CHAPTER 4 CONCLUSIONS
3.1 Determination of number of chemical fingerprints
The 13 eigenvalues and the cumulative percent variance for each are shown in Table 29. The groundwater mixing model consisting of more than four end-members would account for greater than 91% of the total observed variance in the data set as shown in Table 29. The number of end-members was determined by evaluation of normalized varimax loadings and CD scatter plots in “Quick Start”.
Normalized varimax loadings are listed in Table 29. There is a sharp drop from Factor 5 to Factor 6 and no factor beyond six has steep drop. This indicates a minimum of 5 end-members.
The “coefficient of determination” (CDs) of all variable is presented in Table 30 which indicates that a minimum of six end-members are required to accurately back-calculate all 13 analytes. The final analyte to obtain a reasonably high CD is cis-1,2-dichloroethene which progresses from a CD of 0.36 at 5 end-members to a CD of 0.66 at 6 end-members. CD scatter plots for five and six end-members are presented in Figs. 6 and 7. Figure 1 indicates a good fit for most analytes, but poor fit for a number of analytes, most notably cis-1,2-dichloroethene (CD = 0.36 ). For six end-members, cis-1,2-dichloroethene improves to 0.66 as presented in Fig. 7.
In summary, the criteria used to determine the number of fingerprints in the model
indicate a minimum of five end-members, and as many as six. The data were submitted to “Quick Start” for resolution of a six end-member model.
3.2 End-member fingerprint composition and geographic distribution
End-member compositions and sample mixing proportions were resolved for a six end-member model. End-member fingerprint compositions shown in a bar chart of Fig. 8 to Fig. 13.
Maps showing the geographic distribution of the six end-members (EM-1 to EM-6) are presented in Fig. 8 to Fig. 13. These figures are “bubble-maps”. The shaded circles or bubbles shown at each sampling station have a diameter proportional to the amount of that fingerprint in each sample. The chemical composition, geographic distribution, and interpretation of each fingerprint are discussed below.
3.2.1 End-Member 1
Through the analyses of PVA, the corresponding results show that the mixing proportions of EM-1 in each sample are less than 35% as indicated in Fig.8. In other words, the concentrations of EM-1 in each sample are small. The most likely explanation is that the EM-1 may be considered as background concentration. The background concentration is actually derived from the data transformation that the
“non-detect” values are replaced by one-half of the reported limit of detection. The
background concentration generally exists in samples, and their values are usually very small. Accordingly, we conclude that EM-1 is the background concentration.
3.2.2 End-Member 2
This fingerprint is related to a predominantly TCE source. The highest concentration of EM-2 is observed in the monitoring wells (MWs) 19, 29, 30, 31, 32, and 39 located in the center area shown in Fig. 9. This suggests a contribution from TCE.
3.2.3 End-Member 3
EM-3 is dominated by Trichlorofluoromethane and Chloroform. EM-3 is noted in highest proportions in the MWs 9 and 11 and spread in the center area shown in Fig.
10.
3.2.4 End-Member 4
This fingerprint is dominated by 1,1-dichloroethene. In geographic distribution, EM-4 is noted in highest proportions in the MWs 8, 9, 17, 20, 27 and 28 spreading in the center area shown in Fig. 11.
3.2.5 End-Member 5
This fingerprint is related to a predominantly Chloroform source. EM-5 is noted
in highest proportions in the MW 25 located in the west region shown in Fig. 12.
This suggests a contribution from Chloroform.
3.2.6 End-Member 6
This fingerprint is related to a predominantly 1,1-dichloroethene source. EM-6 is noted in highest proportions in the MW 35 located in the center area shown in Fig. 13.
This suggests a contribution from 1,1-dichloroethene.
CHAPTER 4 CONCLUSION
In groundwater pollution investigation, the measured items for sampling water may not be easy to identify the chemical contaminant sources and map multiple plumes with overlapping geographic distributions. Through the use of PVA, a six end-member (eigenvector) mixing model was developed to analyze the chemical composition or geographical distribution of contributing sources. The results of this study demonstrate the application of PVA in chemical data obtained from a complex environmental system and the end-members pattern can be used to identify the contaminant source. The analyzed results indicate that the method could produce a groundwater mixing model by searching for end-members. In addition, identification of chemical contaminant sources and mapping multiple plumes with overlapping geographic distributions can be simulated by the groundwater mixing model.
