- [Voiceover] Let's briefly talk about core data. It's obviously important to verify core measurements with logs. Sometimes, there's an argument as to what kind of emphasis should you put on cores. Should you force the logs to agree with the cores, and the answer is, no you shouldn't, because there are a number of caveats if you can't correlate. And so let's look at some of those. First of all, it's a volume problem. Core samples are small. The logs are measuring at least a cubic foot, and probably tens or hundreds of cubic feet. And so, there's a scalar problem, a minimum of a thousand to one, and maybe as high as a million to one. So to expect a correlation between logs and cores may be dreaming. If you get one, it might be just serendipity, rather than real. Let's look at some suggestions that I have. Core water saturation might be affected by mud filtrate, and by expulsion from the core due to gas expansion from the oil. And that's what I think probably happened in that Nye blower case, where the water saturations are somewhat lower than petrophysics. And it could be just that the water's being kicked out as the core was being brought to surface. Apparent grain densities from cores will be reduced in the presence of gas. So you can't necessarily correlate, and say, well I've got to adjust the petrophysics to agree with the cause, that may not be true. We've spent quite a bit of time addressing the issue of upscaling. It's been around for a long time, upscaling core data to try and get it to agree as best you can with logs. And so, what we do is we take a moving window of different levels, and give major emphasis to what is in the middle, and diminishing emphasis to what is on the ends. It's remarkable what comes out of that. So let's look at the next slide. What's the use of this? Well, we'll see in just a moment. It's used to better define core shifts, as compared with raw data, to derive better core/log comparisons, and we'll see that getting better in some cross plots. Also, any of you that are involved in equity studies that involve calibration of cores to logs, be extraordinarily careful of, is raw data being used, or upscaled data being used? We have shown, in a theoretical case, that it affects equity by 20, 25%. So just be careful of that. Okay, here's an example from the Panhandle field, and we're showing on the left raw data, upscaled core data in the middle, on the right, upscaled and shifted. Put a circle around the upscale data, but not shifted, you can see a little bit of white appearing. On the upscale data, those little bits of white have disappeared totally. They're subtle shifts, but they're real. Then, on the raw data, there, bad grammar, there ain't no way you would see those core shifts from the raw data. It's impossible. We strongly recommend upscaling of core data. Of course, it only works if you've got continuous cores. It won't work on sidewall cores. Here's an example of permeability comparisons. The horizontal axis is core data, the vertical axis is log equivalence with permeability. The raw data, you can see a lot of scatter, R squared of .25. The upscaled data is still scattered. You can see that a lot of that scatter has disappeared, and R squared is .54. We now go to the porosity. Again, same deal, raw data, quite a bit of scatter. R squared of .39. And the upscaled data has reduced it at least to some degree, and the R squared has gone up significantly to .65. So, upscaling the core data we think is important.