AN ATTEMPT TO INTERPRET THE SPREAD OF ELEMENT CONCENTRATION IN ANTARCTIC SURFACE SNOW - THE SAME ELEMENT IN A GIVEN TEST FIELD, SEVERAL ELEMENTS IN DIFFERENT SAMPLING FIELDS

Sampling surface snow on a large test field always leads to a spread o f analyte concentration data which partly follows a Gaussian distribut ion and partly a rectangular one as can be observed from the analysis of literature data. The spread depends on the nonuniformity of the air -snow interface in the field and on the extent of reproducibility of a ll the procedures used from sampling to analysis. Consequently a sampl e relevant to a restricted surface might be poorly representative of t he surrounding area. Contamination of the sample during the gathering and storing steps is assumed to be the main source of nonrandom result s (outliers). Using various statistical tools rye were able to evaluat e which part of the spread was due to the snow surface nonuniformity i n the case of many samples collected in the same test field. In the ca se of samples gathered in different geographical areas, the possibilit y of finding correlations among points is greatly enhanced when three or more analytes are considered for each sample. When the same correla tion is found for some analytes and a variable tentatively tested, inf ormation can be gained about the source of chemical content of snow sa mples. The use of UV pretreatment of snow samples has been proven to c ut down the interference of organics on the electrochemical process in DPASV, allowing one to obtain accurate and reproducible data. (C) 199 6 Academic Press, Inc.