Fast Infrared Exoplanet Spectroscopy Survey Explorer

NIRSpec  @    

10,000s  

NIRSPec  @    

10,000s  

FINESSE  @  

orbit-­‐to-­‐orbit  

FINESSE  @  

orbit-­‐to-­‐orbit  

NIRSpec  @  2  weeks   NIRSpec  @  2  weeks  

FINESSE  @  

orbit-­‐to-­‐orbit  

FINESSE  @  

orbit-­‐to-­‐orbit  

NIRSpec   NIRSpec  

FINESSE   FINESSE  

©Copyright  2011.  All  rights  reserved.