timeseries plotting errors

[eclare108213]

The timeseries plotter appears to be picking up values that it should not.

These issues were noted in #533, for nonstandard plots (so the problem might be my editing of the timeseries script, rather than a problem originally in the script). They occur with all types of forcing (JRA, CORE, NCARbulk). I'm attaching two examples, one for sst and one for the area-weighted total energy plots.
examples.zip

In the sst example,

grep sst cice.runlog* > sst_gx1_jra_all
grep "sst (C)" cice.runlog* > sst_gx1_jra_C
diff sst_gx1_jra_all sst_gx1_jra_C > diff_sst
less diff_sst

<  min/max sst, frzmlt
<   min, max, sum =  -1.96869665044293        32.9603346231369        1525878.86583381      sst
33d30
<   min, max, sum =  -1.90458264992426        32.7596970527221        1530936.61448477      sst
64d60
<   min, max, sum =  -1.90458264992426        31.7854126938785        1534787.65564628      sst
95d90
<   min, max, sum =  -1.90458264992426        31.9466864214512        1526806.43659672      sst

The last values (~1.5e6) in the latter four lines in the diff_sst file appear to correspond with the four spikes in the plot. The timeseries plot was created with timeseries_all.csh, which greps on the "ssh (C)" form, so I don't think it should be picking up these values. But maybe I don't understand how the script works. Could this be a line-continuation problem?

A similar thing seems to be happening in the arwt_tot_energy plot (but not for the arwt_tot_energy_chng plot).

I didn't try the python script.

Also, it might be nice to add options for plotting more variables to the script (or add scripts). Just a thought.

[mattdturner]

@rallard77 Had contacted me last week about a similar issue. I think I have a quick fix for both the timeseries.csh and timeseries.py scripts that should resolve the issue with these fields. I'm testing it now, and will create a PR once I'm satisfied with the change.

Also, it might be nice to add options for plotting more variables to the script (or add scripts). Just a thought.

A while back I briefly tried adding an option to allow users to pass variables as command line arguments. I forget the issues that I ran into, but I can look at it again.