Talk:Gaussian grid

This article is rated Start-class on Wikipedia's content assessment scale.
It is of interest to the following WikiProjects:

Maps High‑importance

	This article is within the scope of WikiProject Maps, a collaborative effort to improve the coverage of Maps and Cartography on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.MapsWikipedia:WikiProject MapsTemplate:WikiProject MapsMaps
High	This article has been rated as High-importance on the project's importance scale.
	This article had an importance automatically assigned to it. Once it has been checked by a human, please remove `\|autoi=yes`.

Naming conventions

It's wrong to speak of a "T62 Gaussian grid" or similar. T62 describes the truncation of spherical harmonics, which have no relation to Gaussian grids except that they are often used together in numerical weather prediction or climate models (spherical harmonics for atmospheric physics, Gaussian grids for surface physics and parameterizations). There may be advantages to use certain types of Gaussian grids together with a given truncation, but there's no 1:1 relation. A Gaussian grid is defined by the numbers of latitudinal and longitudinal grid points alone. And the vertical resolution of a model has no relation to Gaussian grids at all. --89.55.4.99 (talk) 21:00, 27 September 2008 (UTC)[reply]

sample points

“

The gridpoints along each longitude (meridian) are equally spaced, while they are unequally spaced along each latitude (parallel), where they are defined by their Gaussian quadrature. There are no grid points at the poles. By contrast, in the "normal" geographic latitude-longitude grid, gridpoints are equally spaced along both latitudes and longitudes

”

Huh? Gaussian quadrature is a way of adjusting the sample points for numerical integration; what's the function to be integrated along a line of latitude? —Tamfang (talk) 00:32, 7 July 2012 (UTC)[reply]

This is really confusing. The diagram appears to show an ordinary (naive) lat/lon grid (specifically, it looks like a regular rectangular grid in the plate carree projection). It exhibits the typical problem that the poles are oversampled, because when points are equispaced in longitude coordinate they are not equispaced in real distance on the globe (because the meridians converge toward the poles). This seems like the simplest scheme one could possibly choose, optimised for convenience not numerical efficiency. What is Gaussian about it?!? Wouldn't Gaussian quadrature produce irregular spacings, rather than fixing regular increments for each coordinate (lat and lon)?

Moreover, directly contrary to the quoted text, the image shows points that are equally (not unequally) spaced along each parallel. Was the article phrasing intended to convey that the spacing differs between different parallels? Or is the diagram inappropriate?

Maybe the reference to Gaussian Quadrature merely denotes that the constant grid spacing along a meridian (between longitude values of 0° and ±90°) is chosen to span one grid half-unit less than 90° (to avoid having an entire "row" of points located literally on the pole where the coordinate system is degenerate)? This is similar to how Gaussian Quadrature apparently may involve samples with regular spacing that spans slightly less than the full domain? Cesiumfrog (talk) 12:13, 6 March 2026 (UTC)[reply]

Terminology

“

In a regular Gaussian grid, the number of gridpoints along the longitudes is constant, usually double the number along the latitudes. In a reduced (or thinned) Gaussian grid, the number of gridpoints in the rows decreases towards the poles, which keeps the gridpoint separation approximately constant across the sphere.

”

What is "along the longitudes"? Does it mean north/south along a meridian (e.g. increasing longitude), or east/west along a parallel (e.g. constant longitude)?

Also, what are "the rows"? Are the parallels rows and the meridians columns? (Or are there rows of points in all directions?)

Article really needs to rely more on standard geographic terminology (meridians and parallels) and to (at least in passing) give a definition for any other jargon it uses.

Also the image does not match the description of "approximately constant" gridpoint separation. In projection space, the separation looks to be exactly constant. On the globe itself, the separation is not even remotely constant (the sampling density varies immensely between the equator and the poles). I think part of the problem is that the article isn't clear whether it is speaking in a projected coordinate space (rather than the surface in 3D) or referring to only one dimension, etc. Cesiumfrog (talk) 11:32, 6 March 2026 (UTC)[reply]