Skip to content
Prev 21064 / 29559 Next

Create a simple spatial grid

On Tue, May 20, 2014 at 5:41 PM, Remi Genevest <rgenevest at free.fr> wrote:
Step 1 is "buy a bigger computer". Europe is about 3,000km EW and NS
(wikipedia gives the area as about 10,000,000 square km - that might
be the land area? Anyway, orders of magnitude...), which is 150,000 of
your 20metre cells by 150,000. Or twice as many if that was just the
land area. That's 22 billion cells. And you haven't put any data in
them yet.

 Step 2 is the tricky step. You can't create what's commonly thought
of as a 'regular grid' on a sphere. You can make each cell centre 20m
from its nearest neighbours (I think, at least until you run up
against the pole), but it won't be 40m from its next-nearest
neighbours when measured along the great circle distance. You can use
a coordinate system other than WGS84, and make a 20m spaced-out set of
cells, but the distances computed along the sphere won't be right.

 There are some standard map projections that european-wide data seems
to use, mostly based on Lambert conformal conical projections. You
might want to copy the grid used for CORINE data:

http://www.eea.europa.eu/data-and-maps/data/corine-land-cover-2000-raster-3

 - which goes to 100x100m resolution (or 250x250 if you don't want to
have to buy a new PC to load the whole continent...).

Barry