"Variance". To create a variance or quality array the NDF must already exist.
TRUE, the initial records of the unformatted file are interpreted as a FITS header (with one card image per record) from which the shape, data type, and axis centres are derived. The last record of the FITS-like header must be terminated by an END keyword; subsequent records in the input file are treated as an array component given by COMP.
!) value for NOPEREC causes the size of first dimension to be used.
"Data"the NDF is modified rather than a new NDF created. It becomes the new current NDF.
[40,30,20]would create 40 columns by 30 lines by 10 bands. It is only accessed when FITS is
"_UWORD"corresponding to signed byte, signed word, real, integer, double precision, unsigned byte, and unsigned word. See SUN/92 for further details. An unambiguous abbreviation may be given. TYPE is ignored when COMP =
"Quality"since the QUALITY component must comprise unsigned bytes (equivalent to TYPE =
"_UBYTE") to be a valid NDF. The suggested default is the current value. TYPE is also only accessed when FITS is
ngc253.datto the NDF called ngc253. The input file does not contain a header section. The NDF is two-dimensional: 100 elements in x by 60 in y. Its data array has type _REAL. The data records each have 8 values.
ngc253q.datto an existing NDF called ngc253 (such as created in the first example). The input file does not contain a header section. The NDF is two-dimensional: 100 elements in x by 60 in y. Its data array has type _UBYTE. The data records each have 100 values.
ngc253.datto the NDF called ngc253. The input file does not contain a header section. The NDF has the current shape and data type is unsigned word. The current number of values per record is used.
spectrumto the NDF called zz. The input file contains two header records that are ignored. The NDF is one-dimensional comprising 200 elements of type _REAL. There is one data record containing the whole array.
spectrum.listo the NDF called ZZ. The input file contains one header record, that is ignored, followed by a FITS-like header section, which is copied to the FITS extension of the NDF. The shape of the NDF is controlled by the mandatory FITS keywords NAXIS, AXIS1, …, AXISn, and the data type by keywords BITPIX and UNSIGNED. Each data record has AXIS1 values (except perhaps for the last).
the unformatted-file array is written to the NDF array as selected by COMP. When the NDF is being modified, the shape of the new component must match that of the NDF.
If the input file contains a FITS-like header, and a new NDF is created, i.e. COMP =
"Data", the header
records are placed within the NDF’s FITS extension. This enables more than one array (input file) to be
used to form an NDF. Note that the data array must be created first to make a valid NDF, and it’s the
FITS structure associated with that array that is wanted. Indeed the application prevents you from
The FITS-like header defines the properties of the NDF as follows:
Other data item such as HISTORY, data ORIGIN, and axis widths are not supported, because the unformatted file has a simple structure to enable a diverse set of input files to be converted to NDFs, and to limitations of the standard FITS header.