Appendix 16. LAN PHA Documentation (continued)

 

A16.3 Outputs

 

The main outputs from program PHAGEN are three files containing the track, pulse, and MFSA data sets. These files are used to produce the PHA production plots via the PHA display programs. Though the output files may be written to either tape or disk (this is installation dependent), disk is the preferred storage medium. The files are not usually very big and the entire PHA analysis process is considerably faster using disk storage. In particular, the analysis software handles the disk file names (using the date of the requested data) which results in fewer keystrokes to run the programs. The user need be armed with only a date in order to access the file. In addition, if output is written to disk, PHAGEN will create files for each day's worth of data processed. This is a crude but effective method to avoid massive data loss due to any abnormal end of job (power loss, hardware/software failure, etc.).

 

A16.3.1 Track File

 

The first output file described is formally named the LAN Species Pseudo Count Rate Track file, otherwise known as the track file (for brevity). This file contains count rates derived through the combination of high duty cycle, broad channel count rates and low duty cycle, pulse height history data. The count rates in the track file are organized by energy bins and species. In addition, undefined events are included which may be indicative of the instrument health and the species energy mapping correctness.

 

The track file is a binary file containing the LAN common header block and the species track data. The header block contains the header values for the beginning and ending times of the track data record. Each track data record is 11736 bytes in length (2934 4-byte longwords) including the 1664 byte common header block.

 

Each record in the track file contains the five fields below immediately following the common header block:

 

1. Sectored species count rates for hydrogen, helium 3 and 4. These ion species contain 13 energy bins and 9 sectors, normal sectors 1 through 8, and the total across all sectors stored in "sector" 0. Five values are stored for each count rate, the rate itself (counts/second), number of events actually sampled and sample time for two possible contributing channels. The Fortran declaration for the sectored species data is

 

real*4         sector_species (5,0:8,13,3)

 

which organizes the data as 5 values (pseudo rate, n1, time1, n2, and time2) by 9 sectors (1-8, 0=totals) by 13 energy bins by 3 ion species. The n1 and n2 parameters are the number of PHA events analyzed for a particular species, energy bin and sector from any two WART channels possibly contributing events into the same species, bin, sector. Time1 and time2 are the sample times for the corresponding rate channels. Note: Most species, energy bins do not split channel boundaries, and therefore the majority of the n2 and time2 accumulations will not be used (i.e., 0). It is only those species and energy bins which split channel boundaries that will contain data in n2 and time2. The data have been kept separate for proper error analysis.

 

The energy bins are defined as having the following lower bounds for each bin:

 

Table A16-1 Energy bins

 

 

2. Unsectored species count rates for eight ion species containing 13 energy bins. The energy bins are defined the same as the sectored rates. Similar to the sectored data, five values are stored for each rate, the rate itself (counts/sec) and the n1, time1, n2, and time2 parameters. The Fortran declaration for the unsectored species track data is

 

real*4        species (5,13,8)

 

which organizes the data as 5 values (rate,n1,time1,n2,time2) by 13 energy bins by 8 ion species. The Fortran indexing is thus species (j,bin,species) where the species indexing is

 

Index	Species
1	Generic low energy CNO group
2	Carbon
3	Nitrogen
4	Oxygen
5	Neon
6	Silicon
7	Sulfur
8	Iron

 

3. The total number of events analyzed for nine Wart channels (W1-W8, 0=channel totals) and nine sectors (normal 1-8, 0=sector totals). The Fortran declaration of the event counts is

 

integer*4        chan_events(0:8,0:8)

 

which organizes the data as 9 channels (W0, W1-W8) by 9 sectors (0, 1-8). The Fortran indexing is thus chan_events(ich.isec) where ich is the channel number and isec is the sector number.

 

4. The total number of rate channel counts for nine Wart channels (W1-W8, 0=channel totals) and nine sectors (normal 1-8, 0=sector totals). The Fortran declaration of the event counts is

 

real*4         chan_counts(0:8, 0:8)

 

which organizes the data as 9 channels (W0, W1-W8) by 9 sectors (0, 1-8). The Fortran indexing is thus chan_counts(ich,isec) where ich is the channel number and isec is the sector number.