REFERENCES
Barabas, N.; Adriaens, P.; Goovaerts, P. (2004a), Modified Polytopic vector analysis to identify and quantify a Dioxin dechlorination signature in sediments.1. Theory, Environmental Science & Technology, 38, 1813-1820.
Barabas, N.; Goovaerts, P.; Adriaens, P. (2004b), Modified Polytopic vector analysis to identify and quantify a Dioxin dechlorination signature in sediments. 2.
Application to the Passaic River, Environmental Science & Technology 2004, 38, 1821-1827.
Bright, D. A.; Cretney, W. J.; Macdonald, R. W.; Ikonomou, M. G.; Grundy, S. L.
(1999), Differentiation of Polychlorinated Dibenzo-p-dioxin and Dibenzofuran Sources in Coastal British Columbia, Canada. , Environmental Toxicology and Chemistry, 18, 1097-1108.
Bureau of Environmental Protection Hsinchu (2006), An Investigation of the Source of Copper and Arsenic in Discharged Wastewater and the Effect to Oyster Living in Hsiensan Area, Final report. (in Chinese)
Cattell, R. B. (1966), The scree test for the number of factors, Multivariate Behav.
Res., 1, 245-276.
Deane, J. M. (1992), Data reduction using principal components analysis, In Brereton, R. G. (ed.), Multivariate pattern recognition in chemometrics, Elsevier, New York,
125-177.
DeCaprio, A. P.; Johnson, G. W.; Alice, M. T.; Carpenter, D. O.; Chiarenzelli, J. R.;
Morse, G. S.; Santiago-Ricera, A. L.; Schymura, M. J. (2005), Polychlorinated biphenyl (PCB) Exposure Assessment by Multivariate Statistical Analysis of Serum Congener Profiles in an Adult Native American Population, Environmental Research, 98, 284-302.
Ehrlich, R.; Full, W. (1987), In: Use and Abuse of Statistical Methods in Earth Sciences, Size, W., Ed., Oxford, London.
Evans, J. C.; Ehrlich, R.; Krantz, D.; Full, W. (1992), A Comparison between Polytopic Vector Analysis and Empirical Orthogonal Function Analysis for Analyzing Quasi-geostrophic Potential Vorticity, J. Geophys. Res., 97, 2365-2378.
Full, W. E.; Ehrlich, R.; Klovan, J. E. (1981), Extended QMODEL – Objective Definition of External End Members in the analysis of Mixtures, J. Math. Geol., 13, 331-344
Full, W. E.; Ehrlich, R.; Bezdek, J. C. (1982), Fuzzy QModel – A New Approach for Linear Unmixing, J. Math. Geol., 14, 259-270.
Han, B. C.; Leng, W. L.; Chen, R. Y.; Fang, G. T.; Hung, T. C. and Tseng, R. J. (1998), Estimation of Target Hazard Quotients and Potential Health Risks for Metals by Consumption of Seafood in Taiwan, Archives of Environmental Contamination and
Toxicology, 35, 711-720.
Huang, S, J; Hsu, S. P.; Hsieh, H. L.; Chen, C. P. and Shiu, S. J. (2006), The Heavy Metal Distribution and Relationship among Hisnchu Science Park, Keya Stream and green Oysters, Conference on Biodiversity of Hsinchu Seacoast, 9-18. (in Chinese)
Huntley, S. L.; Carlson-Lynch, H.; Johnson, G. W.; Paustenbach, D. J.; Finley, B. L.
(1998), Identification of Historical PCD/F Sources in Newark Bay Estuary Subsurface Sediments using Polytopic Vector Analysis and Radioisotope Dating Techniques, Chemosphere, 36, 1167-1185.
Imbrie, J. (1963), Factor and Vector Analysis Programs for Analyzing Geologic Data, Technical Report No. 6, Office of Naval Research
Johnson, G. W. (1997), Ph.D. Dissertation, University of South Carolina.
Johnson, G. W. and Ehrlich, R. (2002), State of Art Report on Multivariable Chemometric Methods in Environmental Forensics, Environmental Forensics, 3, 59-79.
Johnson, G. W.; Jarman, W. M.; Bacon, C. E.; Davis, J. A.; Ehrlich, R.; Risebrough, R.
W. (2000), Resolving polychlorinated biphenyl source fingerprints in suspended particulate matter of San Francisco bay, Environmental Science & Technology, 34, 552-559
Kaiser, H. F. (1958), The varimax criterion for analytic rotation in factor analysis, Psychometrika, 23, 187-200.
