Anomalies over user-specified time steps
Added by Nathanael Harwood over 5 years ago
Hi all - I'm aware of the YDAYARITH etc operators that let you subtract a time series from the climatology
I'm trying to do the same with 5-day averages as the time step, with the following steps:
cdo -timselmean,5 -select,season=DJF 850tNHem.nc 850_5day.nc #create 5 day averages
cdo -timselmean,5 -select,season=DJF -ydaymean 850tNHem.nc 850_ymean.nc #create multi-year mean of the same steps (climatology)
cdo ydaysub 850_5day.nc 850_ymean.nc 850_anomaly.nc #sub time series from climatology - create anomaly.nc
Is this the correct approach? I'm getting Error (cdf_put_vara_double): NetCDF: Numeric conversion not representable
More than happy to provide subset of the data if needed
Replies (3)
RE: Anomalies over user-specified time steps - Added by Karin Meier-Fleischer over 5 years ago
Hi Nathanael,
please, upload test data to reproduce your problem.
-Karin
RE: Anomalies over user-specified time steps - Added by Nathanael Harwood over 5 years ago
Hi Karin,
No problems - Attached is a 2 year subset of the data, 850hPa temperature.
Many thanks
| 850tNHem.nc (45.6 MB) 850tNHem.nc |
RE: Anomalies over user-specified time steps - Added by Karin Meier-Fleischer over 5 years ago
The warning tells you what is going wrong and what to do to solve the problem:
cdo ydaysub (Warning): Some data values (min=-190905 max=-176216) are outside the
valid range (-32768 - 32767) of the used output precision!
Use the CDO option -b 32 or -b 64 to increase the output precision.
cdf_put_vara_double : name=t type=NC_SHORT minval=-190905.000000 maxval=-176216.000000
Error (cdf_put_vara_double) : NetCDF: Numeric conversion not representable
Add -b 32 to your last cdo command
cdo -b 32 -ydaysub 850_5day.nc 850_ymean.nc 850_anomaly.nc #sub time series from climatology - create anomaly.nc
-Karin