Defects Index | Documentation Table of Contents
98/05/06 DEFECT in program page4
SYMPTOM: When all integers were fixed on some baslines, a segment of
the phase residuals would become large outliers.
PROBLEM: Two defects were found:
1) page was not correctly tracking loss of lock (data gaps and
user input breaks). This caused the selection of the
(arbitrary) "satellite to have its phase ambiguity set
to zero" to vary as integers were fixed. (see note below).
This resulted in "new" ambiguities appearing erroneously.
2) The search of the integer file had a logic error which
created a hole. If two (or more) integers were listed
with ending times one epoch apart, the second integer
was "missed" and the now erroneous earlier integer was
held until the next check of the file.
CORRECTION: For defect 1:
in obsdd.f: modify the kntbia array, which had stored observation
counts if no integer was fixed, to store counts
as a position number if no integer is fixed and as
a negative number if integers are fixed. Thus
the obs count is always available and the sign
of that count can serve as a flag of a fixed integer.
in swbias.f:use modified kntbia to forward reduce and restart
ambiguities if float; simply follow phase continuity
if fixed.
in swbrk.f: use modified kntbia to forward reduce and restart
ambiguities if float; simply follow phase continuity
if fixed.
For defect 2:
in rdint.f: changed logic on test for next time to search
integer file.
FOUND BY: Baseline Advisory Group 98/04/30
FIXED BY M. Schenewerk 98/05/06
VERSION: 9805.06
SOURCE: /g1/HPUX.10/Src/Page4
/g1/HPUX.09/Src/Page4
EXECUTABLE: /g1/HPUX.10/Bin
/g1/HPUX.09/Bin
NOTES: Recall that because the phase ambiguities are constants and
page does double differencing, the solution of those ambiguities
is degenerate and singular, i.e. one phase bias can not be
estimated. The specific phase ambiguity and its value are
arbitrary in this system of equations. page chooses the current
reference satellite and sets its ambiguity to zero.
980506.page4
June 25, 1999
Steve Hilla