hi list, thanks to all who replied to my question regarding stratified kriging posted on august 4, 2003. the original question was:
I have question regarding stratified (ordinary, simple, simple updating, etc.) kriging. when dividing my entire data set in different strata based on additional information like geographical classification of natural landscapes, soil maps, aquifer or watersheds (or whatever is suitable and justifed) and each subset is modelled and interpolated seperately, how do I have to handle the effect that estimation points close to the boundary of each polygon have most of their supporting values defined by the search strategy in the adjacent polygon? applying a buffer around each dividing polygon to get the observations close to the boundaries and clip the interpolation results? would this be an option when using additional information that rarely has sharp boundaries in reality like soils? many many thanks in advance. I got the following answers on this: from Yetta Jaeger: On my web site is a Pattern-plus report in which we describe normalizing by the sill (variance) of each stratum, using a normalized common model for the normalized data (have to use same model form) and kriging the normalized values. Yetta from Ralf Stosius: I'm working on the same problem but in the context of glaciological data. Yesterday I found the following article, which might be very helpful for you. Boucneau, G., Van Meirvenne, M., Thas, O. und Hofman, G. (1998): Integrating Properties of Soil Map Delineation into Ordinary Kriging .- European Journal of Soil Science, 49, 213-229 from Pierre Goovaerts: You are right that such stratified kriging will create discontinuities close to the boundaries and if it does not make sense you can always perform the kriging on the residuals and add back the stratified means. An example is given in the paper you can download from http://www.terraseer.com/services/courses/geostats/geoderma.pdf from Paulo Justiniano Ribeiro You can approach this by including all these information (soil type etc) as covariates in your analysis. The covariates would act as what is called "external trend" in the geostatistical literature. For instance the data set ca20 (calcium content in the 0-20 cm soil layer) in the package geoR (www.est.ufpr.br/geoR) has 2 covariates: region (3 soil management) and altitude. The help file for the data set has further information and exemples. from Isobel Clark: You can approach this by including all these information (soil type etc) as covariates in your analysis. The covariates would act as what is called "external trend" in the geostatistical literature. For instance the data set ca20 (calcium content in the 0-20 cm soil layer) in the package geoR (www.est.ufpr.br/geoR) has 2 covariates: region (3 soil management) and altitude. The help file for the data set has further information and exemples. from Hannes Reuther: Gehts auch auf deutsch ? ISATIS ( www.geovariances.fr) has a functionwhich allows to use FAULTS ( wheter they are boundaries of landforms, soil units or whatervers to be included in a stratified kriging.. i'm not sure that gslib can do. regards, oliver -- * To post a message to the list, send it to [EMAIL PROTECTED] * As a general service to the users, please remember to post a summary of any useful responses to your questions. * To unsubscribe, send an email to [EMAIL PROTECTED] with no subject and "unsubscribe ai-geostats" followed by "end" on the next line in the message body. DO NOT SEND Subscribe/Unsubscribe requests to the list * Support to the list is provided at http://www.ai-geostats.org
