+ All Categories
Home > Documents > The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

Date post: 13-Jan-2016
Category:
Upload: caron
View: 27 times
Download: 0 times
Share this document with a friend
Description:
Gaia First Look: Description and Status Report. The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004. - PowerPoint PPT Presentation
Popular Tags:
35
The Three Dimensional Universe with GAIA The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004 Paris-Meudon, October 4-7, 2004 Gaia First Look: Gaia First Look: Description and Status Description and Status Report Report Stefan Jordan, Uli Bastian, Helmut Lenhardt, Hans Stefan Jordan, Uli Bastian, Helmut Lenhardt, Hans Bernstein, Sonja Hirte (Astronomisches Rechen- Bernstein, Sonja Hirte (Astronomisches Rechen- Institut, Heidelberg) & Michael Institut, Heidelberg) & Michael Biermann (Landessternwarte, Heidelberg Biermann (Landessternwarte, Heidelberg ) ) Acknowledgement: Acknowledgement: - GIS routines by Lennart Lindegren (LL-37) - GIS routines by Lennart Lindegren (LL-37) - simulation data from Barcelona group (Eduard - simulation data from Barcelona group (Eduard Masana et al.) Masana et al.)
Transcript
Page 1: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

The Three Dimensional Universe with GAIAThe Three Dimensional Universe with GAIAParis-Meudon, October 4-7, 2004Paris-Meudon, October 4-7, 2004

Gaia First Look:Gaia First Look:Description and Status ReportDescription and Status Report

Stefan Jordan, Uli Bastian, Helmut Lenhardt, Hans Bernstein, Stefan Jordan, Uli Bastian, Helmut Lenhardt, Hans Bernstein, Sonja Hirte (Astronomisches Rechen-Institut, Heidelberg) Sonja Hirte (Astronomisches Rechen-Institut, Heidelberg)

& Michael Biermann (Landessternwarte, Heidelberg& Michael Biermann (Landessternwarte, Heidelberg))

Acknowledgement:Acknowledgement:- GIS routines by Lennart Lindegren (LL-37)- GIS routines by Lennart Lindegren (LL-37)- simulation data from Barcelona group (Eduard Masana et al.)- simulation data from Barcelona group (Eduard Masana et al.)

Page 2: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 2First Look, Status Report

Often heard here:Often heard here:

• GAIA is a self calibrating instrumentGAIA is a self calibrating instrument

Page 3: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 3First Look, Status Report

Often heard here:Often heard here:

• GAIA is a self calibrating instrumentGAIA is a self calibrating instrument• This is true ….This is true ….

Page 4: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 4First Look, Status Report

Often heard here:Often heard here:

• GAIA is a self calibrating instrumentGAIA is a self calibrating instrument• This is true ….This is true ….• However, GIS possible only after 6 months However, GIS possible only after 6 months

Page 5: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 5First Look, Status Report

Often heard here:Often heard here:

• GAIA is a self calibrating instrumentGAIA is a self calibrating instrument• This is true ….This is true ….• However, GIS possible only after However, GIS possible only after 6 months6 months• Gaia can only solve for Gaia can only solve for calibrationcalibration, , attitudeattitude

and and source parameterssource parameters simultaneouslysimultaneously..

Page 6: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 6First Look, Status Report

The basic problemThe basic problem

• It is difficult to immediately assess proper It is difficult to immediately assess proper functioning of all elements functioning of all elements (on the (on the μμas level)as level)

• But But very importantvery important, else risk of losing , else risk of losing many months of data and mission time.many months of data and mission time.