Klovan, J. E.; Miesch, A. T. (1976), EXTENDED CABFAC and QMODEL, Computer Programs for Q-mode Factor Analysis of Compositional Data, Computers and Geosciences, 1, 161-178.
Lin, S. and Hsieh, I. J. (1999), Occurrences of Green Oyster and Heavy Metals Contaminant Levels in the Sien-San Area, Taiwan, Marine Pollution Bulletin, 38, 960-965.
Malinowski, E. R. (1991),Factor analysis in chemistry, John Wiley and Sons, Inc., New York, 350 p.
Miesch, A. T. (1976a), Q-mode Factor Analysis of Geochemical and Petrologic Data Matrixes with Constant Row Sums, U.S. Geological Survey Professional Paper, 574-g:47
Miesch, A. T. (1976b), Interactive Computer Programs for Petrologic Modeling with Extended Q-mode Factor Analysis, Computers and Geosciences, 2, 439-492.
Reyment, R.A. and Joreskog, K. G. (1993), Applied factor analysis in the natural sciences. Cambridge university Press, Cambridge, 371 p.
TABLE 1. The copper concentrations of oyster and sediment in Hsianshan area form
1987 to 2006
Sampling time Copper concentration (ppm) source
(year/month) oyster sediment
Note: The symbol – represents that the data is not available.
TABLE 2. The location of sampling points a.
Name Distancea (m) / Location Reference source
Nu -40 BEPH, 2006
N0 -35 BEPH, 2006, Huang et al., 2006
Nd -5 BEPH, 2006
Ku -40 BEPH, 2006
Kd1 80 BEPH, 2006
Kd2 710 Huang et al., 2006
Y 7,780 (at Yuchegou creek) BEPH, 2006
Kd3 7,900 BEPH, 2006
Kd4 8,900 Huang et al., 2006
Kd5 9,300 BEPH, 2006
P 2nd period science park Huang et al., 2006 S a municipal sewer in the eastern
part of Hsinchu city
Huang et al., 2006 I intertidal zone of Keya stream Huang et al., 2006 C coast area of Hsianshan Huang et al., 2006
Note: The distance is measured from the confluence of Nanman creek and Keya stream. The minus symbol represents the upstream direction.
TABLE 3. The copper concentrations from February to December, 2006 collected from 8 sampling points along Keya stream (BEPH, 2006).
Sampling date Sampling points (ppm)
2006 Nu N0 Nd Ku Kd1 Y Kd3 Kd5
February 0.002 0.023 0.016 0.003 0.018 - 0.005 ND
March 0.017 0.018 0.017 0.001 0.015 - ND 0.014
April 0.006 0.021 0.02 0.004 0.023 - 0.015 0.022
May 0.004 0.009 0.016 0.01 0.012 - 0.009 0.008
June 0.003 0.012 0.009 0.013 0.018 - 0.005 0.021
July 0.007 0.024 0.024 0.007 0.023 - 0.018 0.021
August 0.005 0.006 0.082 0.006 0.085 0.004 0.005 0.007 September 0.003 0.004 0.005 0.010 0.008 0.005 0.007 0.013 October 0.022 0.006 0.019 0.017 0.018 0.011 0.015 0.028 November 0.004 0.015 0.007 0.011 0.007 0.075 0.005 0.014
Note: ND (non-detects) represents the value below method detection limits. The symbol – represents that the data is not available.
