The Indicator Approach To Estimation Of Spatial Distributions

Few Earth Scientists realize fully the assumptions they are making, willing or not, when pooling into a same histogram data coming from different locations and considering such histogram as an estimate of a probabilistic distribution. First they need the hypothesis of stationarity in space; second the presence of spatial auto-correlations implies that data be unequally weighted according to their relative locations and the degree of spatial continuity of the underlying phenomenon. The novel technique of indicator kriging allows construction of local or global histogram which can be considered as unbiased estimators of precisely defined probabilistic distributions, full account being given to the underlying continuity. Such estimates of local distributions can be used in mine planning for estimation of local recoverable reserves.