Opened 18 years ago
Closed 16 years ago
#78 closed defect (fixed)
Provide some diagnostic output from PGE based QC (and an overall diag % ok)
| Reported by: | marq | Owned by: | Huw Lewis |
|---|---|---|---|
| Priority: | normal | Milestone: | 1.2 |
| Component: | ropp_1dvar | Version: | 0.8 |
| Keywords: | Cc: |
Description
The routine ropp_1dvar_pge() provides an estimate of PGE for all data points, and the calculated PGE can optionally be used for quality control purposes as well.
Similar to ropp_1dvar_bgqc(), the PGE based QC should also give some overall information along the line of "how much of the data has been rejected", and set diag %ok to .false. is the number of rejects - or rather downweights - becomes too large. This, however, is not yet implemented.
Change history (3)
comment:1 by , 18 years ago
| Component: | ROPP (all) → 1DVar |
|---|
comment:2 by , 17 years ago
| Milestone: | 1.1 → 1.2 |
|---|---|
| Owner: | changed from to |
| Status: | new → assigned |
comment:3 by , 16 years ago
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
Note:
See TracTickets
for help on using tickets.
The PGE QC check described is essentially an O-B check since PGE is only a function of O-B/sigma_O-B. The O-B check applied in ropp_1dvar_bgqc rejects points where O-B > bg_reject_factor * sigma_O-B. Setting a PGE 'threshold' to reject data would just be replicating this functionality. Weights are set to zero in ropp_1dvar_bgqc if data are rejected. Weights depending on PGE are only applied to non-rejected data in ropp_1dvar_pge. For simplicity, I propose using the O-B QC as a tool for rejecting data only. PGE is then only used as a weighting tool.