Transcript Holsclaw

UVIS Calibration Update
Greg Holsclaw, Bill McClintock
Jan 6, 2009
Outline
•
•
•
•
•
Recent calibration observations
FUV degradation and Spica variability
Stellar flux comparisons with SOLSTICE
The flat field for small / unresolved targets
Changes in the sensitivity at Lyman alpha
Recent UVIS Calibrations
• The last two Spica calibrations suffered
data dropouts due to DSN issues:
– FUV2008_265_23_40_33_UVIS_086IC_ALPVIR001_PRIME
– FUV2008_309_21_10_33_UVIS_091IC_ALPVIR001_PRIME
• 2008_265 had enough data to continue
the previous analysis of tracking sensitivity
degradation and stellar variability
Illustration of data dropouts
• FUV2008_265_23_40_33_UVIS_086IC_ALPVIR001_PRIME
• FUV2008_309_21_10_33_UVIS_091IC_ALPVIR001_PRIME
Spica variability
Background on Alpha Vir (Spica)
•
Spica is a non-eclipsing double-lined spectroscopic binary system
–
–
–
Though not spatially resolvable, each component is detectable through
measurements of out-of-phase Doppler shifts in the constituent spectral lines
Non-eclipsing due to large apparent orbital inclination of ~70 degrees
Both stars are of a similar spectral class:
•
•
•
Spica is the brightest rotating ellipsoidal variable star
–
–
–
–
–
•
Primary: B1V
Secondary: B4V
The stars have a distorted ellipsoidal shape due to mutual gravitation effects
As the components revolve, the visible area (and thus the observed flux)
changes with orbital phase
Since this is a geometric effect, it should be roughly wavelength-independent
Orbital period is 4.01454 days
Amplitude of flux variation in V-filter ~3%
http://observatory.sfasu.edu
The primary of Spica is a Cepheid variable
–
–
–
–
Periodic variation in the pulsating primary star is much shorter than the system’s
orbital period and about a factor of 2 less in magnitude
Period is 4.17 hours
Amplitude of flux variation in V-filter ~1.5%
This short-term variation, identified in 1968, became undetectable in the early
1970’s (but may return again due to precession of the primary’s rotation axis
relative to the orbital plane, which has a period of 200 years [Balona, 1986])
Ellipsoidal variation model
Variation in flux is given by [Shobbrook, 1969; Sterken et al, 1986]:
dE = A M2/M1 (R/D)3 (1+e cos(TA+Φ))3 (1-3cos2(TA+TA0+Φ) sin2i )
Where:
A=0.822 (wavelength dependent “photometric distortion”)
M2/M1 = 1/1.59 (ratio of masses)
R = 7.6 Rsun = 5.2858e6 km (polar radius of primary)
D = 1.92916e7 km (mean separation between stars)
e = 0.14 (orbital eccentricity)
TA (true anomaly)
T0 = 4.01454 days (orbital period)
TA0 = 150 degrees (apparent angle to line of apsides in
year 2005, has precession period of 128 years)
i = 65.9 degrees (orbital inclination)
Φ = empirical phase shift, a free parameter to match with data
One period of the expected variation in
flux from Spica
Normalized signal vs time
• The left plot shows the total FUV signal vs time (normalized to the
mean), with a line fit
• The right plot shows the same data with this linear trend removed,
along with a theoretical model of the Spica ellipsoidal variation that
has been fit to the curve (optimizing only the magnitude and phase
offset parameters)
Stellar flux comparisons
Stellar flux comparisons
• We can now compare many of the stellar
irradiance measurements from UVIS with those
made by SOLSTICE
• Background on the SOLSTICE instrument:
– An FUV/MUV spectrometer onboard the SORCE
spacecraft
– Built at LASP
– Absolutely calibrated at the NIST-SURF facility
– Routinely measures stellar fluxes in order to track
sensitivity degradation
– Irradiance spectra reduced and calibrated by Marty
Snow at LASP
name
canopus
type
F0
UVIS
x
SOLSTICE
alp cma
alp cru
sirius
A1
B1
x
x
x
x
alp gru
alp eri
alnair
achernar
B6
B3
x
alp leo
alp lyr
alp pav
regulus
vega
peacock
B7
A0
B2
x
x
x
alp peg
Markab
B9
alp psa
alp vir
fomalhaut A4
spica
B1
alp car
bet cen
bet cma
bet cru
bet ori
del cen
del cyg
mimosa
rigel
del sco
x
x
x
x
B1
B1
B0
B8
B2
B9
x
x
B0
x
B2
B0
B0
x
x
x
x
adara
alnilam
eta uma
alcaid
B3
x
gam ori
kap ori
kap vel
lam sco
pi sco
sig sgr
tau sco
zet cen
zet oph
zet ori
zet pup
bellatrix
saiph
B2
B0
B2
B2
B1
B2
B0
B2
O9
O9
O5
x
x
Nunki
alnitak
naos
x
x
x
x
eps cma
eps ori
eps per
shaula
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
List of stars
observed by UVIS
and SOLSTICE
• Stars must be very
hot in order to
produce significant
flux in the FUV
• Therefore, most
stars observed by
UVIS are of B and
A spectral class
• Nine stars were
observed by both
UVIS and
SOLSTICE
UVIS spectra
• This shows the
spectral
irradiance
curves as
measured by
UVIS
• A variety of
shapes (driven
by spectral
class) and
magnitudes
are seen
Star parameters
• We can use a catalog of stellar parameters to estimate
the spectral irradiance from any star using the Kurucz
model
alp
alp
alp
alp
alp
alp
alp
del
eta
zet
name
vir
psa
cma
cru
leo
lyr
pav
sco
uma
cen
Class
B1
A3
A0
B0
B7
A0
B2
