Hi Patrick, I think we could probably do this but first I want to understand more about your grid. Is your grid actually curvilinear in longitude-time? Or are you thinking of plots using the 2-dimensional longitude coordinates from a grid that is curvilinear in xy and a time variable that is somehow generated? If the curvilinear nature of the grid is in the horizontal dimensions, then to make a Hovmoller plot in X-T, I think you'd want to use the curvilinear sampling function, SAMPLEXY_CURV, to get a subset of the data at a particular latitude, and then plot that. So if your data looks like yes? show datayou could define some longitude and latitude pairs to sample: let xpts = x[x=2:360:2]This would return U sampled at each of the longitude-latitude locations (2,31), (4,31) , ... (360,31), for all times on the time axis of U. The result of the function is on an abstract axis in the X direction, so you would need to define the equivalent longitude axis, and put u_at_31N onto that axis, and then you could plot it as a longitude-time plot with the regular 1-argument plot commands. Ansley Brockmann Patrick wrote: Hi all, |