Loading ...
Sorry, an error occurred while loading the content.

[ai-geostats] Hugly reality

Expand Messages
  • marenostrum@libero.it
    Dear Geostat I m a beginner and I had the unliky case of a tridimensional multivariate geochemical dataset to start my geostatistical adventure...moreover data
    Message 1 of 1 , Nov 3, 2004
    • 0 Attachment
      Dear Geostat
      I'm a beginner and I had the unliky case of a tridimensional multivariate geochemical dataset to start my geostatistical adventure...moreover data are taken from sediment samples of a high polluted port and present a high variability...
      My question is:
      histogram of variables doesn't reveal nor a gaussian distribution neither a lognormal one, that is that variables are not well distributed on any of these distributions...do I have to force to find some distribution that could well represent my data? The condition based on the fact that Kriging performs well on gaussian variables is so constraining? Or maybe the main thing is the shape of variogram?...Do I have to try to lay variables back to gaussian shape or I can direct my strengths on studying and modelling variogram?...And if so, if variogram are unbounded, do I have to consider variables in a non stationary framework or it's possible that such a bad distribution could false the shape of the variogram? (For example for log transformed variable, even if histogram do not get better, some variograms are better shaped and not so unbounded)...
      I know I'm confused but I need some practical suggests to come from all this good and perfect theory to my hugly reality...
      Thanks
      Simone
    Your message has been successfully submitted and would be delivered to recipients shortly.