Hi All, Just a bit more between Mario and I about his data, which we thought we'd pass on to finish this thread.  It turned out that Mario had named a variable T, which caused most of his trouble, but there was some useful discussion of curviilnear grids as well. If the lon, lat temp are just one-dimensional lists of points, then the 3-argument shade command can't use the curvilinear grid to plot the data. I'll show more about this at the end of this message. To look at your data in Ferret try these commands.    yes? USE test2.cdf    yes? show data    yes? show grid lon   What axis is the data on?  If it's on the X axis, look at some of the data,     yes? list/i=1:10 temp, lon, lat Is your variable named T?  If so, that could be causing confusion as Ferret expects T to represent the time coordinate of a variable.  You could either change the name of the variable when the data is corrected, or with ncdump and ncgen.  Or, if you really want to use T, then it can be referenced by putting it in single quotes:  yes? list/i=1:10 'T', lon, lat Ferret expects curvilinear data to be on a 2 dimensional grid, I x J, where at each i,j point you have the latitude and longitude point corresponding to the curvilinear grid, and the value of the data field. Here is some output from a Ferret session. The variables are GEOLON, GEOLAT, and ZVAR.  These are all on a 180x173 point grid with axes XAX (i=1 to 180) and YAX (j = 1 to 173)    yes? use lonlat.nc      yes? show data         currently SET data sets:        1> ./lonlat.nc  (default)     name     title                     I      J      K      L     GEOLON   geographic longitude     1:180  1:173  ...    ...     GEOLAT   geographic latitude      1:180  1:173  ...    ...     ZVAR     variable on the grid     1:180  1:173  ...    ...       yes? show grid geolon        GRID GMZ1     name       axis              # pts   start                end     XAX       X                  180 r   1                    180     YAX       Y                  173 r   1                    173     normal    Z     normal    T      ! Visualize the lon,lat curvilinear grid    yes? set view upper    yes? plot/vs/sym=dot/j=140:173 geolon, geolat      ! Shade plot of a variable on the grid    yes? set view lower    yes? shade/j=140:173 ztp, geolon, geolat   Mario Germano wrote: ```Hi everybody, I am experiencing difficulties in inserting grid points in a cdf file (using FORTRAN) because these points have all different lon-lat coordinates: they are not uniformly distributed on an orthogonal grid, but rather on a rotated grid. Ideally I'd like to be able to insert points one by one using a syntax like: insert_point(lon,lat,wind_speed, temperature,...) I ignore if such a thing is possible. Thanks in advance for answering my question. Mario. ```