Bias correcting SST observations in OBS

[jwhile]

This code change will allow NEMO to read in fields of SST biases and remove theses biases from the observations before the observations are compared to the model.

This is a fairly minor change that has no impact on model dynamics. My plan for testing is to run the standard NEMO SETTE tests, but also conduct a run where the model is given a bias field and show that the observations change appropriately between the input observation files and the output feedback file.

The code will be reviewed by Arthur Vidard (avidard)

[jwhile]

The wiki ticket location for this ticket is https://forge.ipsl.jussieu.fr/nemo/wiki/ticket/1328

[jwhile]

Arthur, Could you confirm that you are happy with my planed tests

[avidard]

The proposed tests seem appropriate.

The SETTE tests will check that it does not affect the dynamics, and the other test will check that the new feature does what it should.

[jwhile]

To add more detail about the proposed test for the bias correction:

Two runs will be made in ORCA025, one run will be given a bias field, while the other will not have a bias field. In all other aspects the runs will be the same. If the SST bias correction is working correctly then the difference between the output SST feedback files will be identical to the negative of the bias correction field that was read in. This can easily be check by a visual inspection of a random selection of the observations.

[jwhile]

demonstration of bias in AMM12 model

[jwhile]

Demonstration of bias in ORCA2 model

[jwhile]

The updated code has been submitted to ​https://forge.ipsl.jussieu.fr/nemo/browser/branches/2014//dev_r4650_UKMO14.5_SST_BIAS_CORRECTION.

The code passes the SETTE tests as is detailed on the standard wikki page ​https://forge.ipsl.jussieu.fr/nemo/wiki/ticket/1328.

[jwhile]

To test the new functionality I have done tests in the ORCA2 and AMM12 configurations. In both cases I gave a model a bias field of +5 degrees in one half of the domain and -5 degrees in the other (the division is N/S in AMM12 and E/W in ORCA2). To pass the test this bias should have been subtracted from the observations in the output feedback files. As can be seen in the following images both tests passed:

AMM12 (NOTE: real observations were used for this test)

ORCA2 (NOTE: I used synthetic observations all of the same value)

[jwhile]

Artur, given the above could you please fill out a reviewers twiki for this code change. I have already set one up for you at:

​https://forge.ipsl.jussieu.fr/nemo/wiki/ticket/review/1328

[jwhile]

The code has been approved by the reviewer and will be committed as is.

The reviewer has raised important points about variable names within the OBS code. These should be addressed by a future rationalising of the OBS code.

[jwhile]

Ticket closed as code committed long ago.