Some compromise must be foundSome compromise must be found

Page 7: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 7First Look, Status Report

Overall QL/FL SchemeOverall QL/FL Scheme• Quick Look (QL) - Quick Look (QL) - ESOCESOC• Science Quick Look (Sc-QL) – Science Quick Look (Sc-QL) – ESOC + FL taskESOC + FL task

Sc-QL monitorSc-QL monitor Sc-QL evaluatorSc-QL evaluator

• Early science processing steps - Early science processing steps - Core ProcessingCore Processing Ingestion, cross-matching, initial data treatmentIngestion, cross-matching, initial data treatment

• First-Look pre-processing (FLP) - First-Look pre-processing (FLP) - FL taskFL task ODIS (one-day iterative solution)* or ring solutionODIS (one-day iterative solution)* or ring solution

• Detailed First Look (DFL) - Detailed First Look (DFL) - FL taskFL task First-Look monitorFirst-Look monitor First-Look evaluatorFirst-Look evaluator * formerly called 1-day * formerly called 1-day

GISGIS

Page 8: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 8First Look, Status Report

ScanScancirclescircles

RGCRGC

at the node:at the node:differentdifferent

scan directionsscan directions

90 deg off the node:90 deg off the node:only oneonly one

scan directionscan direction

Scan directions and locationlocation on the sky

Different scan directions: Across-scan unknowns solvableDifferent scan directions: Across-scan unknowns solvable

Equal scan directions: Across-scan unknowns undefinedEqual scan directions: Across-scan unknowns undefined

Page 9: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 9First Look, Status Report

Two basic ways of attackTwo basic ways of attack

1.1. Some reduced variant of GIS, now named Some reduced variant of GIS, now named „ODIS“, One-Day Iterative Solution„ODIS“, One-Day Iterative Solution

2.2. Direct solution of the full problem of the 1-day Direct solution of the full problem of the 1-day astrometric reduction, named „astrometric reduction, named „Ring SolutionRing Solution“ “

being performed in parallelbeing performed in parallel

Note: Neither way can produce a full astrometric Note: Neither way can produce a full astrometric solution. solution.

Page 10: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 10First Look, Status Report

ODIS versus Ring SolutionODIS versus Ring Solution

• Ring SolutionRing Solution::

- allows to easily and clearly detect and remove the degeneracies- allows to easily and clearly detect and remove the degeneracies- can in principle produce full covariances- can in principle produce full covariances- may be much more demanding in terms of storage and computing- may be much more demanding in terms of storage and computing

• ODISODIS::- a block-iterative scheme does not necessarily show its degeneracies- a block-iterative scheme does not necessarily show its degeneracies- a block-iterative scheme does not necessarily converge- a block-iterative scheme does not necessarily converge- the degeneracies are not easily pinned down- the degeneracies are not easily pinned down

Page 11: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 11First Look, Status Report

First-Look Preprocessing using Ring First-Look Preprocessing using Ring Solution:Solution:

Direct elimination of source unknownsDirect elimination of source unknownsSources Attitude Cal.

6 000 000

6 000

900

0

0

x x

f u l l

Page 12: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 12First Look, Status Report

ODIS versus Ring SolutionODIS versus Ring Solution

• Ring SolutionRing Solution::

- allows to easily and clearly detect and remove the degeneracies- allows to easily and clearly detect and remove the degeneracies- can in principle produce full covariances- can in principle produce full covariances- may be much more demanding in terms of storage and computing- may be much more demanding in terms of storage and computing

• ODISODIS::

- a block-iterative scheme does not necessarily show its degeneracies- a block-iterative scheme does not necessarily show its degeneracies- a block-iterative scheme does not necessarily converge- a block-iterative scheme does not necessarily converge

Page 13: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 13First Look, Status Report

Initial estimates for 24 hours of data, using (various) present-day Initial estimates for 24 hours of data, using (various) present-day 2 GHz Pentium III (4 GByte memory) PCs2 GHz Pentium III (4 GByte memory) PCs

•About About 5 GByte5 GByte of observed data of observed data•→→50 million parameters50 million parameters (attitude, calibration (attitude, calibration parameters, source parameters, parameters, source parameters, global parametersglobal parameters))•Computing time (extrapolated to 50 million elementary observations) without data base overhead:•Ring solution: 1600 hours•ODIS: 40 hoursAt time of GAIA: faster computersParallelization possible!

