regression kriging with non-euclidean distances

Don & Facundo,

This is something that I'm interested in as well for coastal marine
applications.

The only solution I've found so far is a cumbersome "barrier" function
available in the kriging operation of ArcGIS. I would be interested in
seeing an (better) implementation of the idea in R.

I was able to find a few references on the subject that I pass along to you
and anyone else that might find it useful:

Krioruchko, K. and A. Gribov. 2002. Geostatistical interpolation and
simulation in the presence of barriers. X. Sanchez-Vila et al. (eds.), *geoENV
IV -- Geostatistics for Environmental Applications,* pp. 331-342. Kluwer
Academic Publishers: the Netherlands.

Curriero, F.C. 1997. The use of non-Euclidean distances in geostatistics,
PhD thesis.

Little, LS, D Edwards and DE Porter. 1997. Kriging in estuaries: as the crow
flies or as the fish swims? Journal of experimental marine biology and
ecology. 213:1-11.

Rathbun, S. 1998. Spatial modeling in irregularly shaped regions: kriging
estuaries. Envirometrics. 9:109-129.

Jensen, OP, MC Christman and TJ Mller. 2006. Landscape-based geostatistics:
a case study of the distribution of blue crab in Chesapeake Bay.
Envirometrics. 17:605-621.

Regards,

Alisa Coffin


On Thu, Mar 20, 2008 at 2:42 PM, Don MacQueen <macq at llnl.gov> wrote:

  

Facundo,

I am interested in essentially the same thing,
though my application is inside buildings. I am
wondering what technique(s), software, etc. you
would be using to calculate your distances.

I asked about this about a month ago, here on
r-sig-geo, and received some helpful replies. The
subject line at that time was
   [R-sig-Geo] Shortest path around obstacles (OT)

-Don

At 7:05 PM +0100 3/20/08, Facundo Mu?oz wrote:
    

Hello All,

I'm performing  regression kriging (using both gstat and geoR)  in an
urban environment. Thus, i have discontinuities (buildings) in the
prediction space which make distances to be non-euclidean.

I need to estimate variogram and make predictions using distances
calculated by myself and passed to the algorithms (say, in a distance
matrix), instead of letting them calculate distances based on
      

coordinates.
    

?Has anyone faced this issue before?

Thanks in advance,
    Facundo Mu?oz

--
Departament d'Estad?stica i Investigaci? Operativa Universitat de
Val?ncia (Estudi General)
Facultat de Matem?tiques, Dr. Moliner 50, 46100 Burjassot, Spain.
(+34) 96 354 3987, fax: (+34) 96 354 3238
e-mail: Facundo.Munoz at uv.es, web: http://www.uv.es/~famarmu/<http://www.uv.es/%7Efamarmu/>

--
--------------------------------------
Don MacQueen
Environmental Protection Department
Lawrence Livermore National Laboratory
Livermore, CA, USA
925-423-1062