Opened 15 years ago
Closed 14 years ago
#225 closed defect (fixed)
Bug in ropp_pp_spectra
| Reported by: | Ian Culverwell | Owned by: | Ian Culverwell |
|---|---|---|---|
| Priority: | normal | Milestone: | 5.1 |
| Component: | ropp_pp | Version: | 4.1 |
| Keywords: | ropp_pp_spectra | Cc: |
Description
User João Francisco Galera Monico
has discovered this bug in ropp_pp_spectra:
! 3.4.1 Compute wave number
IF (ic == 1) k = 2.0_wp * pi * f_L1 / C_Light
IF (ic == 1) k = 2.0_wp * pi * f_L2 / C_Light (It would not be (ic==2)
?????
ic is channel number, so the 2nd line should say IF(ic == 2) ...
This is what's done in OCC (thanks to Kjartan Kinch at DMI for checking).
João Francisco finds that correcting this makes some significant differences to his results - see attached. (I think he means it changes L2 results, not L1.)
Attachments (3)
Change history (7)
by , 15 years ago
| Attachment: | SUBROUTINE ropp_pp_spectra.docx added |
|---|
comment:1 by , 15 years ago
| Component: | ROPP (all) → Preproc |
|---|
comment:2 by , 15 years ago
| Summary: | Buf in ropp_pp_spectra → Bug in ropp_pp_spectra |
|---|
comment:3 by , 14 years ago
| Milestone: | 6.0 → 5.1 |
|---|
This is an obvious bug. Correcting it makes a significant (~10-20%) difference to the L1 power spectrum |ln U| almost everywhere. (I was wrong before: it is the L1 results that are wrong.) The differences in the phase are just about discernible by eye (see the left hand panels):
before
after
Happily, it doesn't cause any failures to the standard ropp_pp test procedures.
Implement in ROPP5.1.
by , 14 years ago
| Attachment: | ropp_pp_spectra_dt_cntl.gif added |
|---|
by , 14 years ago
| Attachment: | ropp_pp_spectra_dt_test.gif added |
|---|
comment:4 by , 14 years ago
| Resolution: | → fixed |
|---|---|
| Status: | new → closed |
Afterword: I've updated the reference figures for the "make test" part of ropp_pp accordingly.
SUBROUTINE ropp_pp_spectra.docx