WSPC04137 - Laserfiche WebLink

Home Browse Search

I,U :}.J , , HIERARCHICAL CLUSTER ANALYSIS (HCA) OF COMPLETE GEOCHEMICAL DATA SET . IDENTIFIES THREE DISTINCT GROUPS OF WELLS The purpose of HCA is 10 group multi\'ariale data such that underlying links between the groups can be discerned (Davis, 1973. p. 456). This ~ouping is accomplished by calculating a similarity distance of all \'arlables contained in the data set between all possible pairs of samples Aher sample distances have been compUled. the IWO most similar samples are linked and this linkage continues until all the samples and clusters hne been linked. Identical samples .....ould ha\'e a similarny \'alue (SV) of 1.0 and the most dissimilar sample/c1usler in the data set would hne a SV of 0.0. Many methods exist for establishing the linking belween samples and clusters. with sc\'en clustering options available in the PIROUETTE software package (lnfometri.ll. 1992. p. 4-6 10 4.7). Results of HCA are displayed in the form of a dendogram consuucled with the SV scale on lOp decreasing from 1.0 (most similar) to 0.0 least similar (figure below). After log transformation and autoscaling of the ANETIi data set. HCA was applied. The complete link clustering option (Inforntlrix. 1992, p. 4-19 to 4.20) resu\led in the mosl discernible clusters. Thee clusters (designated groups 1.2. and 3) are distinguished using a SV of 0.45 defined by the dotted vertical line on the dendogram (figure below), The group 3 cluster contains the five OFB samples. Group I and 2 clusters contain all the samples from the NavajO aquifer Three of the four samples from wells in the Upper Paleozoic aquifer are in the group I cluster and the remaining sample is in the group 2 cluster. . :~~::~ ~ GROUP >I;~ ~~ ONE )". '.. " -~y A~A~ ~A~A ..AAAAA ~- ..">h{)......A. ~AAA~AA GROUP TWO ~~ ~ ~""AA">:'" ~ A....v..A. ~A.,("..,A . GROUP II THREE ~. SIMILARITY VALUE, UNITLESS 1.0 0,8 0.2 . 0.0 0.6 o,~ f - f- l-}-- - -, , 'I-- l- I- Aneth Progress Report..Fehmary /995