## GEOSTATS: cdf and ccdf

Expand Messages
• Dear all, I have some questions on the difference between the cumulative distribution function (cdf) and the conditional cumulative distribution function
Message 1 of 1 , May 18, 1999
Dear all,

I have some questions on the difference between the cumulative
distribution function (cdf) and the conditional cumulative
distribution function (ccdf).

I am not sure when I do infer the cdf and when do I infer the
ccdf by using various geostatistical tools.

If I want to infer the cdf of my continous variable Z at location u
(Z(u)) I have to calculate the sample histogram of all samples z(u_n,
n=1,...,n). Then I can determine the cdf-mean (expected value) which
is the sample mean (possibly the declustered mean)? And the a priori
covariance/variance is my cdf-covariance/variance?

When I use kriging I get the ccdf-mean (?) as an least squared error
estimate which is my kriging estimate z*(u) (using the neighbouring
sample values z(u_n)). The ccdf-covariance (?) is inferred through
the covariance/variogram model that is specified before.

In contrast, indicator kriging provides the whole ccdf (?) at an
unsampled location u inferred through the kriging of the transformed
sample values around.

So in both approaches
1) I use the neighbouring data (to infer the ccdf)?
2) in the first case I only infer the first two moments of the
ccdf (mean, covariance), in the indicator approach the whole ccdf is
inferred?

I would be grateful for any comments to see if I am on the right or
on the wrong way.

Thanks so far.

Ulrich

><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><><>

Ulrich Leopold

Department of Soil Science
The University of Trier

E-mail: leop6101@...
phone: 0049-(0)651-140764