snranal
John Gipson
Dan MacMillan

Abstract:

Table of contents:

1   Overview

2   Usage

3   Description of snranal algorithm.

4   Snranal Output

4.1   Tables of snranal results
4.2   SUMRY program output
4.3   SKED program output
4.4   SNR histograms.
4.5   Sigma histograms.
4.6   SNR by time and source.

5   History

1   Overview

The basic formulation of snranal is to express each observed source flux as the product of the predicted flux and a source factor. The achieved antenna SEFD is expressed as the predicted SEFD divided by the square root of an antenna SNR factor since SNR is inversely proportional to the square root of the SEFD. For each observation, the ratio of the achieved SNR and the predicted SNR is then expressed as the product of the source factor and the antenna factors of the two observing antennas. The factors for each antenna and each source are determined by performing a least squares fit over all the observations. The snranal algorithm is discussed in more detail below. The program snranal has the following features: - Command line driven - Processes both S and X band data - Compares predicted and measured station and source performance - Graphics output can be sent to the screen, to the printer, or not at all - Histograms can be created of measured, achieved, and corrected SNRs and sigmas by station - Time history of SNRs by station can be plotted

2   Usage

Snranal is command line driven. If you just type snranal without any arguments, it will remind you of the expected syntax: snranal -sched <directory> -database <name> -master <directory> -skedprog <filename> [-plot <screen,print, or none>] [-outdir <directory>] [-outfile <filename>] [-histogram <yes or no>] [-observations <yes or no>] [-status <filename>] -sched schedule file directory. Here the schedule file directory name does not include './year/sesscode/' (snranal determines this from the database name). -database database name -master masterfile directory -skedprog sked executable name -plot The default for the flag -plot is none. Otherwise plots will be printed or displayed on the screen. -outdir All snranal output files are placed in the -outdir directory. -outfile Filename for snranal result tables. Default is that the file will be written to the -outdir directory -histogram If histogram is turned on, snranal will produce histograms of measured and expected SNR. The default is 'off'. -observations If observations is turned on snranal will produce plots of the SNR by observation for different stations. The default is 'off'. -status optional run status filename Full pathnames are required (-master, -skedprog,-outdir) except for -sched as noted above. The remaining flags, given in [...], are optional. Flags are case insensitive, as is the database name. The database name does not need the $ character as the first character although it is allowed.

3   Description of snranal algorithm.

1. The first thing snranal does is to run sked to generate a summary of the expected observation schedule. This generates a file of the form: Experiment_code.sked_list 2. Following this, snranal runs sumry on the S-band and X-band databases to get the achieved SNRs and integration times from the databases. It generates files of the form: Experiment_code_x.sumry Experiment_code_s.sumry 3 Following this, snranal goes through these three files, matching up the observations. For each observation it extracts the stations, sources, predicted and measured SNR at each band, and predicted and actual integration time. 4 The model it uses for the station and source behavior is that during each experiment, the stations are a constant multiplicative factor weaker (or stronger) than predicted and so are the sources. Therefore, the measured SNRs are related to the predicted SNRs by: SNRmeas = SNRpred*Stat1_fact*Stat2_fact*Source_fact where the last three terms on the right hand side are correction factors. Snranal takes the log of this equation to obtain: log (SNRmeas) = log(SNRpred)+log(Stat1_fact)+log(Stat2_fact) +log(Source_fact) This equation can be solved in a least squares sense for the log(Stat_fact) and log(Source_fact). Snranal does this, assuming that "on average" all sources and all stations behave as expected. This last assumption is necessary because if all the sources got X times as strong, and all the stations sqrt(X) as weak, the measured SNR would not change. You need some constraint to fix this "X" factor. Snranal imposes the constraint that on average the adjustments are 0: Sum log(stat_fact)- 2 * Sum log(src_fact) = 0. Snranal imposes the additional constraint that any source that is observed less than 5 times has its nominal strength.

4   Snranal Output

4.1   Tables of snranal results

Snranal produces tables that summarize the expected versus measured performance for all sources and stations. Any sources or stations that differ by more than 30% from their expected value are flagged. The source tables show the estimated source factors and the achieved source fluxes. Currently, the station tables show the antenna SNR factors from the experiment. A future update will show the achieved SEFDs based on the antenna SNR factors and the predicted SEFDs. This summary is put in the output file specified on the command line. If no output file is specified, the output goes into the output directory specified on the command line. The file name has the form Experiment_code.snranal

4.2   SUMRY program output

Snranal calls the program SUMRY to get parameters including the achieved SNRs, integration times, and quality codes. SUMRY creates a table of this information for each observation. The X-band and S-band tables are saved with names of the form in the output directory specified on the command line Experiment_code_x.sumry Experiment_code_s.sumry

4.3   SKED program output

Snranal also calls the SKED program to make an output of scheduling parameters such as source flux models and antenna SEFDS, and a listing of scheduled observations including predicted SNRs and scan lengths. This SKED output is saved in a file in the output directory given in the command line with the form Experiment_code.sked_list

4.4   SNR histograms.

Snranal bins the S- and X-band into bins which are 5 wide. It also calculates the cumulative distribution of SNR, e.g., what percentage of points has an SNR below 35. These are written out to files on a station by station base. These have filenames of the form and are written to the output directory from the command line Experiment_code.station_name.SNR If requested on the command line, these files are printed to the laser printer, or plotted on the screen.

4.5   Sigma histograms.

Snranal calculates the S- and X-band sigmas based on the SNR and the RMS spanned bandwidth. It bins these in these which are 5 ps wide. It also calculates the cumulative distribution of SNR, e.g., what percentage of points has a sigma below 35ps. These are written out to files on a station by station basis. These have filenames of the form and appear in the output directory Experiment_code.station_name.SIG If requested on the command line, these files are printed to the laser printer, or plotted on the screen.

4.6   SNR by time and source.

SNR creates files which contain each observation a station is in by time and source. These have filenames of the form and appear in the output directory Experiment_code.station_name.obs

5   History

This program has a long history going back to Clara Kuehn. It is fair to say that about the only thing new snranal has in common with the old version is the name. A fundamental difference between this version of snranal and previous versions is that this version is station oriented, as opposed to baseline oriented. WHO WHEN WHAT J. Gipson 1995.04.14 Wrote D. MacMillan 2001.01.16 Changed interface. The program is integrated to Calc/Solve


Questions and comments about this document should be sent to:

Dan MacMillan (dsm@leo.gsfc.nasa.gov)

Last update: 2001.01.16