TABLE 4 The concentration data of 6 chemical items of Data Sets 3 to 12 (BEPH,
Detected limitation 2.5 0.5 2.0 0.013 0.0010 0.00043
27, February Nu 8.4 5.9 11.2 9.22 0.002 0.0016
TABLE 4(conti.) The concentration data of 6 chemical items of Data Sets 3 to 12
TABLE 4(conti.) The concentration data of 6 chemical items of Data Sets 3 to 12
TABLE 5. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 1, 6×18 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
TABLE 6. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 1, 6×18 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
P 0 0 0 1 0 0
N0 0 0 1 0 0 0
S 0 1 0 0 0 0
Kd2 0 0 0 0 1 0
Kd4 0 0 0 0 0 1
I 1 0 0 0 0 0
TABLE 7. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 2, 7×18 data matrix, 7 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 EM-7
TABLE 8. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 2, 7×18 data matrix, 7 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 EM-7
P 0 0 0 1 0 0 0
N0 0 1 0 0 0 0 0
S 0 0 1 0 0 0 0
Kd2 0 0 0 0 0 1 0
Kd4 0 0 0 0 0 0 1
I 0 0 0 0 1 0 0
C 1 0 0 0 0 0 0
TABLE 9. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 3, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
SS 20.7 91.5 21.9 34.8 36.9 15.9
DO 5.7 4.0 15.4 9.0 17.3 6.6
BOD 19.6 0.7 29.2 10.5 17.0 24.2
NH3-N 12.6 1.5 24.1 20.6 11.0 14.5
Cu 23.8 0.4 5.2 8.4 8.0 18.8
As 17.5 1.8 4.2 16.7 9.8 20.0
TABLE 10. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 3, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 1 0 0 0
N0 1 0 0 0 0 0
Nd 0 0 0 0 0 1
Ku 0 0 0 0 1 0
Kd1 0.76 -0.1073 -0.0925 0.2891 0.0646 0.0861
Kd3 0 0 0 1 0 0
Kd5 0 1 0 0 0 0
TABLE 11. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 4, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 8.9 44.4 85.7 46.8 18.0 19.4 DO 4.6 19.0 1.8 9.9 5.6 3.9 BOD 2.4 13.0 3.2 11.9 4.5 5.0
NH3-N 36.2 9.5 2.0 1.3 45.6 29.6
Cu 13.6 2.9 5.8 25.2 0.5 10.5 As 34.3 11.2 1.5 4.9 25.7 31.6
TABLE 12. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 4, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 1 0 0 0
N0 0.6561 0.0241 0.2185 -0.1042 -0.1591 0.3646
Nd 1 0 0 0 0 0
Ku 0 1 0 0 0 0
Kd1 0 0 0 0 0 1
Kd3 0 0 0 0 1 0
Kd5 0 0 0 1 0 0
TABLE 13. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 5, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 21.3 84.1 52.2 57.6 16.3 19.3 DO 5.2 3.8 6.5 6 6.8 6.1 BOD 6.5 5.4 26.5 5.4 10.7 5.7
NH3-N 39.5 2 4.2 1 38.8 37.7
Cu 19.9 2.3 7.5 24.9 19.6 21.8
As 7.7 2.4 3.1 5.1 7.7 9.6
TABLE 14. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 5, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 1 0 0 0
N0 1 0 0 0 0 0
Nd 0 0 0 0 0 1
Ku 0 1 0 0 0 0
Kd1 0.4291 -0.2728 -0.0074 0.4043 0.3438 0.1031
Kd3 0 0 0 0 1 0
Kd5 0 0 0 1 0 0
TABLE 15. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 6, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 5.1 46.5 7.1 2.8 28.8 9.1 DO 7.6 11.3 5.6 6.9 11.0 8.5 BOD 6.4 21.9 5.7 6.5 21.3 6.0 NH3-N 53.2 10.7 64.4 51.6 18.7 50.4
Cu 13.9 6.6 8.5 16.7 13.7 11.2 As 13.8 3.0 8.6 15.5 6.5 14.8
TABLE 16. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 6, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 1 0 0 0 0
N0 0 0 0 0 0 1
Nd 0 0 0 1 0 0
Ku -0.1522 -0.1462 0.4552 0.0286 0.1207 0.6939
Kd1 1 0 0 0 0 0
Kd3 0 0 1 0 0 0
Kd5 0 0 0 0 1 0
TABLE 17. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 7, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 6.9 87.4 70.2 34.7 6.6 2.3 DO 5.9 2.9 5.5 24.9 8.6 6.8 BOD 5.8 2.8 0.9 12.1 6.7 5.0