B0
B3
B2
Vmag
0.98
1.17
-1.44
0.77
1.36
0.03
1.94
NaN
1.85
NaN
Temp
22387
8318
9550
24547
10965
9333
12023
NaN
11220
NaN
Grav
3.67
4.19
4.27
3.57
3.77
3.98
3.48
NaN
3.78
NaN
Radius
7.94
1.86
1.78
10.00
3.89
2.69
6.31
NaN
3.89
NaN
Plx
12.44
130.08
379.21
10.17
42.09
128.93
17.80
NaN
32.39
NaN
Temp is the effective temperature in Kelvin
Radius is in units of solar radius
Plx is the parallax in arcseconds as measured by Hipparcos
Metallicity assumed to be zero (solar)
Stellar data from: Allende Prieto C.,
Lambert D.L., Astron. Astrophys. 352,
555 (1999), data accessed through
VizieR
Kurucz Stellar Models
• The Kurucz model stellar irradiance at any wavelength is
computed in a coarse grid of three variables:
temperature, gravity, and metallicity
• A convenient routine for accessing these models is
available from:
– IUEDAC IDL Software Libraries
– http://archive.stsci.edu/iue/iuedac.html
• The routine KURGET1 allows a user to specify a
measured temperature, gravity, metallicity and angular
size to arrive at a model spectral irradiance from any
star:
–
–
–
–
–
Wavelength range: 9.1 nm to 106 μm
Total wavelength steps: 1221
Resolution: ~1nm in the EUV, FUV, MUV
Default units: [ergs/cm2/s/Angstrom]
Allows linear interpolation between most model grid points
UVIS, SOLSTICE, KURUCZ
• Three
Kurucz
models
show good
consistency,
two do not
Ratio of UVIS to SOLSTICE
• This plot shows
the ratio of UVIS
spectra (1.1nm
bins) to
SOLSTICE over
130-180 nm
• Here, UVIS
spectra were
reduced using
the flat-field
• While the
magnitude
varies, a
consistent shape
in the ratio is
apparent
WITH
flat-field
Q - Is the discrepancy dependent on spectral
type, signal level, or position on the detector?
Ratio of UVIS to SOLSTICE
• Normalized
to a mean
value of one
Ratio of UVIS to SOLSTICE
• This plot shows
the average ratio
of UVIS to
SOLSTICE for
all stars
observed
(except Alp PsA)
Average of all stars
Ratio of UVIS to SOLSTICE
• This plot shows
the average ratio
of UVIS to
SOLSTICE for
all stars
observed
(except Alp PsA)
• Expanded
vertical scale
Wavelength shift
• Due to the small uncertainty in
spacecraft pointing, a star image
could be centered anywhere
within six spectral pixels using
the low-resolution slit
• This is effectively an uncertainty
in the wavelength scale of the
spectrum
• With an FUV dispersion of 0.078
nm per pixel, this translates to a
potential variation of ~0.5 nm
Star
image
UVIS detector pixels are 100 x 25 microns (H x W)
UVIS-FUV Lowresolution entrance slit
Low-res slit is 6 pixels wide
Wavelength shift adjustment
• The phase shift between the measured UVIS spectrum
and SOLSTICE can be estimated by finding the peak of
the cross-correlation function
• This is done by the following procedure:
– Shift the UVIS spectrum in one-pixel increments
– Smooth the UVIS spectrum to 1 nm
– Interpolate the UVIS spectrum to the SOLSTICE wavelength
scale
– compute the linear correlation coefficient
– Find the shift value where the maximum correlation occurs
• This is a small effect for 1 nm resolution spectra
Cross-correlation analysis
• This shows the
correlation
coefficient
versus UVIS
pixel shift
• A peak in the
correlation
coefficient is
considered a
wavelength
match
• All matches
occur within four
pixels
Average UVIS/SOLSTICE ratio
• No wavelength shift
• With individual
wavelength shifts
Ratio of UVIS to SOLSTICE vs
SOLSTICE irradiance
•
•
•
UVIS appears to
measure fluxes that are
higher at lower signal
levels
This could be the result
of a need to remove an
offset, which would
preferentially affect
measurements of lower
signal
There is currently no
attempt to adjust for
nonlinearity due to the
detector dead-time; an
adjustment would have
the effect of increasing
the UVIS signal of
brighter stars
Ratio of UVIS to SOLSTICE vs
surface temperature
• The dimmer
stars also
happen to
be the
cooler ones
Effective temperatures from: Allende Prieto C., Lambert D.L., Astron. Astrophys. 352,
555 (1999), data accessed through VizieR
Stars are imaged in the star-burned rows
•
Flat-field
corrector
•
Spatial
distribution
of stars
•
This plot shows the
spatial distribution of
light on the detector
from each star, along
with the inverse of
the average row-torow correction from
the post-burn flatfield
The stars were
imaged in the center
of the starburned
region of the
detector
It is here that the
sensitivity correction
is largest, and the
spectral shape of the
correction the most
significant
Spatial distribution of stellar flux
measurements
No Flat-field
applied
WITH Flat-field
applied
• These plots show the distribution of light on the FUV
detector for each observation of stellar irradiance
• The curves are normalized to the total signal
• The images are centered at about row 32
Star comparisons – TO DO
• Compare spectra from measurements where the star
image is located outside of the starburned rows (Alp Vir,
Alp Leo, Alp PsA, others?)