Data volume, computing timeData volume, computing time

Page 14: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 14First Look, Status Report

Lessons already learnedLessons already learned

• Either FLP method (ODIS or ring solution) needs a Either FLP method (ODIS or ring solution) needs a „zeroth“ iteration„zeroth“ iteration to identify the degenerate dimensions to identify the degenerate dimensions

• Either FLP method can conveniently useEither FLP method can conveniently use quaternions quaternions in in the RGC system for the attitude representation the RGC system for the attitude representation (GAIA-SJ-002)(GAIA-SJ-002)

• Convergence behaviourConvergence behaviour of ODIS depends on block of ODIS depends on block sequence, on block size, on block composition, .....sequence, on block size, on block composition, .....We‘ll use more than 4 blocks (split AL/AC), and omit the We‘ll use more than 4 blocks (split AL/AC), and omit the ‚global‘ block (GAIA-SJ-001)‚global‘ block (GAIA-SJ-001)

• In smaller test runs: distribution of observations over time In smaller test runs: distribution of observations over time interval very important interval very important

Page 15: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 15First Look, Status Report

ODIS IterationsODIS Iterations

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Before first iteration:Before first iteration:

large scale calibration wronglarge scale calibration wrong

FOV 1FOV 1 FOV 2FOV 2

-23000 sources-23000 sources

-125000 elementary -125000 elementary “observations”“observations”

Page 16: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 16First Look, Status Report

ODIS IterationsODIS Iterations

After first iteration:After first iteration:

large scale calibration adjustlarge scale calibration adjust

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 17: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 17First Look, Status Report

ODIS IterationsODIS Iterations

After second iteration:After second iteration:

large scale calibration adjustslarge scale calibration adjusts

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 18: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 18First Look, Status Report

ODIS IterationsODIS Iterations

After 24 iteration:After 24 iteration:

large scale calibration adjusted…large scale calibration adjusted…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 19: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 19First Look, Status Report

ODIS IterationsODIS Iterations

After 50 iteration:After 50 iteration:

large scale calibration adjusted…large scale calibration adjusted…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 20: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 20First Look, Status Report

ODIS IterationsODIS Iterations

After 50 iteration:After 50 iteration:

large scale calibration adjusted…butlarge scale calibration adjusted…but

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 21: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 21First Look, Status Report

ODIS IterationsODIS Iterations

Both FOV afterBoth FOV after

50 iteration:50 iteration:obse

rved

-com

pute

d (a

long

-sca

n)/

obse

rved

-com

pute

d (a

long

-sca

n)/ μμ

asas

time/daystime/days

large scale calibration large scale calibration

Page 22: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 22First Look, Status Report

ODIS IterationsODIS Iterations

FOV 1 afterFOV 1 after

50 iteration:50 iteration:obse

rved

-com

pute

d (a

long

-sca

n)/

obse

rved

-com

pute

d (a

long

-sca

n)/ // μ

μ asas

time/daystime/days

large scale calibration large scale calibration

Page 23: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 23First Look, Status Report

ODIS IterationsODIS Iterations

FOV 2 afterFOV 2 after

50 iteration:50 iteration:obse

rved

-com

pute

d (a

long

-sca

n)/

obse

rved

-com

pute

d (a

long

-sca

n)/ μμ

asas

time/daystime/days

large scale calibration large scale calibration

Page 24: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 24First Look, Status Report

ODIS IterationsODIS Iterations

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Before first iteration:Before first iteration:

large scale calibration wronglarge scale calibration wrong

FOV 1FOV 1 FOV 2FOV 2

Page 25: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 25First Look, Status Report

ODIS IterationsODIS Iterations

Only CCD calibrationOnly CCD calibration

If we do it right…If we do it right…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 26: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 26First Look, Status Report

ODIS IterationsODIS Iterations

After 2 iterationAfter 2 iteration

If we do it right…If we do it right…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 27: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 27First Look, Status Report

ODIS IterationsODIS Iterations

After 1 iterationAfter 1 iteration

If we do it right…If we do it right…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 28: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 28First Look, Status Report

ODIS IterationsODIS Iterations

After 5 iterationAfter 5 iteration

If we do it right…If we do it right…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 29: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 29First Look, Status Report

ODIS IterationsODIS Iterations

After 10 iterationAfter 10 iteration

If we do it right…If we do it right…

observed-computed (along-scan)/observed-computed (along-scan)/μμasas

Page 30: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 30First Look, Status Report

Lessons already learned (cont.)Lessons already learned (cont.)