NH3-N 42.2 0.3 1.2 14.0 3.4 38.1
Cu 16.7 4.1 18.2 11.3 15.9 13.8 As 22.5 2.5 4.1 3.0 58.8 34.0
TABLE 18. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 7, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 0 1 0 0
N0 0 0 0 0 1 0
Nd 0 1 0 0 0 0
Ku 0 0 0 0 0 1
Kd1 1 0 0 0 0 0
Kd3 -0.2957 0.7521 -0.3093 0.0333 -0.0818 0.9014
Kd5 0 0 1 0 0 0
TABLE 19. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 8, 7×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 19.7 64.1 53.3 35.2 45.5 11.5 DO 9.5 5.1 12.3 7.3 11.2 10.3 BOD 13.0 6.9 10.6 23.2 18.3 16.4
NH3-N 4.0 2.0 3.2 3.0 2.9 4.2
Cu 36.9 17.3 14.6 21.6 12.7 40.6 As 16.9 4.6 6.0 9.7 9.4 17.1
TABLE 20. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 8, 7×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 1 0 0 0
N0 0.1373 0.1003 0.1253 -0.1209 -0.0766 0.8346
Nd 0 0 0 0 0 1
Ku 0 0 0 0 1 0
Kd1 1 0 0 0 0 0
Kd3 0 0 0 1 0 0
Kd5 0 1 0 0 0 0
TABLE 21. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 9, 8×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6 SS 12.9 33.5 13.1 20.2 38.8 3.7 DO 6.2 18.3 8.7 24.5 9.7 3.7 BOD 5.0 18.6 7.2 22.7 26.5 2.9
NH3-N 36.2 10.8 37.4 4.4 9.7 23.5
Cu 5.4 13.1 6.3 14.4 10.1 46.8 As 34.4 5.8 27.3 13.7 5.2 19.4
TABLE 22. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 9, 8×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 1 0 0 0 0
N0 0.7556 -0.3462 0.4184 0.1763 -0.0308 0.0267
Nd 0 0 0 0 0 1
Ku 1 0 0 0 0 0
Kd1 0.7385 -0.0528 -0.6948 0.0828 0.0412 0.8852
Kd3 0 0 1 0 0 0
Kd5 0 0 0 0 1 0
Y 0 0 0 1 0 0
TABLE 23. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 10, 8×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
SS 8.1 14.4 82.8 39.8 5.7 50.2
DO 6.8 17.6 4.5 11.7 8.9 14.7
BOD 5.5 43.0 3.5 9.6 28.4 13.2
NH3-N 45.8 3.5 0.6 8.7 47.3 9.7
Cu 5.3 13.1 6.8 23.1 5.7 6.5
As 28.5 8.4 1.8 7.1 4.0 5.6
TABLE 24. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 10, 8×6 data matrix, 6 end-member model.
Station EM-1 EM-2 EM-3 EM-4 EM-5 EM-6
Nu 0 0 0 0 0 1
N0 0 0 0 0 1 0
Nd 1 0 0 0 0 0
Ku 0 0 1 0 0 0
Kd1 0.7194 -0.0904 0.0677 0.2024 0.1193 -0.0185 Kd3 0.9321 0.0945 0.1914 0.031 -0.1633 -0.0858
Kd5 0 0 0 1 0 0
Y 0 1 0 0 0 0
TABLE 25. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 11, 8×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM1 EM2 EM3 EM4 EM5 EM6 SS 28.9 33.5 72.4 15.4 7.9 20.4 DO 14.9 12.4 2.8 14.2 5.0 8.3
BOD 12.7 5.0 4.5 5.8 2.6 6.6
NH3-N 9.8 5.8 4.0 7.4 56.6 26.7
Cu 26.5 33.9 13.5 14.4 16.2 21.9
As 7.2 9.4 2.7 42.9 11.6 16.2
TABLE 26. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 11, 8×6 data matrix, 6 end-member model.
Station EM1 EM2 EM3 EM4 EM5 EM6
Nu 0.4443 -0.4571 0.1497 -0.3428 0.552 0.6539
N0 0 0 0 1 0 0
Nd 0 0 0 0 1 0
Ku 0 1 0 0 0 0
Kd1 -0.1168 -0.0044 -0.0421 0.061 0.822 0.2805
Kd3 0 0 0 0 0 1
Kd5 0 0 1 0 0 0
Y 1 0 0 0 0 0
TABLE 27. End-member fingerprint compositions (in percent) analyzed through PVA of Data Set 12, 8×6 data matrix, 6 end-member model. Each column sums to 100%.
Item EM1 EM2 EM3 EM4 EM5 EM6
SS 4.7 67.6 8.2 45.8 41.6 3.5
DO 6.1 9.5 6.8 13.3 9.5 6.2
BOD 2.7 6.6 8.0 6.6 13.9 2.1
NH3-N 42.4 7.8 3.9 4.9 11.1 39.3
Cu 5.9 5.6 71.0 20.3 19.2 6.2
As 38.2 2.9 2.1 9.0 4.7 42.7
TABLE 28. End-member fingerprint compositions (in decimal percentages) analyzed through PVA of Data Set 12, 8×6 data matrix, 6 end-member model.
