R. Jimenez-espinosa et M. Chica-olmo, Application of geostatistics to identify gold-rich areas in the Finisterre-Fervenza region, NW Spain, APPL GEOCH, 14(1), 1999, pp. 133-145
Three univariate geostatistical methods of estimation are applied to a geoc
hemical data set. The studied methods are: ordinary kriging (cross-validati
on), factorial kriging, and indicator kriging. These techniques use the pro
babilistic and spatial behaviour of geochemical variables, giving a tool fo
r identifying potential anomalous areas to locate mineralization. Ordinary
kriging is easy to apply and to interpret the results. It has the advantage
of using the same experimental grid points for its estimates, and no addit
ional grid points are needed. Factorial kriging decomposes the raw variable
into as many components as there are identified structures in the variogra
m. This, however, is a complex method and its application is more difficult
than that of ordinary or indicator kriging. The main advantages of indicat
or kriging are that data are used by their rank order, being more robust ab
out outlier values, and that the presentation of results is simple. Neverth
eless, indicator kriging is incapable of separating anomalous values and th
e high values from the background, which have a behaviour different to the
anomaly. In this work, the results of the application of these 3 kriging me
thods to a;set of mineral exploration data obtained from a geochemical surv
ey carried out in NW Spain are presented. This area is characterised by the
presence of Au mineral occurrences. The kriging methods were applied to As
, considered as a pathfinder of Au in this area. Numerical treatment of Au
is not applicable, because it presents most values equal to the detection l
imit, and a series of extreme values. The results of the application of ord
inary kriging, factorial kriging and indicator kriging to As make possible
the location of a series of rich values, sited along a N-S shear zone, cons
idered a structure related to the presence of Au. (C) 1998 Elsevier Science
Ltd. All rights reserved.