DAILY PAPER REVIEW

20110813_Characterisation of Alpine lake sediments using multivariate statistical...

 

 

20110813_Daily Paper Review by Aamir Alaud-din
Title and Author:
Title: Characterisation of Alpine lake sediments using multivariate statistical techniques
Authors: Sara Comero a,b,*, Giovanni Locoro b, Gary Free b,1, Stefano Vaccaro b,2, Luisa De Capitani a, Bernd Manfred Gawlik b
a Department of Earth Sciences “Ardito Desio”, University of Milan, Via Botticelli 23,20133 Milan Italy
b European Commission, Joint Research Centre, Institute for Environment and Sustainability, Via Enrico Fermi, 21020 Ispra, Italy

Summary of Paper
Positive matrix factorization (PMF) is a recent technique in multivariate analysis. Unlike other multivariate methods, it does not involve removal of outliers and is not data sensitive techniques. Rather, it gives weight to samples and without removing outliers it gives results.
Three multivariate statistical techniques were applied in this study on X-ray fluorescence (XRF) analyzed samples taken from alpine lake sediments. Algorithm for PMF is as given below
X = GF + E
where x are the elements of input data matrix, G are the elements of factor scores and F are the factor loadings and E are the residuals.
Two different error estimates were used which have already been published. Percentages/concentrations of different metals being studied are shown in table 1.
Table 1
    Na (%)    Mg (%)    Al (%)    Si (%)    P (%)    S (%)    Cl (mg/kg)    K (%)    Ca (%)    Ti (tg%)
Fac 1    0.00    0.00    0.00    0.75    0.01    0.09    6.56    0.00    0.18    0.00
Fac 2    0.11    0.05    0.42    2.59    0.00    0.00    5.81    0.13    0.00    0.02
Fac 3    0.01    0.13    0.03    0.45    0.01    0.01    13.41    0.03    3.23    0.00
Fac 4    0.02    0.07    0.49    0.66    0.00    0.00    0.27    0.17    0.00    0.02

    V
(mg/kg)    Cr (mg/kg)    Mn (mg/kg)    Fe (%)    Co (mg/kg)    Ni (mg/kg)    Cu (mg/kg)    Zn (mg/kg)    As (mg/kg)    Pb (mg/kg)
Fac 1    2.54    1.36    6.29    0.12    0.11    0.00    0.08    21.19    0.00    12.09
Fac 2    2.69    3.90    11.08    0.10    0.29    0.21    0.03    0.53    0.43    3.09
Fac 3    1.30    0.00    20.10    0.03    0.27    0.68    1.94    2.37    0.91    2.81
Fac 4    5.15    3.77    31.16    0.27    0.92    0.04    3.44    13.87    0.51    6.04

Factor 1 explained variation in S, Pb, Zn, and P. Maximum variation was found in S equal to 70% whereas the second most variation was in Pb (>30%). Factor 2 accounted for most of the variability in Ca (>80%). This factor also explained variation in Mg and Cl about 30%. Factor 3 explained the variability in Na, and Si and major and to a lesser extent in Ti and Al. Factor 4 revealed 30% to 50% variability in Al and K and some transition elements Ti, V, Mn, Fe, and Co.
By: Aamir Alaud-din
aamiralauddin@gist.ac.kr

첨부 (1)
20110813_Characterisation of Alpine lake sediments using multivariate statistical techniques.pdf
100.6KB / Download 1