Hi Richard
It is not uncommon to see negative values on variables that are
supposed to be non-negative (such as dust concentration,
mixing ratios of gases, etc)
because the data was interpolated from the original
grid, normally a model grid, where it did not have negative values,
to a more user friendly grid, say, a lat-lon grid,
using an algorithm that may overshoot the original values,
for instance a cubic spline algorithm,
or the now fashionable "flux conservative" algorithms.
In the ECMWF case (which if I remember right uses a spectral algorithm),
I would guess that the conversion from spectral to
spatial domain may have something to do with the negative values as
well, because the conversion relies on discrete Fourier Transforms
(DFTs/FFTs) and Legendre polynomials, which can also generate spurious
negative values.
We have seen plenty of this, on data coming from spectral models,
from models that use cubed-sphere grid, etc.
Fortunately the spurious negative values tend to be localized
in areas where the overall values of the data are already low (e.g.
water vapor in the stratosphere), suggesting that the negative values
are just a numerical artifact, and those data points should not be used
for scientific purposes.
One workaround, which we have used, is to "rectify" any negative values
to zero, or to some very small positive number
below the normal range of that variable.
You can easily do this with NCO or CDO.
I hope this helps.
Gus Correa
On 11/04/2017 10:07 AM, Richard Cloete wrote:
Hi all,
Pretty new to NetCDF and ECMWF data.
Just a quick (probably with an obvious answer) question: What do I do
with negative values? For example, I am using aermr04, aermr05 and
aermr06 (dust) data from ECMWF. I have noticed many negative values.
What do I do with these data? Shall I just treat them as zeros? I ask
because I am calculating the concentrations using the following formula:
https://software.ecmwf.int/wiki/display/CKB/How+to+compute+concentration+from+a+mixing+ratio+field
Many thanks,
Richard
_______________________________________________
NOTE: All exchanges posted to Unidata maintained email lists are
recorded in the Unidata inquiry tracking system and made publicly
available through the web. Users who post to any of the lists we
maintain are reminded to remove any personal information that they
do not want to be made public.
netcdfgroup mailing list
netcdfgroup@xxxxxxxxxxxxxxxx
For list information or to unsubscribe, visit:
http://www.unidata.ucar.edu/mailing_lists/
