[R-sig-Geo] Correspondence between gstat and ArcGIS functionality?

[Edzer Pebesma]

Hi,

telling us what the drop-down menus tell you leaves me a bit of guess
work. Universal probably means universal kriging; linear drift a first
order trend in the coordinates (pass degree = 1 to krige, and something
like value~x+y to variogram, when long/lat are named x and y). Does the
first "Linear" in "Linear with linear drift" refer to a fitted linear
variogram model?

Prof. Jeffrey Cardille wrote:
> Hello all,
> 
> This is my first posting to R-sig-geo; I looked for the answer in the archives, but have not had any luck for my particular question. 
> 
> I have a newbie question about the correspondence between gstat and ArcGIS.  I have a problem that gives very very satisfying results in ArcGIS, and I would like to write a public function in R for others to use that does the same thing.  The choice I used in Arc was "Universal" and in the dropdown menu it says "Linear with Linear Drift".  My data matrix is pretty standard:  2000 x 2000, with some NA, and values between 0 and 255.
> 
> So I have a few questions.  Maybe I'll enumerate them.
> 1. Has anyone made a comprehensive list between ArcGIS functionality and how to do it in gstat?

Great idea!

> 2. If not: does anyone know a simple correspondence between what ArcGIS is doing and how to set up a gstat call for "Universal/Linear with Linear Drift"?  I have looked online for precise details about Arc's behavior, but haven't seen anything detailed enough.

Well, you have a license, doesn't that come with access to some sort of
documentation?
> 
> 3. Are there any tricks in gstat that make interpolation with gstat especially fast or slow?  For example, if the numbers were treated faster if they were between 0 and 1, or if NA should be recoded, or if I should never use a data frame but always a matrix.  That sort of thing.
>
Not the things you mention; defining a local neighbourhood matters --
small neighbourhoods are faster, but at some stage also produce worse
interpolations.

For filling in missing strips in images I'd try to use segmented
neighbourhoods, to guarantee values are taken from all sides. Quadrant
search is available in the gstat C code, but not interfaced through the
R package. I will implement it if you volunteer to test it.

> If not, I'll resort to trial and error and diving deeper into the gstat documentation.  But it seems like a good idea to check in first.  For anyone interested, I am trying to repair Landsat satellite photos, which have big broken strips of nodata due to a mechanical failure.  The strips are up to 14 pixels wide.  I need to do this 2000x2000 interpolation about 2000 times-- so speed differences of even a few seconds or minutes are quite important..

I'm interested in your findings!

BTW, just out of curiosity, why don't you do the whole thing with ArcGIS?

> 
> Thanks!
> Jeff
> 
> ------------------------------------------
> Prof. Jeffrey Cardille
> jeffrey.cardille at umontreal.ca
> 
> ******************************************************************************************
> **  http://www.geog.umontreal.ca :                Web                                   **
> **  http://sites.google.com/site/jeffcardille :   Plans de Cours et Recherche           **
> **                                                                                      **
> **  "Petit à petit l'oiseau fait son nid ..."    N'hésitez pas à corriger mon français  **
> *********************************************************************************************************
> **     Département de Géographie                        **     Bureau:                                 **
> **     assistant professor / professeur adjoint         **     Salle 440                               **
> **     Université de Montréal                           **     Pavillon Strathcona                     **
> **     C.P. 6128                                        **     520, chemin de la Côte-Ste-Catherine    **
> **     Succursale Centre-ville                          **     Montreal, QC H2V 2B8                    **
> **     Montréal, QC, H3C 3J7                            **     Télé: (514) 343-8003                    **
> *********************************************************************************************************
> **
> **  Pour aider à Haiti avec votre intérêt de géographie, visitez:
> **  http://groups.google.com/group/udem-geo-aide/
> **
> *********************************************************************************************************
> 
> 
> 
> 
> 
> 	[[alternative HTML version deleted]]
> 
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo

-- 
Edzer Pebesma
Institute for Geoinformatics (ifgi), University of Münster
Weseler Straße 253, 48151 Münster, Germany. Phone: +49 251
8333081, Fax: +49 251 8339763  http://ifgi.uni-muenster.de
http://www.52north.org/geostatistics      e.pebesma at wwu.de