A STATISTICAL-METHOD FOR LITHO-FACIES IDENTIFICATION

In this paper a procedure for the estimation of the lithological descr iption of the stratigraphical formation by wells is described; in this procedure the results of wireline measurements and a litho-facies dat abase are combined. The litho-facies are defined from log values and h istograms and are translated into gaussian distributions. In order to optimise type lithologies we have used the discriminant function refer red to as the Bayes rule. In addition, several logic rules have been a pplied in order to discriminate between different litho-facies (test o f shale volume and bore hole conditions). The wireline logs used are d ensity, sonic transit time, gamma ray, spontaneous potential and resis tivity. The litho-facies groups represented in this database are sands tone, calcareous sand, shaly sand and shale. The procedure has been ap plied to two wells in a fluvial environment (sand-shales series). The results have been compared with drilling rates, cutting descriptions a nd Dipmeter information. For each level a lithological column can be p roduced with a computed probability. The examples are used for the stu dy of sequences and facies in the Tertiary formations in the foreland Chaco basin of Bolivia (latitude 19-20-degrees-S).