Compares a pair of NDFs for equivalence
"FALSE"
if
they are significantly different in any way, and to "TRUE"
if they are not
significantly different.
If they are not similar, a textual description of the differences is written to standard output, and to any file specified by Parameter REPORT.
The two NDFS are compared in the following ways. Each test has an integer identifier, and the list of tests to be used can be controlled by Parameters DOTESTS and SKIPTESTS. Tests that are not included by default are indicated by the test number being in square brackets. Some tests have parameters that control the exact nature of the test. These are listed in parentheses at the end of the description test listed below.
"V"
–- The numerical value is a signal-to-noise value. The absolute difference in pixel
data value is divided by the square root of the smaller of the two variances
associated with the pixels (one from each input NDF). If the resulting ratio is
smaller than the ACCDAT value, then the two pixel data values are considered to
be equivalent. An error is reported if either NDF does not have a VARIANCE
component.
"R"
–- The numerical value is a relative error. The absolute difference between the two
pixel data values is divided by the absolute mean of the two data values. If the
resulting ratio is smaller than the ACCDAT value, then the two pixel data values are
considered to be equivalent. To avoid problems with pixels where the mean is close to
zero, a lower limit equal to the RMS of the data values is placed on the mean value
used in the above ratio.
"A"
–- The numerical value is an absolute error. If the absolute difference in pixel
data value is smaller than the ACCDAT value, then the two pixel data values are
considered to be equivalent.
If no character is included in the ACCDAT string, "R"
is assumed. ["1E-6 R"]
[0.2]
"R"
–- The numerical value is a relative error. The absolute difference in variance
value is divided by the absolute mean of the two variance values. If the resulting
ratio is smaller than the ACCVAR value, then the two pixel variances are considered to
be equivalent.
"A"
–- The numerical value is an absolute error. If the absolute difference in variance
values is smaller than the ACCVAR value, then the two pixel variances are considered to
be equivalent.
If no character is included in the ACCVAR string, "R"
is assumed. ["1E-6 R"]
TRUE
, then string comparisons are case sensitive.
Otherwise they are case insensitive. [TRUE]
[!]
!
) if all
tests are to be included in the initial list. This initial list is modified by
excluding any tests specified by Parameter SKIPTESTS. [!]
"R"
–- The numerical value is a relative error. The absolute difference in the
number of bad values is divided by the mean number of bad values in both NDFs
(for the QUALITY array, the total number of pixels in the NDF is used as the
denominator in this ratio). If the resulting ratio is smaller than the NBAD value,
then the two NDFs are considered to be equivalent for the purposes of this
test.
"A"
–- The numerical value is an absolute error. If the absolute difference in the
number of bad values is smaller than the NBAD value, then the two NDFs are considered
to be equivalent for the purposes of this test.
If no character is included in the NBAD string, "R"
is assumed. ["0.001 R"]
[!]
!
) value is supplied, the initial
list is left unchanged. [15]
FALSE
on
exit if any of the used tests indicate that the two NDFs differ. TRUE
, then trailing or leading white space is ignored when
comparing strings. [FALSE]