Applied Geostatistics with SGeMS : Errata

The following are errors that have been found in the first edition of the SGeMS user's guide.

Acknowledgements

• We referred to Mohan Srivastava as Dr., Mr. Srivastava kindly reminded us that he did not stay at Stanford long enough to get his PhD (to the great regret of Prof. A. Journel we might add.)

Chapter 3

p.38 line 18

an estimated map, similar but not…

p.39 line 29

…(Goovaerts, 1997, p.372). In the framework of an algorithm-driven random function, the L realizations define the random function; they are useful only if the process (algorithm) used for their generation reflects the joint dependence between data and unknowns. The arithmetic average of the L simulated…

p.42 line 8

…the determination of the conditional pdf Prob{Z(u) = z | n(u)} of the..

p.47 line 3 from bottom

…anchored on different variogram models;…

p.49 line 16

…1998, p.25). A worthwhile addition would be the power model.

p.50 line 2

…involving only two locations in space at a time…

p.56 line 5

Note that such cosine trend…

p.57 line 8 from bottom

…estimation paradigm if it could be applied to…

p.59 line 2

B; D^t denotes the data…

p.59 line 3

…u_α which can be of point support P or block support B; n(u) denotes…

p.60 line 6

…The Markov screening hypothesis calls for only the secondary datum co-located with the point being estimated. This restricts…

p.61 line 5

this is written as:

p.61 line 17

One possible solution ….. original variables.

p.63 line 11 from bottom

…providing the implicit weight given…

p.64 line 4 from bottom

the conditional probability, that is Bayes relation:

p.65 line 7

Indicator kriging, when…identifies Bayes relation (Eq.(3.38)). Simulating the value…

p.67 line 3

…and is normally performed…

p.68 line 12

can be drawn (in first approximation) independently…

p.69 line 7 from bottom

…independent of (lack of correlation is not enough) the estimated…

p.72 line 7 from bottom

…in the course of simulation. Because scanning and the search tree storage do not retain the location information, the training image should be “stationary”. That is, any particular pattern should be found at many places throughout the training image. Issues of non stationarity can be dealt with by considering separately different stationary training images, see later p.178. Details about the search tree….

p.74 line 19

Polyakova and Journel, 2007 (pp 715-733).

p.77 line 13

account for the interaction of data event D_i with all information already considered.
The exact expression of the….is (Polyakova and Journel, 2007 (pp 715-733)):

p.77 line 5 from bottom

…would correspond to no interaction between the data event…

p.77 line 3 from bottom

…could be read as a measure of how data D_i modifies the information brought by the previous data…

p.78 line 12

…which is independent of the data sequence D₁, D₂, …,D_n . Also evaluation…

p.79 line 12

of data interaction for any specific set…