Документ взят из кэша поисковой машины. Адрес
оригинального документа
: http://www.adass.org/adass/proceedings/adass96/verkhodanov1.html
Дата изменения: Tue Jun 23 21:17:24 1998 Дата индексирования: Tue Oct 2 03:49:35 2012 Кодировка: Поисковые слова: ultraviolet |
Next: NOVIDAS and UVPROC II-Data Archive and Reduction System for Nobeyama Millimeter Array
Previous: The OPUS Pipeline Toolkits
Up: Software Systems
Table of Contents - Index - PS reprint - PDF reprint
O. V. Verkhodanov
Special Astrophysical Observatory, Nizhnij Arkhyz,
Karachaj-Cherkessia, Russia, 357147
The RATAN-600 (Parijskij & Korolkov 1986) is a transit radio telescope with a variable profile allowing simultaneous multi-frequency observations using different feed horns. At present, the wide-band radiometers at 0.968, 2.31, 3.95, 7.69, 11.1, and 22GHz are mounted on feeder cabin No. 1. The large number of frequencies allows observations of instantaneous spectra of radio sources. A high sensitivity to extended sources permits deep sky surveys, detection of new radio sources, and studies of the fluctuations of the microwave background radiation. At the higher frequencies (11.1, 22GHz) we may account for the influence of the atmosphere, and at lower frequencies (0.968GHz) we may obtain the Galactic background level, and clean the data at intermediate frequencies (Parijskij & Korolkov 1986).
The result of a single observation at the RATAN-600 from one receiver is a one-dimensional scan of the sky. This data vector contains points of measured antenna temperatures of the sky with a constant registration interval in sidereal time.
Based on these features we have created a ``Flexible Astronomical Data Processing System''(FADPS, Verkhodanov et al. 1993), operating under different flavors of UNIX (Linux, Solaris, etc.).
The usual stages of the RATAN-600 data reduction are the following. After observation, the registration program automatically removes interference spikes by resampling the raw data using robust methods (Shergin et al. 1997).
The user may then further smooth the data using the same algorithms to estimate the low noise component or the background (Figure 1). After subtraction of the background from the primary data, Gauss-fit analysis may be used to find components of the source and estimate their parameters.
Figure: Four iterations of the FADPS background curve calculation.
Original PostScript figure (29kB).
Figure: The screen of the fgr program with the menu and drawn data.
Original PostScript figure (1.55MB).
Our approach in the construction of FADPS follows the main principles of operation of programs in a data processing system and the rules of operation in UNIX.
The command languages of UNIX (Bourne shell and
C shell) are used as the command languages of our system. These languages
have loop, condition, and local transition operators. The process
of data reduction is a consecutive application of operators (commands)
to the user's data recorded as equidistant measurements and stored
in separate files. Each operator corresponds to a specific
command called a ``module.'' Using a set of such modules
and a basic knowledge of UNIX the user may construct his/her
own data processing system or simply type a sequence of modules
to obtain the desired result. The command modules can read
data from standard input and write data to standard
output and may be used as filters in UNIX pipes.
Let us consider some vector containing observational data
of a radio source.
To obtain the parameters of a source we have to fit a Gaussian to
our record. The standard analysis
consists of the consecutive application of
two operators: the background subtraction operator and
the Gauss-fit operator
:
where D is the initial vector and R is the resulting vector. Under UNIX the application of both operators corresponds to the following command string:
where bgd is the program for background subtraction, gauss is the program for Gaussian fitting, the key `` -w'' defines that the subsequent parameter is the size of the window for smoothing, the key `` -n'' defines that the subsequent parameter is the lower amplitude level of the Gauss-fit.
The system includes the commands to perform the following: Gauss-fit analysis (to estimate the parameters of sources and to decompose them into separate components), fitting of tabular beam patterns, non-linear smoothing and averaging on the basis of robust algorithms (Shergin et al. 1997), background calculation and subtraction, interpolation, convolution, Fourier cleaning, visualization in X-Windows or in PostScript, and reading and writing files in FITS and FITS BINARY TABLE format.
These commands may be executed as UNIX scripts or using our own graphics shell for our data processing system in the X-Window system. The shell scripts may be used in ASCII terminal mode. The interactive mode of the FADPS is available in the graphics program fgr (Files GRaphics), where a user may select required operations from a menu and see the result of the execution in the X-Windows screen of fgr (Figure 2). fgr also permits the execution of external users' filters.
FADPS also operates with spectra of radio sources recorded as FITS TABLEs. The system supports fitting a spectrum with a set of curves using different weights for different spectral points. It calculates spectral indices and fluxes at any desired frequency. The graphics program spg (SPectra Graphics) operates with these functions (see Figure 2 in Verkhodanov et al. 1997). The spg menu provides a least-squares fit of a spectrum with several curves. The choice among these curves is either automatic or via mouse clicks. The mouse also allows a manual fitting (when the cursor delineates the curve) and the deletion of unreliable or unwanted data points (or attaching a zero weight to them). Data points may be weighted in different ways: using equal weights, weights by flux density errors, or on a point-by-point basis. This system is connected with the CATS database of radio catalogs (Verkhodanov et al. 1997) operated at the RATAN-600.
At present the FADPS is used at the RATAN-600 and in some others Russian institutes as the system to operate on one-dimensional vectors of data recorded in FITS format. An additional Tcl/Tk interface over the FADPS modular structure is being developed. The software is written in the algorithmic language ``C'' and may be freely distributed.
This work was supported partly by grant No 96-07-89075 of the Russian Foundation for Basic Research. The author thanks ISF-LOGOVAZ Foundation for the travel grant, the SOC for the financial aid in the living expenses and the LOC for the hospitality. The author is thankful for H. Andernach for critical notes on this paper.
Parijskij, Y. N., & Korolkov, D. V. 1986, Ap.Sp.Sci.Rev., Ser. D 5, 40
Shergin, V. S., Verkhodanov, O. V., Chernenkov, V. N., Erukhimov, B. L., & Gorokhov, V. L. 1997, this volume
Verkhodanov, O. V., Erukhimov, B. L., Monosov, M. L., Chernenkov, V. N., & Shergin, V. S. 1993, Bull. of SAO, 36, 132.
Verkhodanov, O. V., Trushkin, S. A., Andernach, H., & Chernenkov, V. N. 1997, this volume
Next: NOVIDAS and UVPROC II-Data Archive and Reduction System for Nobeyama Millimeter Array
Previous: The OPUS Pipeline Toolkits
Up: Software Systems
Table of Contents - Index - PS reprint - PDF reprint