Cosmic Origins Spectrograph Instrument Handbook for Cycle 17

12.1 COS Data Products

---

12.1.1 FUV TIME-TAG data

Raw data

COS FUV TIME-TAG raw data consists of detected events in sequential order, based on the time the event was detected. Each event includes:

• Dispersion location (x value),
• Cross-dispersion location (y value),
• Time of the event, and
• Pulse-height value.

The raw data include source counts, sky background, detector background, and stim pulses. The data from the two detector segments are interleaved in the flight electronics, but they are later separated in the ground software.

Corrected TIME-TAG data

The corrected TIME-TAG data consist of seven quantities:

• x and y, the location of the event in detector coordinates, corrected for thermal distortion (characterized by the stim pulses) and geometric distortion (determined during ground testing). These new x and y values are now floating point instead of integer.
• x_d, the x-position of the event in non-integral pixels corrected for orbital and heliocentric doppler effects. This allows for the spectrum to be in either detector coordinates (x, y) or wavelength space (x_d, y). The examination of both images allows for detector features, such as a hot spot, to be evident, or for a wavelength-dependent feature to be sharp.
• t, the time for each event. As noted earlier, the time is recorded to the nearest 32 msec. However, the sequence of photon events in the TIME-TAG list is the true order in which they occurred, although it is possible for the ordering between segments to not be maintained because of the interleaving of events.
• h_p is the pulse height for an event. The value of h_p is the charge extracted from the micro-channel plate for that photon, or, alternatively, the electron gain for that event. The pulse height can be used later as a filter to select the most significant events as a way of reducing noise.
• is the sensitivity or weighting term for a photon event. It combines pixel-to-pixel response variations and the dead-time correction.
• q represents the quality factor for a given event, based on its location. The value of q is determined from a list of detector blemishes and the like. For example, an event within a bounding box around a hot spot will be assigned a value of q to indicate that it is probably from that hot spot. The q values are assigned during ground processing from a reference file.

Corrected image

The corrected image is, like the raw data, 16384 × 1024 in size, and consists of an effective counts per second which is the sum of all the factors associated with a given pixel, divided by the exposure duration. This corrected image is formed from the corrected TIME-TAG data using x_d and y values and standard threshold values for the pulse heights. Thus this corrected image is in wavelength space and corrected for doppler smearing. The width of each pixel is uniform in physical space, i.e., in microns. This image is not corrected for background counts.

Error array

The error array is also 16384 × 1024 in size and includes the errors in the corrected image for each pixel. These errors are calculated using Poisson statistics from the gross counts, and are corrected for flat field and dead-time. The units are the same as for the corrected image, namely effective counts sec^-1.

Science spectrum

The extracted science spectrum consists of:

• , the central wavelength for a pixel,
• C_G, the gross count rate,
• C_B, the background count rate,
• C_N, the net count rate,
• F, the flux,
• , the error in F,
• MDQ, the maximum data quality in a resolution element, and
• ADQ, the average data quality.

These quantities will be explained in detail in the Data Handbook. The count rates are computed for bins of equal physical width. The background rate is determined from regions of the detector immediately above and below the science spectrum and is averaged over a larger region than just one pixel to improve the statistics since the rate is very low. This background is measured away from any sky and so does not include that.

The net count rate is converted to flux using calibration reference files. The final error is in flux units and includes all the known sources of error. The MDQ is the maximum of all the quality flags for the individual events, while the ADQ is the average of those.

If multiple spectra have been obtained using FP-POS, the individual spectra are weighted by their relative exposure duration and then merged into a single file.

12.1.2 NUV TIME-TAG data

Raw NUV TIMED-TAG data

For the NUV, no pulse heights are recorded, and so the raw data consist of x, y, and t.

Corrected TIME-TAG list

For the NUV, the corrected list of TIME-TAG events consists of x, y, x_d, t, , and q.

Corrected NUV TIME-TAG image

A corrected 1024 × 1024 image is formed from the corrected list of TIME-TAG events. The value in each pixel is effective counts sec^-1, which is the sum of all counts in a pixel multiplied by its factor and divided by the exposure duration.

Error array

The error array is also 1024 × 1024 and is based on Poisson statistics from the gross counts, correcting for flat field effects and the dead time. The units are effective counts sec^-1.

Science spectrum

The science spectrum is a table with the same quantities as for FUV TIME-TAG data.

12.1.3 FUV ACCUM data

Raw FUV ACCUM data

The raw data is an array of dimensions 16384 × 1024 pixels. Each pixel is 16 bits and so can handle up to 65,535 counts. Note that the actual image size sent from the instrument is much smaller than 1024 in height to minimize data quantities, but the full image size is maintained in the ground processing to allow for future movements of the spectrum on the detector.

Corrected FUV ACCUM data

The corrected image is also 16384 × 1024 pixels and is corrected for doppler motion of the spacecraft (done on-board), flat-field response, and detector dead time. The value in each pixel is effective counts sec^-1, which is the total counts in a pixel multiplied by that pixel's factor and divided by the exposure duration. The images are corrected for thermal drift and geometric distortion, but not for background.

Error array

The error array is another 16384 × 1024 image that has per-pixel errors computed from Poisson statistics using the gross counts and correcting for flat-field effects and the dead time.

Science spectrum

This includes the same quantities described for FUV TIME-TAG data.

12.1.4 NUV ACCUM data

The data products for NUV ACCUM mode are all as for the FUV except that the images are 1024 × 1024.

12.1.5 NUV ACQ/IMAGE data

Raw data

The raw data for ACQ/IMAGE is 1024 × 1024 × 16 bits.

Corrected image

The corrected image is 1024 × 1024 × 16 bits and includes effective counts sec^-1, the gross counts multiplied by a pixel's factor and divided by the exposure duration.