LASCO C3 Level 1 Processing

Action Item list from July 2002 meeting at NRL

(1) Check impact of "3m_clcl_all.fts" during MISSING BLOCKS procedures.

 None: img[zz] = newimg[zz]

• img is the returned image
• zz represents the missing blocks (only missing blocks are changed)

(2) Convert "C3[2]nullblock.sav" to ASCII.

• c2nullblocks.txt
• c3nullblocks.txt

(3) Check dynamic range of processed images to see if STORAGE REQUIREMENTS can be reduced.

• This introduces a rounding error.
• To implement, scale image between 2.0e-13 and 6.5e-9 in reduce_level_1.pro to maximize dynamic range

BSCALE  =          9.91806E-14 / Data value = FITS value x BSCALE + BZERO
BZERO   =          3.24995E-09 /

• Currently, image statistics in header (DATA[MIN,MAX,AVG,P01-99]) are from unscaled image:

DATAMIN =          2.05145E-12 / Minimum Value Not Equal to Zero
DATAMAX =          9.56944E-09 / Maximum Value NE Saturated
DATAZER =               183539 / Number of Zero Pixels
DATASAT =                   91 / Number of Saturated Pixels
DSATVAL =          6.50000E-09 / Value used as saturated.
DATAAVG =        7.4424799E-11 / Mean of Image
DATASIG =        2.2080698E-10 / Standard Deviation of Image
DATAP01 =        8.5300000E-12 / Percentile Value
DATAP10 =        1.2640000E-11 / Percentile Value
DATAP25 =        1.6150000E-11 / Percentile Value
DATAP50 =        2.5180000E-11 / Percentile Value
DATAP75 =        5.1540000E-11 / Percentile Value
DATAP90 =        1.3580000E-10 / Percentile Value
DATAP95 =        2.6102000E-10 / Percentile Value
DATAP98 =        5.5190000E-10 / Percentile Value
DATAP99 =        9.3757000E-10 / Percentile Value

Results from 10 images (1024x1024) from 1-10 Dec. 1996  (darr = unscaled - scaled) (units DN/sec):

• Example: part of a row illustrating effects of BSCALE
• IIlustration of effect of BSCALE on the mean of images
• Histogram of difference between images processed with and without BSCALE (1024x1024x10 pixels)
• Tail of histogram of difference between images processed with and without BSCALE for 10 images; 893 pixels are not shown, but scattered beyond the right axis
• Number of saturated pixels
• GIF movie showing 10 darr images:
• scale is +-0.09 DN/sec.
• rings appear in X# (Hcompress?) images

Results from series of one image every 20 days (43 total):

• DATAMIN from one image every 20 days
• Illustration of effect of BSCALE on mean
• Number of saturated pixels

(4) Look at GEOMETRIC DISTORTION CORRECTION, specifically the rings in the warped image. (w/ JEFF MORRILL and MIKE ANDREWS) [CHAIR]

IDL[calliope]>rndm=randomn(5,1024,1024)
; RANDOMN uses the Box-Muller method for generating normally-distributed (Gaussian) random numbers,
; here with a seed of 5.
IDL[calliope]>maxmin,rndm
      4.86010  0.000247973     -5.02657

IDL[calliope]>for i=31,1000,32 do rndm[*,i]=2
IDL[calliope]>for i=31,1000,32 do rndm[i,*]=2
IDL[calliope]>tvscl,rndm
IDL[calliope]>wrndm2=c3_warp(rndm,h0)
IDL[calliope]>tvscl,wrndm2

• color table for these images

Ring in warped image (such as here):

IDL[calliope]>rdif=wrndm2-rndm

; ** to find circle which corresponds to zero in C3 distortion correction
IDL[calliope]>cf = DISTORTION_COEFFS ('C3')
IDL[calliope]>r=sqrt(-1*cf[0]/cf[1])
IDL[calliope]>print,r/.021
      455.903
; Apply circle to rdif using TVCIRCLE
; **

Pattern in warped image

• pattern apparently not affected by changing gridsize or /quintic in WARP_TRI.pro
• I estimate magnitude of pattern to be about 10% (of noise?)
• ( warped image - bkg ) <3>(-2) DN/sec   ; background is min of 27 daily medians
• closeup of pattern in gridded warped image (wrndm2)

Bottom line: pattern is IN THE NOISE, and much less significant than other errors

(4.5) Where is distortion function from?

TBD
 

(5) Develop procedure to map MISSING BLOCKS from header and apply to image

•  Done. See ADD_MISSING_BLOCKS.pro