Wrap around longitudes?

[milankl]

Not sure why this only popped up in a nc file created by SpeedyWeather but the first and last longitudes aren't connected in a periodic fashion. I can live with the gap at the pole ... I know ERA5 just averages the northernmost latitude ring and writes that data as an additional ring at 90˚N (which seems like a waste though tbf)

[TheJeran]

Depends on the extent. The texture is scaled based on the extent and resolution and just samples the texture based on the scaling.

Can you take a picture of how Browzarr calculated the extent?

settings --> adjust extent (at bottom)

[milankl]

I guess it's all because we do lat north to south and longitude 0 to 360?

[TheJeran]

No that's not the issue. I gotta revisit the logic. I thought I coded it to treat the min and max values as the centers of the pixel. But I think maybe it's treating it as the edge of the canvas since a lot of the datasets I work with have extents -180 to 180. Which doesn't make sense if they were the centers since it would have a duplicate column.

I'm fairly certain that's whats happening here and if that's the case then the 3.75 resolution wouldn't be enough to close the gap. You would need to double it to 7.5. (The resolution is just the edge padding. Doubling it won't break the sampling)

I will revisit this because I don't think it's the correct behavior.

[milankl]

Yes our data, if at all resolution of dx is would for example run from 0 to 360-dx or for other grids go from dx/2 to 360-dx/2.

Similar in latitude. The spacing may be uneven but the first and last row of cells (we say ring) extend to the pole so there's always a vertex at the poles but never a centre.