Rejection of entire profiles containing low background pressures/temps in FM

[cburrows]

Currently the operational Met Office model runs with 70 vertical levels and this raised model top gives rise to pressure and temperature values which are below the lower limits set by ROPP. When running ropp_fm_bg2ro_1d, ropp2ropp sets these values to the missing data indicator, and ​ropp_fm_roprof2state.f90 then rejects the entire profile, because it contains a the missing data value which is negative (line 165 and onward).

Example output:

---

ROPP Forward Model - v5.0

---

INFO: Processing profile 1 of 1 INFO: (OC_20110729180000_META_U999_METO)

WARNING (from ropp_fm_roprof2state): One or more pressure values are negative.

INFO (from ropp_fm_roprof2state): (Minimum pressure value = -0.1E+09)

INFO (from ropp_fm_roprof2state): Check input data and valid_range attributes

INFO (from ropp_fm_roprof2state): Set status flag state%ok to FALSE

The first action should be to change the ranges for background temperature and pressure. Changing the minimum allowed pressure value to 0.001hPa instead of 0.01hPa, and the minimum temperature value to 80K instead of 150K would result in the acceptance of all the profiles currently being tested (UCAR profiles from 29/7/11). Whatever the range is modified to, this must clearly be announced to users, as data which is currently being rejected, may now be accepted.

Also, perhaps, the test for negative values needs a secondary condition that the negative value is not equal to the missing data indicator to avoid rejection of entire profiles because of one missing value.

Note that until this is addressed, no recent Met Office backgrounds will be accepted by the ROPP forward model.

[cburrows]

This was found (partly) to be a data issue - the processing carried out by ropp2ropp checks the data range in the netcdf file rather than intrinsic values. The pressure limit in ROPP was reduced considerably at changeset 2184.

Whole profiles will still be ignored if a single missing data value is present, which may be undesirable.

Finally, the range of Met Office temperatures seems to be falling outside the ROPP range at 70km and above. This seems to be mostly confined to the winter hemisphere. Here are some profiles for 29/7/11 (note, the top level is 80km and the next highest is 72km):

Means and standard deviations are shown in these plots, as well as the current minimum temperature in ROPP.

[cburrows]

It seems that a fix for the pressure has already been implemented, but there is still an issue with the temperature range. Currently the valid range is 150-350K. Met Office background temperatures of 80K at the highest level are probably spurious, but the rest of the profile is largely valid, so it shouldn't be rejected in its entirety. Hopefully a combination of customised config files and -no_ranchk flags will address the problem.

[cburrows]

The Met Office configuration file has been updated to reduce the minimum temperature to 75K to allow more profiles to pass range-checking. No other modifications to ROPP are needed - the reason for the low temperatures will be investigated at the Met Office.