• No across-scan instrument calibration possible within a No across-scan instrument calibration possible within a short amount of timeshort amount of time

• →→Commissioning and initial in-orbit calibration would Commissioning and initial in-orbit calibration would greatly benefit from a greatly benefit from a „zero-nu-dot“„zero-nu-dot“ scanning mode (i.e. scanning mode (i.e. scan axis remaining on the ecliptic, at 50 degrees from scan axis remaining on the ecliptic, at 50 degrees from the sun)the sun)

Page 31: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 31First Look, Status Report

““Zero-nu-dot mode”Zero-nu-dot mode”Coverage of the ecliptic pole areaCoverage of the ecliptic pole area

Note: z-axis of the satellite points toward the top (on Note: z-axis of the satellite points toward the top (on day 1); sun is either 50º to the right or 50º to the left day 1); sun is either 50º to the right or 50º to the left

of it.of it.

about 8000 starsabout 8000 stars

Page 32: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 32First Look, Status Report

Lessons already learned (cont.)Lessons already learned (cont.)

• Commissioning and initial in-orbit calibration would greatly Commissioning and initial in-orbit calibration would greatly benefit from a benefit from a „zero-nu-dot“„zero-nu-dot“ scanning mode (i.e. scan axis scanning mode (i.e. scan axis remaining on the ecliptic, at 50 degrees from the sun)remaining on the ecliptic, at 50 degrees from the sun)

• This will provide us with a first calibration also across-scanThis will provide us with a first calibration also across-scan• Needed: Needed: Pre-GAIA astrometric observationsPre-GAIA astrometric observations of some 8000 of some 8000

stars around the ecliptic polesstars around the ecliptic poles

Page 33: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 33First Look, Status Report

Lessons already learned (cont.)Lessons already learned (cont.)

• Commissioning and calibration would greatly benefit from Commissioning and calibration would greatly benefit from an „an „AC calibrationAC calibration“ CCD sampling mode (i.e. strongly “ CCD sampling mode (i.e. strongly reduced across-scan binning in AF; in practical terms this reduced across-scan binning in AF; in practical terms this can be implemented by temporarily using the WY00 can be implemented by temporarily using the WY00 windows over a larger magnitude range)windows over a larger magnitude range)

Page 34: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 34First Look, Status Report

First-Look Preprocessing using ODIS:First-Look Preprocessing using ODIS:Preliminary conceptPreliminary concept

• First, produce a global solution for the First, produce a global solution for the along-RGC unknownsalong-RGC unknowns only, already omitting those stars with only 1 FoV transit. only, already omitting those stars with only 1 FoV transit.• In this process, collect the In this process, collect the list of actual scan directionslist of actual scan directions for for each star each star• For those stars with sufficiently differing scan directions,For those stars with sufficiently differing scan directions, solve solve across-scan coordinatesacross-scan coordinates in subsequent solutions. in subsequent solutions. • For all across-scan attitude unknowns collect the number For all across-scan attitude unknowns collect the number of stars with across-scan solutions contributing to them. of stars with across-scan solutions contributing to them.• For those across-scan attitude unknowns with sufficiently For those across-scan attitude unknowns with sufficiently many such stars contributing, make a first solution. many such stars contributing, make a first solution.• In all further iterations, omit or „Huber away“ discordantIn all further iterations, omit or „Huber away“ discordant observations, using all unknowns selected before. observations, using all unknowns selected before.

• After first GIS: the first+last of these steps suffice (probably)After first GIS: the first+last of these steps suffice (probably)

Page 35: The Three Dimensional Universe with GAIA Paris-Meudon, October 4-7, 2004

GST13, ESTEC, 2004-09-13 35First Look, Status Report

The first look task is The first look task is not a closed not a closed

session session – – collaboration is collaboration is

welcomewelcome

ENDEND


Recommended