|
Provisional Flats & Darks, and ground comparisons | |
o Below we compare the latest darks and flats derived from flight data versus
those from MIC2 testing. Performance using comparisons of coadds made from
re-calibrated frames and statistics are also given. The MIC2 test products
are from the second test (Mar/Apr 2009) and use the engineering model
electronics.
o All flight flats appear to be remarkably stable. For early mission analyses,
see: Flight Responsivity Maps
o The W1,W2 darks appear remarkably stable too. However some time-variability
is seen in the W3,W4 darks when these are computed from different datasets.
It is not known yet whether this is tied to the pedigree of the input
data (e.g., astrophysical background variations), or is really instrumental
in nature. In the end, we have a good set of working darks. Further
residuals are expected to be removed by dynamic calibration using
"sky-offset" products. These products are created from a running block
median along every scan, re-baselining to zero, and then subtracting these
from the respective input frames. This functionality has not yet been
validated.
o In a nutshell, the w3,w4 darks were derived from flight data using the
following method. A simple detector observation model for pixel i can be
written:
O_i ~ G_i*S_i + D_i, where
G_i = the gain (= relative flat here),
S_i = true sky,
D_i = dark+bias.
A proxy for the sky S_i is taken to be the median raw-frame (L0) signal
over all array pixels minus some (unknown) absolute dark level DL, i.e.,
S_i ~ L0 - DL.
The observation model then becomes:
O_i ~ G_i*L0 + C_i, where
C_i = D_i - (G_i*DL).
The flat (slope) G_i and intercept C_i are estimated from fitting to L0 data
that follows the change in zodiacal background. The dark signal per pixel
is then: D_i ~ C_i + (G_i*DL). For now, we assumed values of the absolute
dark level DL from MIC2 testing. One can do a better job and determine DL
using a global minimization fit. This is planned for later. Regardless of
the absolute dark level, what's important here is to capture the dark
variation over each array, e.g., the banding structure. Note that this
model ignores the linearity correction. This will be folded in later.
For now, the values of D_i from this method vastly improve the instrumental
calibration for W3 and W4.
o Actual FITS files of products are in:
phoebe:/Users/fmasci/01.20.10_provisional/
 |
 |
 |
 |
 |
 |
 |
 |
 |
 |
The images below are W3 coadds made from L1 frames processed with ground calibrations [left] versus those processed using the latest flight calibrations [right] - mainly darks, flats, and a correction for droop. The galaxy at the center is NGC1097. The purpose here is to show the reduced instrumental residuals when the new calibrations are used. Further below is the W4 coadd comparison. Each coadd is of the same dimension of a typical Atlas Image (~1.56 degrees on a side).
For additional testing of new vs. old flats on individual frames, see the animations at the end of: Flight Responsivity Maps.
 |
 |
The most useful comparisons here are the "MEDIAN 1-SIGMA UNC" values (data derived uncertainties, not from a model). To get a feel for the achieved accuracy, the relative error is given by: 100 * (MEDIAN 1-SIGMA UNC/MEDIAN)%. DARKS: FLIGHT: w1: TRIMMED MEAN = 126.790045764026 MEDIAN = 126.854270935059 TRIMMED STDDEV = 0.495638944760485 MEDIAN 1-SIGMA UNC = 0.0182627849280834 w2: TRIMMED MEAN = 126.831773534848 MEDIAN = 126.888732910156 TRIMMED STDDEV = 0.464302403097224 MEDIAN 1-SIGMA UNC = 0.0139244571328163 w3: TRIMMED MEAN = 267.210005242453 MEDIAN = 267.955871582031 TRIMMED STDDEV = 12.2105976976511 MEDIAN 1-SIGMA UNC = 2.18255567550659 w4: TRIMMED MEAN = 283.375599575752 MEDIAN = 282.940551757812 TRIMMED STDDEV = 5.94605463475657 MEDIAN 1-SIGMA UNC = 1.08535397052765 -- MIC2: w1 TRIMMED MEAN = 126.731373656845 MEDIAN = 126.79020690918 TRIMMED STDDEV = 0.482345103471963 MEDIAN 1-SIGMA UNC = 0.0847787484526634 w2: TRIMMED MEAN = 126.822081763272 MEDIAN = 126.875679016113 TRIMMED STDDEV = 0.462724204947326 MEDIAN 1-SIGMA UNC = 0.0796581357717514 w3: TRIMMED MEAN = 265.42783784427 MEDIAN = 265.169158935547 TRIMMED STDDEV = 4.85252146790588 MEDIAN 1-SIGMA UNC = 0.57173627614975 w4: TRIMMED MEAN = 282.206094879428 MEDIAN = 281.670013427734 TRIMMED STDDEV = 5.84730234722142 MEDIAN 1-SIGMA UNC = 0.306508123874664 FLATS: FLIGHT: w1: TRIMMED MEAN = 0.990915229715353 MEDIAN = 1.00114178657532 TRIMMED STDDEV = 0.0450786573033271 MEDIAN 1-SIGMA UNC = 0.00547813484445214 w2: TRIMMED MEAN = 0.98984747479498 MEDIAN = 0.999574363231659 TRIMMED STDDEV = 0.0729365545581228 MEDIAN 1-SIGMA UNC = 0.00246186321601272 w3: TRIMMED MEAN = 0.999937674691483 MEDIAN = 1.0005464553833 TRIMMED STDDEV = 0.0157995175459051 MEDIAN 1-SIGMA UNC = 0.00106819916982204 w4: TRIMMED MEAN = 0.999982556490866 MEDIAN = 1.0022965669632 TRIMMED STDDEV = 0.0174808210201523 MEDIAN 1-SIGMA UNC = 0.00114717206452042 -- MIC2: w1: TRIMMED MEAN = 0.994243404738707 MEDIAN = 0.999867618083954 TRIMMED STDDEV = 0.0296867655476971 MEDIAN 1-SIGMA UNC = 0.000689558743033558 w2: TRIMMED MEAN = 0.993606854545473 MEDIAN = 0.99871826171875 TRIMMED STDDEV = 0.0518478689763131 MEDIAN 1-SIGMA UNC = 0.000801462447270751 w3: TRIMMED MEAN = 1.00144394199958 MEDIAN = 1.00002634525299 TRIMMED STDDEV = 0.0135510509124029 MEDIAN 1-SIGMA UNC = 0.000152157081174664 w4: TRIMMED MEAN = 0.997730377076971 MEDIAN = 1.0010598897934 TRIMMED STDDEV = 0.0160948322558013 MEDIAN 1-SIGMA UNC = 0.0000961642144829966