# GIMMS NDVI3g grid cells not entirely square-shaped?

The official data description of the recently updated GIMMS NDVI3g record describes the grid parameters of the dataset as follows.

grid-name: Geographic Lat/Lon
pixel-size: 1/12=0.0833 degrees

size-x: 4320
size-y: 2160

upper-left-lat: 90.0-1/24
upper-left-lon: -180.0+1/24
lower-right-lat: -90.0+1/24
lower-right-lon: 180.0-1/24


This means (at least to my understanding) that in x-direction, the cell size equals

> (180-1/24) * 2 / 4320
[1] 0.08331404


whereas in y-direction, the cell size equals

> (90-1/24) * 2 / 2160
[1] 0.08329475


I.e., the single grid cells are not entirely square-shaped. Am I missing something in my calculation or does the officially described pixel size of "1/12=0.0833 degrees" simply result from rounding effects of the manually calculated edge lengths in x- and y-direction?

Your understanding is inaccurate. The shifts by $\frac{1^\circ}{24}$ are only relevant for figuring out, where the pixel centers are located and where to start when you are looking to find the right grid index for a certain point. As for the resolution, it is much simpler. We have $4320$ pixels covering $360^\circ$ in longitude direction, i.e. a resolution of $\frac{1^\circ}{12}$ and similarly for the latitudinal resolution.