Hi Abani, What you say is quite correct, but it depends on your criteria for differentiation between measured and indicated resources. As a rough rule of thumb we take the lag distance at two thirds of the of the variogram value (minus nugget) as being a good search distance from a minimum permissable number of samples to be a good guide to decide whether a block is measured or indicated. If one has calculated the kriging variance at this point, one can then define all the blocks (or estimated points) with less than this kriging variance as measured, and those with greater kriging variance value as indicated. However generally one cannot take indicated resources beyond a search distance greater than the range of the variogram, because beyond that point all samples get equal weighting just as in using the arithmetic average and pairs have no corralation. The beauty of using the kriging variance contours to define the limits of the categories is that it mimics the anisotropy of the variography smoothly, whereas on number of samples alone, the shape of the sampling grid tends to be the only governing criteria. Maybe to study your two cases one should manufacture the equivalent experimental variogram model of your correlogram and then model it and see if the two cut off points for measured resources differ. Seeing as your measured resources for the second model were greater in extent, I would say that the lag distance point of the cut off would be longer. One should follow this up on a dummy data base of actuals, to see if your cut off point for measured was reasonable (One could regress the Estimates calculated from a sparse grid with the actuals). Points or blocks estimated from samples at distances beyond the range of the variogram are usually categorised as inferred. Summarising, because your variogram range is shorter, your total measured plus indicated resources will be less in the second model, but your measured resources will be more. Further sometimes the variogram model of a longer range model starts off increasing at a steeper rate than the short range model only to flatten out and start increasing at a lower rate than the short range model (ie anisotropically speaking). I am still looking for examples to explain the reason for this, but suspect it is anisotropy direction at short distances between pairs of samples is over ruled by the anosotropy direction at greater distances between pairs of samples due to some phenominum or process in nature. One can follow this up by creating a regularised grid of the random point samples and see what happens to the anisotropy of the variogram model made from this. Another thought is that your second model could have been less anisotropic than your first (ie more elliptical search radius) and therefore more measured estimates were created. Hope this helps. I am not sure whether my first communication came through, so I am resending this with cc AI-geostats. I think I am having trouble with your address, so I am hoping that AI-geostats sends this on to you. Best regards Bill Northrop
-----Original Message----- From: Abani R Samal [mailto:[EMAIL PROTECTED] Sent: Tuesday, May 29, 2007 8:57 PM To: Bill Northrop Subject: Re: AI-GEOSTATS: Block model differences Hi Bill, My variogram ranges are "shorter" than the earlier correlogram models. I think in my case the variogram values increase at a higher rate than the older model. Isn't that right? If thas true then the samples near the block centers (within the range) should get higher weights than the samples farther from the block centers. Bill, please correct me if I am wrong. Abani ************************************************************ ----- Original Message ---- From: Bill Northrop <[EMAIL PROTECTED]> To: Abani R Samal <[EMAIL PROTECTED]> Sent: Tuesday, May 29, 2007 8:46:12 AM Subject: RE: AI-GEOSTATS: Block model differences Hi Abani, With regards to your "secondly statement", I would say that you are obtaining more measured estimates because your variogram values now increase at a lower rate with increasing lag, thus giving more weight to samples further away than in your previous model.If this is not the case then we must think again. Regards Bill Northrop -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Abani R Samal Sent: Thursday, May 24, 2007 7:42 PM To: [email protected] Subject: AI-GEOSTATS: Block model differences Dear List, I received a block model which used a correlogram model to krig. The ore-body strikes approx. N35E, dips approx. 35 degrees in NW direction. I modeled the variograms and found the followings: 1: My directions of anisotroy are along the inclined ore-body (the earlier correlograms were not along the ore body). 2: My ranges are shorter than the earlier correlogram ranges. earlier correlograms had two structures, which I don't see in my variograms. The likely reason for this is that the earlier person doing correlogram model did not keep its sill below a standard variance/ correlogram line. In my case I kept my variogram sills below the variance line (using ISATIS). 3: My search ellipsoid had same dimensions as earlier search ellipsoid, but my search ellipsoid oriented along the ore-body, where as the earlier one did not. 4: The measured and indicated resources are categorized based on distance (from block center to sample) and min.-max. number of samples used for interpolation of the block. I am getting approx. 30% of more resources in the measured category blocks. I am needing a valid explanation for this: I think, because of the re-orientation of my search ellipsoid (along the ore body), I am able to find more blocks meeting the minimum sample criteria for estimation (than the earlier model): Is this a valid reason? Secondly, also I think as my variograms are having shorter ranges, I am allowing more blocks to be estimated from nearest samples than the earlier model: Is this right? I'll highly appreciate your valuable comments/ suggestions. Regards, Abani R Samal ************************************************************ ABANI RANJAN SAMAL 11183 West 17th Avenue, APt 201 Lakewood, CO 80215 http://myprofile.cos.com/arsamal _____ Moody friends. Drama queens. Your life? Nope! - their life, your story. Play Sims Stories at Yahoo! Games. <http://us.rd.yahoo.com/evt=48224/*http://sims.yahoo.com/> _____ Don't get soaked. Take a <http://tools.search.yahoo.com/shortcuts/?fr=oni_on_mail&#news> quick peak at the forecast with the Yahoo! <http://tools.search.yahoo.com/shortcuts/?fr=oni_on_mail&#news> Search weather shortcut.