• Look into getting extended SOLSTICE spectra (>180nm)
from Marty
• Check other catalogs to see if the effective temperatures
are consistent
• How are the stellar parameters of temperature, gravity,
and metallicity derived? How accurately are they
known?
• Include flux from secondary companion stars
• Compare the flux from UVIS using Don’s FUV sensitivity
curve
Flat-fielding for small targets
Use of the flat-field corrector for
small sources
• The flat-field is meant to correct the highfrequency pixel-to-pixel sensitivity nonuniformity
• It is derived from the along-slit slew scans of
Spica, which approximates a uniform extended
source
• However, it has been noticed that there could be
an issue with using the corrector for small
(unresolved) targets
Effect of the flat-field on an
extended source
•
•
This approximates the spatial
distribution of a uniform
extended source
Sum of all scans, sum in the
spectral dimension
•
•
•
This demonstrates the effect of
applying a flat-field to a uniform
extended source
Sum of all scans, flat field applied,
and sum in the spectral dimension
Overcorrection in the starburned rows
Effect of the flat-field on a point
source
•
•
•
Total signal vs star position
Evil pixels interpolated across
No flat-field applied
• Total signal vs star position
• Flat-field applied to each
frame
Sensitivity at Lyman-alpha
IPH data reduction
• Identify all UVIS observations which are:
–
–
–
–
–
FUV
Low-resolution
Unbinned
Unwindowed
IPHSURVEY
• Average all records in each observation
• Mark all evil pixels as NaNs
• Average all remaining ‘good’ pixels in each
column (row 3 to 60) to form a single spectrum
Filtering IPH data
• All selected LISM
spectra in the
DAPS database
• All LISM spectra which
meet a filtering criteria to
exclude contamination by
stars and data dropouts
Filtering IPH data
• Select all
observations
prior to the year
2000
• Select all
observations
after the year
2000 with an
average
(between pixel
300 and 1023)
signal in the
range
0.0005±0.00001
Normalizing to a fixed LISM signal
• All filtered spectra, with the
offset subtracted (mean of
signal between pixels 990 –
1010)
• Need to remove LISM signal
variation due to distance from
the Sun and pointing.
• Normalize each spectrum to
the signal between pixel
ranges 100-110 and 150-160
Signal normalized spectra
• Spectra are
normalized to
the signal in
the shaded
regions
Ratio to an early spectrum
• This shows
the ratio of all
spectra to an
early
spectrum
• This should
show the
fractional
change in
sensitivity
Ratio to an early spectrum
• Look at the
change as a
function of
time for a few
pixels
Pixel value vs time
•
•
•
•
This shows the count
rate vs time for three
spectral columns
Each curve is
normalized to the
mean of the values
before 2000
The variation is much
larger than the random
uncertainty in the
detected counts
Ideas:
–
Colors here denote different pixels
–
Due to a small
mislocation in the
position of the slit?
This is a known
effect.
Incomplete removal
of stellar spectra?
Poor background subtraction?
•
Raw spectra
(counts/second)
•
Seems unlikely that some pixels
decline more than others. More
likely that the background
subtraction was poor.
•
After subtraction
of an estimate of
the background
(mean of pixels
990-1010)
Summary
• There is a small, approximately linear decline in FUV
sensitivity
• The ellipsoidal variation of Spica continues
• Comparisons of UVIS measurements of absolute stellar
fluxes with SOLSTICE show that there is a variation in
absolute magnitude on the order of 30%, and a
systematic shape difference
• The flat-field should be used with caution for small
targets
• There is a measurable decrease in sensitivity at Lyman
alpha as a function of time, but the shape is difficult to
determine