Isobel,
I keep in mind at all times that wonderful
sentence in Clark (1979), "Systematic errors, be they sampling , analytical,
or whatever, will not be picked up by geostatistics and will be
transferred to any estimates produced." (p. 119)
Isobel, this wisdom should be included
in the disclaimer of any geostatistical report. If one looks at the lead
paper of the Australian Best Practices Volume (proper citation not available
because I have misplaced the book). More than 50% of gold deposits
fail to achieve nameplate output in the first year of production due to
grade shortfall (David Harquail, Graham Clow and and Australian masters
thesis). Then things improve, not from statistics, but using fudge
factors(mine call factor, cutting, whatever).
Mike Armitage, Managing Director of SRK, is on
record in the SRK newsletter stating that the predictability of geostatistics
is no better today than at any time in the past. I took issue with Danie
Krige in the SAJG for saying, in 2001, the framework of geostatistics included
all the necessary elements and no further discoveries need be made.
In my experience, the issue is sampling,
splitting, analysis, and execution. By the time a geological sample gets
to a 30 gram aliquot, the likelihood of that 30 grams representing 1,000,000
grams of rock is probably compromised about 50% of the time. So we
respond, in ignorance, by taking thousands of samples. In my
opinion, 33 samples will suffice in am stope, giving two extra sample for more
confidence. ... Need to take large enough samples to be representative
though. Got to avoid the systematic error of not having a representative
sample for the region, deposit, stope, etc.
I cringe when I see geostatistics applied to
ground water projects or remediation projects, or astrophysics. Ask
yourself whether asteroid size distribution as sampled might suffer a
'nugget effect'.
Francis T. Manns, Ph.D., P.Geo.
(Ont)
Artesian Geological Research
Toronto, Ontario
And the following sense from Jim Tilsley also
need to be integrated....
Fran,
Questioning geo-stats is an advancement in
understanding. This is good for everyone and everything
involved.
The point most overlooked in the 'equation' is
the first half of the term. GEO!
I, as you know, have always objected to
statistics being the main focus. FIRST do the geology. Then we may
find that some sort of statistical treatment can be applied
rationally.
As we have discussed before on many occasions,
Kreiging works wonderfully on mature gold placers of the Wits. The
trouble comes in when attempts are made to apply it to another type of
mineralized zone, one that has not been deposited under the same energy regime
as the Wits.
There can be a relationship between grade - a
parameter fraught with difficulties in the details - and relative position in
respect to another grade, only when there is some sort of energy gradient that
ties the two together - as in the paleo-placers for which the approach was
developed. Since the energy budget correlates with the distance from the
centre of the channel, decreasing toward the limits of flow - (Hydraulics 101)
detrital heavy mineral concentration can be expected to reflect the mechanical
energy applied, and some sort of distance/grade relationship is
plausible.
(The Carbon Leader gold mineralization is another
matter, but if the algae mats acted as chemical traps and there was some
difference in total flow (volume) from a 'channel' axis across the mats, one
could expect to see a similar distribution of values, but the geological
controls and the difference in depositional environment must be
considered.)
Statistics, in my experience, may be applicable
in ore reserve estimation, (and will solve many problems if properly applied)
providing they are used with due reference to the geological controls on
deposition. And this applies to mechanical and chemical mineralization
systems equally. The chemical systems are usually the more difficult to
deal with.
Jim
----- Original Message -----
Sent: Wednesday, February 08, 2006
10:07 AM
Subject: [ai-geostats] Re: Geostats
Scam?
And don't forget the 27 page article by Philip & Watson
"Matheronian geostatistics - Quo Vadis?" in Mathematical Geology, vol 18, pp
93-117. 24 pages on what a confidence trick geostatistics is.
There are response letters in that volume and later volumes from
members of the geostatistical community. A wonderful example of neither side
listening. Bear in mind when you read this article that it was not refereed
before publication.
In fact, this is a very good paper to illustrate the enormous
difference which can occur when you approach a problem from a completely
different direction.
