Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

How to deal with ERROR: Gaussian_latitude_computation?


New member

I recently experienced that WRF did not run on the input files that I had generated by running real.
After checking the met_em files I discovered that I had no data in the SST and SEAICE variables.
I later found out that I had defined the wrong geo information for these data when making the intermediate files using pywinter (documentation: pywinter — pywinter 2.0.6 documentation).
The first time I produced the intermediate files I had assumed a cylindrical equidistant grid instead of the gaussian grid.
In the geo information for the Gaussian grid one must give the southwest corner latitude and longitude for the data. In this instance, I get the values -78.67, 82.85 for the latitude and longitude respectively. I attach a quick plot of one of the data files.
The problem is that when I try to run metgrid on the files with the correct (gaussian) grid defined, I encounter an error message:

Processing 1980-01-01_06
Oops, something is not right with the Gaussian latitude computation.
The input data gave the starting latitude as -78.675.
This routine computed the starting latitude as +- 89.661.
The difference is larger than 0.01 degrees, which is not expected.
ERROR: Gaussian_latitude_computation

Here it seems that the latitude of the corner is not as expected by metgrid, even though this is indeed the latitude of data point.
How do I fix this in order to get metgrid to run again?
Thanks in advance for any tips!


  • SST-1980-01-01_12.jpeg
    59.6 KB · Views: 4
Many apologies for the delay in response. Most of us were off for the holidays and are now trying to get caught up.

Have you gotten past this issue yet? If not, are you able to plot the original data to show the lat/lon coordinates? If possible, can you attach one of the data files? If the file is too large and you don't have another method of sharing, please see the home page of this forum for instructions on sharing large files. Please also let me know the source of the data. Thanks!
hello! Thanks for your reply and no worries about the delay.
Unfortunately I still haven't figured out the issue yet. I have uploaded the intermediate file being read by metgrid to the nextcloud storage. Its called FILE:1980-01-01_06.
The latitude and longitude grid are both 2D, so I have made a quick plot of the lat/long grids here which correspond to the data plotted in the first message.
It is the MPI-ESM1-2-HR dataset, downloaded from
For the SST data I have selected experiment ID=historical, variant label=r1i1p1f1, grid label=gn, Table ID=Oday, variable=tos


  • Latitude_SST.jpeg
    32.3 KB · Views: 2
  • Longitude_SST.jpeg
    32.7 KB · Views: 2
If I use the WPS/util/rd_intermediate.exe utility to look at the data file you shared, it does show the latitude for SST and SEAICE as -78.674881, but all other variables have latitude -89.284225. I now realize you're using only a subset of Gaussian data, which the ungrib program is not able to handle. For Gaussian data, ungrib only supports the use of a full global dataset. Is it possible for you to get the full set?
Ah I see. I actually did not use ungrib to make the intermediate files, it is a python package called pywinter that is run on the data I downloaded.
I wasn't aware that the SST and SEAICE datasets were only a subset of the full global dataset, so I guess I will have to contact the folk at MPI to find out about this. Thanks for the explanation!