adjustmap
ldss3pattern
ldss3pattern removes pattern noise from LDSS3 frames.
USAGE  ldss3pattern framename
 
INPUT 
 
OUTPUT 
 
PARAMETERS 

Details:
The program ldss3pattern estimates the contribution of the pattern noise at every location using a template, and then subtracts this contribution from an LDSS3 data frame. The Fourier Transform of an empty subsection of the data is taken and its power spectrum is computed. This is used as a template power spectrum for computing the contribution of the pattern noise over the entire data section of the array. Then the Fourier Transform of the full data section of the array is taken and its power spectrum is computed. Because the template power spectrum is computed from a smaller array, the template power spectrum is resampled to produce an array with a length equal to the power spectrum computed for the full data section of the frame. This rescaling of the size of the template power spectrum uses a simple interpolation routine because the rescaling in size is by a noninteger amount. Then a simple sigmaclipping routine is performed to identify regions of the data power spectrum that have significantly greater power than in the template. This excess power is due to the presence of real information in the data array on top of the pattern noise. By identifying these regions of the power spectrum that arise from real signal, we go back to the original sine and cosine components of the Fourier Transform of the data section and filter out those components identified by the power spectrum analysis. This should leave you with the pattern noise in Fourier space. A inverse Fourier Transform is performed to produce the pattern noise back in real space.
The above task is performed for the output of both amplifiers. Because the pattern noise is identical in both FITS files, we use a weighted average of the two results as the final estimate of the pattern noise. A weighted average is computed for each pixel for all chips of the pattern noise, each weight being inverse of the pixel value squared. This downweights larger values, such as pixels containing real signal or other artifacts such as cosmicrays, and these are effectively excluded from the resulting estimate of the pattern noise. From this a template of the pattern noise is created and then can be subtracted from the data.