#128 closed defect (fixed)
r_coc range too small
| Reported by: | Dave Offiler | Owned by: | Dave Offiler |
|---|---|---|---|
| Priority: | normal | Milestone: | 1.1 |
| Component: | ropp_io | Version: | 1.1 |
| Keywords: | range check | Cc: |
Description
Martin (DMI) reports that some r_coc (offset radius of curvature centre from Earth's centre) are being set 'missing' with v1.1. Valid range in ropp_io_type.f90 is +/-10km, but Martin wishes to use values ~31km, but the v1.1 range checker is (correctly) rejecting such large values.
Stig confirms calculations show offsets up to 42km near poles. Axel also confirms values in GRAS data can be > 10km.
Change history (3)
comment:1 by , 17 years ago
| Component: | ROPP (all) → I/O |
|---|---|
| Keywords: | range check added |
| Milestone: | → 1.1 |
| Priority: | major → normal |
| Resolution: | → fixed |
| Status: | new → closed |
comment:3 by , 16 years ago
| Milestone: | → 1.1 |
|---|
Note:
See TracTickets
for help on using tickets.
Checked that:
1) the BUFR encoder (ropp2bufr.f90) does not apply this 10km limit, but only uses the gross BUFR descriptor limit (1000km)
2) r_coc range is not used in any calculations within ROPP (apart from udunits scale change on I/O)
so changing the range definition in ropp_io_types should have no other impact.
Agreed with DMI that +/-50km range for r_coc is sufficient to allow all valid values through.
Martin edited his local file to 50km range, recompiled & ran ok, allowing through the larger values, confirming the change works as expected.
Change included in working copy and committed as [1519].