Defects Index | Documentation Table of Contents
97/07/17 16:45 DEFECT in program page4
SYMPTOM: Discontinuities in post-fit plot for satellite PRN01
PROBLEM: Combination of flags exploited a hole in testing for piece-wise,
linear (PWL) flags. Although unusual before, this setup will
become more common. Circumstances:
a) PWL tropo on
b) tropo of first site in sequence off (relative tropo estimates)
therfore its matrix index was initialized to zero.
c) first satellite in sequence required than two biases
In the routine pwlflg, which tests if a parameter is PWL, the
tests for PWL tropo's is:
1) PWL tropo's on AND
2a) matrix index valid OR
2b) matrix+1 index valid
PWL tropo's on; first site index is zero (0) and not valid, BUT
0+1 was valid and 1 was the first satellite bias. Program was
getting first and last bias of satellite correctly, but, when
testing for tropo's, would erroneously advance satellite bias.
CORRECTION: Modified PWL tropo test in pwlflg to be:
1) PWL tropo's on AND
2) matrix index is not zero AND
3a) matrix index valid OR
3b) matrix+1 index valid
FOUND BY: L Hall and S Hilla
FIXED BY: M Schenewerk
VERSION: Beta
SOURCE: pwlflg.f
NOTES: Will go out with next upgrade.
Hilla has put in place for Hall.
970717.page4
June 25, 1999
Steve Hilla