 

5. Undefined event counts for nine Wart channels (W1-W8, W0=undefined channel) and nine sectors (normal 1-8, 0=undefined sector). The counts contain the number of events which could not be identified as a particular species during the PHAGEN process. These counts serve as a check on the validity of the species energy map used in PHAGEN. If these counts are high, the species energy map probably needs recalibration. This could occur if any instrument drift is encountered. The Fortran declaration of the undefined event data is

 

integer*4          undef_events(0:8,0:8)

 

which organizes the data as integer counts for 9 channels (W0,W1-W8) by 9 sectors (0,1-8). The Fortran indexing is thus undef_events(ich,isec) where ich is the channel number and isec is the sector number.

 

To summarize, the LAN track records have the following structure:

 

Table A16-2 LAN track record definition

 

 

A16.3.2 Pulse File

 

The second output file from program PHAGEN is called the Pulse file. This file contains normalized, averaged rates (counts/sec) for each pixel in the 256 x 256 PHA pixel space for the LAN CA experiment. Similar to the track data set, the pulse matrix is a combination of high duty cycle count rate data (WART rates) and low duty cycle PHA event history yielding normalized rates for the time interval requested by the user. In addition, eight accumulation times for channels W1 through W8 are saved in order to calculate the actual counts from the rate data. There is no sector information retained in the pulse data set. The Fortran declaration for the pulse rates matrix and associated times is

 

real*4         times(8), pulses(256,256)

 

with the data organized as 256 rates across detector C (deposited energy) by 256 rows across detector D (delta energy). The Fortran indexing into the pulse matrix is then pulses (ic,id) where ic is the C index and id is the D index.

 

Due to the size of one pulse rates matrix (256 x 256 x 4 bytes), the output logical record spans 16 physical records, each physical record being 18080 bytes (4520 4-byte longwords) and having the following structure:

 

Field #	Length (bytes)	Description
0	1664	LAN Archive header block
1	32	Channel times for W1-W8
2	16384	Rates for next 1/16 of pixel space
Total	18080	(4520 longwords)

 

A16.3.3 MFSA File

 

The third output file from the PHAGEN program is called the MFSA file. This file contains normalized, averaged, sectored rates for the four detectors (F', F, M', and M) spectra accumulations. The four average spectra contain 32 energy channels and 8 sectors each. (The lower sector resolution data from detectors F and M are converted to the higher resolution 8 sectors.) The rates are normalized using the high duty cycle time synchronous rates in the MFSA Archive records (called R8, R4 in the LAN Archive Description) in combination with the low duty cycle spectral counts. The resultant MFSA data set contains 32 channel energy resolution ion count rates by sector. There is no sample time retained in the MFSA data set other than the implied time interval of the record. The Fortran declaration for the MFSA data block is

 

real*4       spec_totals(2,8,32,4), det_counts (2,8,8,4)

 

which organizes the spectra data as 2 values (pseudo rate and actual counts) for 8 sectors for 32 energy channels for 4 detectors. The actual counts acquired during the interval are retained for proper error analysis. The indexing for the spectra data is thus spec_totals(1=rate/2=counts,isec,ich,idet) where isec is sector number, ich is channel number (1-32), and idet is detector number. The detector order is 1:F', 2=F, 3=M', 4=M. In addition, the total detector rate channel counts and sample times are retained for 8 sectors, 8 channels, 4 detectors. The indexing for the rate channel data is thus det_counts(1=counts/2=time,isec.ich,idet) where isec is the sector number, ich is the channel number (note: F' and F only use 7 of the 8 channel locations), and idet is the detector number. The MFSA records contain 10240 bytes (2560 4-byte longwords) and have the following structure:

 

Field #	Length (bytes)	Description
0	1664	LAN archive header block
1	8192	MFSA spectra count rates
2	2048	M and F rate channel counts, time
Total	11904	(2976 longwords)

 

 

Next: A16.4 Processing

 

Return to HISCALE List of Appendices

Return to Ulysses HISCALE Data Analysis Handbook Table of Contents