Notes about the Far Infrared Line Mapper (FILM) Data in IRSA IRTS database (S. Lord, May 2004)*
-----------------------------------------------------------------

WHICH CHANNELS ARE PRESENT IN THE FILM DATA ?

In its design, the IRTS FILM instrument has 4 channels, one channel devoted to the 63um (158.4 cm-1) [OI] line  and the other three associated with the 157.7 um (63.4 cm-1) [CII] line. The three channels associated with [CII] consist of a central channel of higher resolution,  (R=409), and two side channels to measure baseline continuum at  155.27 um (R=130) and 160.2 um (R=130).

Data from only two of these four channels are present in the IRSA/IPAC file set: 1) the  "line data "  from the  [CII] 157.7 um detector and 2) the continuum background data from the  155um detector.

WHAT IS THE SPECTRAL RESOLUTION, THE SPATIAL RESOLUTION, AND  SAMPLING?

The spectral resolution of the central CII channel is R=409 which spans 733 km/s and thus the full range of Galactic disk velocities.
 
The spatial resolution of a single pixel  is 8' x 13', where 13' the resolution in the "in-scan" coordinate direction. The sampling of pixels is Nyquist - pixels are displaced approximately every 4' in the in-scan and cross-scan directions. Over several orbits a multi-strip map was built up. These strips are seen in each observing box. [Note: the term "observing box" is not the official nomenclature of the IRTS mission but a term invented for these pages.] The observing boxes are represented by the 88 red squares in the IRSA IRTS image main window and are 12.8 degrees on a side.

We picture an observation below with detector tracks going across it. The data are resampled into the closest 4'x4' pixel within the 192 pixel x 192 pixel observing box. The box is oriented in galactic (long.., lat.) which is shown as (x,y) on the axes below.  The RA and Dec direction is given by the N,E arrows.  The box width is 12.8 degrees on a side. There are about 88 such independent boxes. The boxes overlap. If you click on a position using the IRSA Atlas service that is in the overlap region, the files for both boxes will be listed.



DATA UNITS

We draw the following conclusions by looking at the comments for the keyword BUNIT in the headers of the intensity maps' FITS files:
The CII Channel is in units of delta_lambda (of the detector) x Intensity_lambda yielding W/m^2/sr
The 155 um Channel is in units of lambda (i.e. 155 um) x Intensity_lambda again yielding W/m^2/sr
Thus to convert the 155 um channel to a flux density one would multiple 8'x13' in sr (8.8 10^-4) and divide by 155 which would yield W/m^2/um

HOW ARE THE DATA ARRANGED?
There are typically  nine  FILM files associated with a particular observing box.

(This is true for all observing boxes except three: the square at (84deg RA, 27deg Dec) has ten 155 um files and four CII files. The square at (286deg RA, 17deg Dec) that has only five 155 um files, and it lies next to a square at (295 deg RA, 1.4deg Dec) that has only four CII files.)

When you select a position, these nine files will be listed, available for display or downloading. Four files will be associated with the CII detector, and five will be associated with the 155um continuum detector. 

It convenient to distinguish these files from each other in two steps:
                                                                   
First, the CII 158 um line data are found in files with a directory path such as: MAPS/FILM/CII/                                    
The 155 um files are found in the path such as MAPS/FILM/155UM/film_155_fits/ or  MAPS/FILM/155UM/film_155_intp/                       
When downloading files, you may wish to separate these files into two directories or rename them, because you will not be able to distinguish
the CII and 155 um data by the filenames.

The second step is to look at the keywords of the FITS header within the file to see which type it is. We list the nine different types of FILM files below.

The files are FITS images, such as the image above, which can be viewed with utilities such as OASIS, ds9, etc..
                   
We list the four CII files in order and give the important identifying keywords:

CII File 1) Samples per pixel Map

BUNIT   = '         '          / # of samples in a pixel
COMMENT   The IRTS/FILM [CII]-line channel sample number in a pixel

This is the number of samples taken that fell into a particular pixel that were used to calculate the pixel value (not interpolated by surrounding pixels). The number typically ranges between 1-4, with 39 indicating no data in a pixel, and the count reaching as high as 20 where the orbit scan tracks are crowed together.

