Sequential Indicator Simulation (SIS), although widely used, is relati
vely slow, and requires tedious inference of a large number of indicat
or variogram models. SIS is designed only to estimate class proportion
s and to reproduce indicator variogram models; the statistics of the c
ontinuous attribute being simulated, z-histogram and variogram, may be
poorly reproduced. Several implementations of the SIS algorithm are p
roposed resulting in better reproduction of statistics yet with better
CPU performance.