Message-ID: <51A759B7.5030202@gmail.com>
Date: 2013-05-30T13:52:55Z
From: Jeroen Steenbeek
Subject: Kriging and datum shifts
In-Reply-To: <51A4CA1D.6050506@uni-muenster.de>
Yes Edzer, I was not using terminology accurately. Apologies, R is new
to me.
Let me dig deeper if this simple example is needed. I will make sure if
gstat kriging methods correctly use great circle distances correctly.
Jeroen
On 2013-05-28 12:15 PM, Edzer Pebesma wrote:
> Jeroen, sp does not have anything to do with kriging, so your comment is
> confusing.
>
> sp does create spatial objects from data.frame's, and may do some sanity
> checks on coordinates in case they are longlat -- you have to specify
> they are longlat, though.
>
> If they are specified as longlat, and e.g. gstat::idw is used for
> interpolation, it should use great circle distances as opposed to
> Euclidian distances to figure out interpolation weights. If this is not
> the case, I'd like to look at a simple, reproducible example. Kriging on
> spherical data is done at your own risk, there are no checks that the
> variogram model can be applied sanely on a sphere, as reported before.
>
>