Hi Bruno,
Thank you for the report. This is a
spurious message for this dataset. When initializing a dataset,
Ferret tests that all of the coordinate variables are
monotonically increasing. A parameter in the internal code was
being incorrectly initialized, and that triggered this false
result on the test of the time axis. As you saw, in fact the time
axis is increasing, and Ferret is not in fact doing any
"micro-adjusting" as it stores the axis information. We will get
this fixed in the next release.
By the way for those who may not have
run into this, if there were repeated values, for instance if we
repeated the second value so that the timesteps are 600432,
600433, 600433, 600434; then Ferret would allow you to work with
the dataset by applying this micro-adjusting,adding a small amount
to the second of the repeated coordinate values:
yes? use repeated_t.nc
*** NOTE: Axis has repeated values --
micro-adjusting time
yes? list/l=1:4/precision=12 t[gt=u]
VARIABLE : T
axis TIME
FILENAME : repeat.nc
SUBSET : 4 points (TIME)
01-JUL-2018 00:00:00 / 1: 600432.000000
01-JUL-2018 01:00:00 / 2: 600433.000000
01-JUL-2018 01:00:00 / 3: 600433.000074
01-JUL-2018 02:00:00 / 4: 600434.000000
Ansley
On 10/19/2018 5:57 AM, Bruno Levier
wrote:
Hi!
Loading a netcdf file (in attachment of this email) with FERRET
v7.43, I've got this warning:
use test.nc
*** NOTE: Axis has repeated values -- micro-adjusting time
But in fact the time variable has no repeated values at all.
This is the time variable listed by Ferret:
list /l=1:5 /pre=12 T[G=u]
VARIABLE : T
axis TIME
FILENAME : test.nc
SUBSET : 5 points (TIME)
01-JUL-2018 00:00 / 1: 600432.000000
01-JUL-2018 01:00 / 2: 600433.000002
01-JUL-2018 02:00 / 3: 600434.000004
01-JUL-2018 03:00 / 4: 600435.000006
01-JUL-2018 04:00 / 5: 600436.000008
And this the time variable in the file:
ncks -C -H -v time test.nc | head -5
time[0]=600432
time[1]=600433
time[2]=600434
time[3]=600435
time[4]=600436
Any idea of what is going on?
Thanks,
Bruno
--