Station EM1 EM2 EM3 EM4 EM5 EM6
Nu 0 1 0 0 0 0
N0 2.8558 0.0251 0.156 0.2922 -0.4468 -1.8823
Nd 0 0 0 0 0 1
Ku 0 0 0 1 0 0
Kd1 1 0 0 0 0 0
Kd3 2.7377 -0.3347 -0.1258 0.4129 0.1687 -1.8587
Kd5 0 0 0 0 1 0
Y 0 0 1 0 0 0
TABLE 29. The 13 eigenvalues, the cumulative percent variance and the normalized varimax loadings for the matrix of 41 samples and 13 organic compounds used indices for determination of the number of significant eigenvectors.
Eigenvector
No. Eigenvalue Cumulative
Percent variance
TABLE 30. Miesch coefficients of determination (CDs) calculated from a matrix of 41 samples and 13 organic compounds. The final analyte to obtain a reasonably high CD is cis-1,2-dichloroethene which progresses from a CD of 0.36 at 5 end-members to a CD of 0.66 at 6 end-members.
Number of end-members
Variable 2 3 4 5 6 7 8 9 10 11 12
1,1,1-trichloroethane 0.33 0.44 0.53 0.55 0.76 0.81 0.81 0.89 1 1 1 1,1-dichloroethene 0.07 0.43 0.76 0.97 0.97 0.98 0.99 0.99 1 1 1
1,1-dichloroethane 0.5 0.53 0.75 0.87 0.94 0.96 0.97 1 1 1 1
1,2,4-trimethylbenzene 0.56 0.55 0.93 0.98 0.98 0.98 0.99 0.99 0.99 1 1 1,3,5-trimethylbenzene 0.55 0.55 0.93 0.98 0.98 0.98 1 1 1 1 1
Benzene 0.51 0.54 0.9 0.98 0.98 0.99 0.99 0.99 0.99 1 1
Chlorobenzene 0.51 0.53 0.89 0.99 0.99 0.99 1 1 1 1 1
Chloroform 0.13 0.68 0.79 0.98 0.98 0.99 1 1 1 1 1
cis-1,2-dichloroethene 0.1 0.09 0.17 0.36 0.66 0.83 0.92 0.95 1 1 1
m,p-xylene 0.36 0.44 0.56 0.6 0.83 0.87 0.9 0.95 1 1 1
Toluene 0.45 0.45 0.72 0.75 0.75 0.9 0.99 1 1 1 1
Trichloroethylene 0.88 0.95 1 1 1 1 1 1 1 1 1
Trichlorofluoromethane -0.02 0.18 0.31 0.95 0.95 0.98 1 1 1 1 1
FIGURE 1. Study area locations.
FIGURE 2. Sampling points location along Keya stream, Nanman creek, and Yuchegou creek.
0
FIGURE 3. The concentration data of 18 heavy metals of water samples collected at P, N0, S, Kd2, Kd4, and I in April, 2006 (Huang et al., 2006).
0
FIGURE 4. The concentration data of 18 heavy metals of suspended solid samples collected at P, N0, S, Kd2, Kd4, I, and C in April, 2006 (Huang et al., 2006).
FIGURE 5. Statistical information of temperature and rainfall in Hsinchu during February to November, 2006
FIGURE 5. (conti.) Statistical information of temperature and rainfall in Hsinchu during February to November, 2006
FIGURE 6. The CD scatter plots for 5 end-member model: an industry park in the northern part of Taiwan, 13x41 matrix. Figure 1 indicates a good fit for most analytes, but poor fit for a number of analytes, most notably cis-1,2-dichloroethene (CD=0.36 ).
FIGURE 7. The CD scatter-plots for 6 end-member model: an industry park in the northern part of Taiwan, 13x41 matrix. For six end-members, cis-1,2-dichloroethene improves to 0.66.
FIGURE 8. Fingerprint composition and geographic distribution for end-member 1.
FIGURE 9. Fingerprint composition and geographic distribution for end-member 2.
FIGURE 10. Fingerprint composition and geographic distribution for end-member 3.
FIGURE 11. Fingerprint composition and geographic distribution for end-member 4.
FIGURE 12. Fingerprint composition and geographic distribution for end-member 5.
FIGURE 13. Fingerprint composition and geographic distribution for end-member 6.
個人資料
姓名：莊敏筠
生日：民國 72 年 2 月 24 日 出生地：新竹縣
聯絡電話：0952620135
email：abandon.ev94g@nctu.edu.tw
住址：新竹縣峨眉鄉峨眉村 2 鄰峨眉街 25 號 學歷：
民國 94 年畢業於國立交通大學土木工程學系 民國 96 年畢業於國立交通大學環境工程研究所