CII File 2) Raw Intensity Map

BSCALE  =       01E-13 / TRUE=TAPE*BSCALE+BZERO
BZERO   =       00E+00 / COMMENT The IRTS/FILM [CII]-line intensity map
BUNIT   = 'W/m**2/sr'  / Intensity (DELTA-LAMBDA*I-LAMBDA) for line

These line fluxes are a fundamental product of the mission. I believe the CII detector data here has had the scaled 155 um continuum removed since the intensities take negative excursions, but I am by no means sure.  This is a significant point since at 158 um the Line/Continuum can be on the order of unity. Therefore, this possibility should be checked.  Further, given the the resolution is quoted as 8'x13' I believe the conversion to W/m**2 should involve a multiplication by:  8.8 10^-4.

CII File 3) Interpolated Intensity Map
BSCALE  =               01E-13 / TRUE=TAPE*BSCALE+BZERO
BZERO   =               00E+00 /
BUNIT   = 'W/m**2/sr'          / Intensity (DELTA-LAMBDA*I-LAMBDA) for line

As shown below, the line intensities have been interpolated and extrapolated to fill in the gaps between scans. Presumably the algorithm used is that one specified in the fifth 155 Fits files discussed below. We do note that in the way that the gaps have been are filled-in, hot adjacent pixels can dominate and systematic high intensity structure can sometimes be seen to result.


CII File 4) Dispersion Error Map

BSCALE  =               01E-13 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =               00E+00 /                                               
BUNIT   = 'W/m**2/sr'          / Error for line   

This error value seems to be a 1 sigma dispersion between the values combined: it is zero when only one data point was used to form the pixel. The error map applies to the intensity map (File 2) not the interpolated intensity map (File 3).

We note that the FITS header variable CDELT2 has a sign inversion between the CII and the 155 um which causes the maps appear flipped in the observing box when compared with each other.

155 um File 1) Samples per pixel MAP
BSCALE  =            1.000E+00 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =            0.000E+00 /                                               
BUNIT   = 'NUMBER   '          / The Number of data in pixel 

This is again the number of 155 um samples falling into a pixel. Here there are more than in the CII case, and never just 1. The number ranges into the 40 and the value 86 represents no data.

155 um File 2) Raw Intensity Map
BSCALE  =      1.000000000E-11 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =      0.000000000E+00 /                                               
BUNIT   = 'W/m**2/sr'          / Intensity (LAMBDA*I_LAMBDA)  

155 um File 4) Interpolated Intensity Map
BSCALE  =      1.000000000E-11 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =      0.000000000E+00 /                                               
BUNIT   = 'W/m**2/sr'          / Intensity (LAMBDA*I_LAMBDA). This image is    
                               /  including interpolated pixel values.  
155 um File 3) Dispersion Error Map    
Notice the BSCALE value is lower for the error map- the FIR 155 um continuum is detected with
significantly higher S/N than the CII line emission.                   
BSCALE  =      1.000000000E-12 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =      0.000000000E+00 /                                               
BUNIT   = 'W/m**2/sr'          / Error of Intensity (LAMBDA*I_LAMBDA)    

155 um File 5) Interpolation Method Map
The larger pixel boxes below are used when there are larger gaps between scans.

BSCALE  =            1.000E+00 / TRUE=TAPE*BSCALE+BZERO                        
BZERO   =            0.000E+00 /                                               
BUNIT   = 'INDEX NO'           / The number of the interpolation index
DATAMIN =                     0 / # of the interpolation index                  
DATAMAX =                    3 /  0 : Original Intensity                       
                               /  1 : The value is linearly interpolated from  
                               /       the nearest 3x3 pixel box               
                               /  2 : The value is non-linearly interpolated   
                               /       from the nearest 5x5 pixel circle       
                               /  3 : The value is non-linearly interpolated   
                               /       from the nearest 7x7 pixel circle     

CALIBRATION

An initial assessment of LWS CII calibration was made by Mayiuti et al. (1997) who site a calibration uncertainty of 40% for the CII fluxes. They find IRTS measurements
are 0.7 and 0.5 as much as those of a balloon mission BICE and COBE/FIRAS


* These note have been prepared by reading the papers below and looking at the data products.
The author is responsible for all errors.

The best source of definitive pre-launch information about FILM (the Far Infrared Line Mapper)
aboard the IRTS satellite is found here:
"Far-Infrared Line Mapper (FILM) on the Infrared Telescope in Space" Shibai, H. et al. 1994, ApJ 428, 377

For an initial look at the post-mission CII data, and for calibration against other observatories the following reference is useful:
"[CII] line observations of the Galactic Plane by the IRTS/FILM," Makiuti, S. et al. 1997, in "Diffuse Infrared Radiation and the IRTS, ASP Conference Series, Vol 124. H. Okuda, T. Matsumoto, and T. L. Roellig, eds..