What baffled me for a long time is that their argument is that where
you have more data, you have more variability and therefore less confidence
in your estimates. This seemed to me, set in my ways as a classical
statistician as counter-intuitive. Some years later I was discussing a
possible geostatistics study with a colleague who was a land surveyor when
the penny dropped! P&W are geographers. They can SEE the surface they
are trying to map. So they MAKE more observations where the land surface is
more variable. Completely logical. Where they have fewer observations, it is
because the surface is flat and they are confident in their estimates. And
vice versa.
Matheron developed geostatistics, just as econometricians and
statisticians developed time series analysis -- for surfaces and series
which cannot be seen. More data does not mean less confidence in local
variability, it just means more data.
As you said, Steve, if geostatistics does not make sense for your
application or the results don't make sense in the context of you data,
there are lots of other things you can do. And, if you are happy with using
geostatistics, keep it simple!
Isobel
http://www.kriging.com/courses
Stephen Henley
<[EMAIL PROTECTED]> wrote:
Hi
Mach -
Jan Merks isn't the only one who has questioned the
fundamentals of
geostatistics. You might also like to look at the work
of Robert F. Shurtz.
Perhaps the most interesting is a slim volume
published by the Mining &
Metallurgical Society of America, "A
Geostatistical Monograph" (in 1997, I
believe, though the book isn't
dated). This is in the form of a debate
between Robert Shurtz
(criticising geostats) and Harry Parker (defending
geostats).
If you want alternatives to geostats, there are
plenty on offer. Problem is,
of course, that after thirty-something
years geostatistics has become the
'accepted wisdom' in the mining
community, and you have an uphill struggle
to justify using any other
method. Unfortunately, also, any geostatistical
practitioner who uses
a fancy new nonlinear kriging method has much less of
a struggle even
if that new method is little better than alchemy. Linear
kriging does
actually have some nice mathematical properties if its
assumptions are
satisfied and when used properly ('best linear unbiased
estimator' and
all that) but what many users seem to forget is that, as soon
as they
tweak the method, they lose some or all of these properties. In my
opinion many of the more exotic variants of kriging are worse than
using
non-geostatistical methods because they encourage - wrongly - a
belief that
they are in some ways superior simply because they are
known as "xxxxx
kriging".
- Steve
Henley
================================
Dr Stephen
Henley
Resources Computing International Ltd
185 Starkholmes
Road
Matlock, Derbyshire DE4 5JA, UK
+44(0)1629 581454
Skype:
silicondale
[EMAIL PROTECTED]
================================
-----
Original Message -----
From: "Mach Nife" <[EMAIL PROTECTED]>
To:
Sent: Tuesday, February 07, 2006 6:58
PM
Subject: [ai-geostats] Geostats Scam?
>
Hi,
>
> This guy (Jan W Merks) seems to devote his life
trying
> to prove that Geostatistics is a scam... If it's
true
> I'm gonna start looking for alternatives.
>
>
http://www.geostatscam.com
>
> Anyone has a point of vue on
this?
>
> machnife
>
>
__________________________________________________
> Do You
Yahoo!?
> Tired of spam? Yahoo! Mail has the best spam protection
around
>
http://mail.yahoo.com
>
>
--------------------------------------------------------------------------------
>*
By using the ai-geostats mailing list you agree to follow its
rules
> ( see http://www.ai-geostats.org/help_ai-geostats.htm
)
>
> * To unsubscribe to ai-geostats, send the following in
the subject or in
> the body (plain text format) of an email
message to [EMAIL PROTECTED]
>
> Signoff
ai-geostats
--------------------------------------------------------------------------------
No
virus found in this incoming message.
Checked by AVG Free
Edition.
Version: 7.1.375 / Virus Database: 267.15.2/253 - Release
Date: 07/02/2006
* By using the ai-geostats mailing list you
agree to follow its rules
( see
http://www.ai-geostats.org/help_ai-geostats.htm )
* To unsubscribe
to ai-geostats, send the following in the subject or in the body (plain
text format) of an email message to [EMAIL PROTECTED]
Signoff
ai-geostats
* By using the ai-geostats mailing list you agree to follow its rules
( see http://www.ai-geostats.org/help_ai-geostats.htm )
* To
unsubscribe to ai-geostats, send the following in the subject or in the body
(plain text format) of an email message to [EMAIL PROTECTED]
Signoff
ai-geostats