1. 1 Evaluation of Groundwater Quality Parameters Using Multivariate Statistics Methods A case Study of Majmaah, KSA Students Hussam Khaled Almubark Abdullah A. Alzeer Majid Mufleh Almotairi A presentation prepared for: Supervisor Dr. SaMeH S. Ahmed Civil and Environmental Engineering Department College of Engineering – Majmaah University

2. Outlines Introduction Objectives and Problems Water Quality Data Methodology Statistical Analysis Multivariate Statistical Analysis Geostatistics Conclusions & Recommendations 2

3. Introduction Groundwater is the main source (95 %) of fresh water in Saudi Arabia. It is important to ensure good and safe water quality for drinking and other purposes. Monitoring water quality is important and any contribution to assist this program is always welcome. Characterisation of the water quality parameters means defining levels, distribution, changes with time, etc. 3

4. 4 Objectives The main objectives of this Project are: To evaluate and map the characterisation of groundwater quality at certain area in Majmaah. To assist the monitoring program by reducing the number of measured variables, using advanced statistical methods (Multivariate Statistics). To classify groundwater in the area into spatial water quality types using modern techniques (Geostatistics).

5. Problems No previous water quality analysis have been conducted at the study area Need for reducing the number of WQP to save cost and time Problem 2 Problem 3 Problem 1 Wells without coordinates Problem 4 Support the monitoring program by dynamic maps revealing WQP in the place

6. Software’s and Analyses MS Office SPSS Surfer Variable analysis Scatter plot Correlations PCA FA Word Power Point Excel: -Descriptive Stat. -Correlation matrix -Graphs -Coordinates -Calculations Correlations PCA Rotation methods Kriging Contour maps 3D figures StatGraph 6

7. Water Quality Data Field, lab., and Analysis 7

8. Study Area 8 Info.Values LatitudeN 25 o LongitudeE 46 o Area21 km2 Distance12 km No. of wells15 Population60,000

9. Sampling in the field 9

10. Laboratory Analyses 10

11. Results of lab. Analysis 11 WellpH EC, (µs/cm) TDS, (mg/l) S, (mg/l) Ag, (mg/l) NO3, (mg/l) Hg, (mg/l) 1 8.4 383.10245.184 30.030.0190.4 2 8.1 341.80218.752 20.030.0170.4 3 8 334.20213.888 30.030.0190.4 4 7.9 333.60213.504 30.030.0190.4 5 7.9 336.80215.552 20.020.0160.3 6 8.2 383.30245.312 20.020.0150.3 7 8.3 383.80245.632 20.020.0150.3 8 8.2 385.10246.464 20.030.0180.4 9 8.3 383.50245.440 30.030.0230.4

12. GPS Data Well Locations 12

13. Well locations 13

14. GPS 14

15. Coordinates of the Wells 15 Well No.NEX, mY, m 1 25 ̊̊ 53.411'45 ̊ 21.375' 4643142863571 2 25 ̊ 48.773'45 ̊ 24.483' 4590992855056 3 25 ̊ 48.769'45 ̊ 24.463' 4591322855018 4 25 ̊ 48.456'45 ̊ 24.109' 4597222854439 5 25 ̊ 48.956'45 ̊ 25.790' 4569162855370 6 25 ̊ 48.684'45 ̊ 25.564' 4572922854867 7 25 ̊ 48.836'45 ̊ 25.951' 4566472855150 8 25 ̊ 49.269'45 ̊ 25.951' 4566492855999 9 25 ̊ 49.087'45 ̊ 25.792' 4569172855612 10 25 ̊ 49.183'45 ̊ 26.067' 4564552855791 11 25 ̊ 48.622'45 ̊ 26.468' 4557812854758 12 25 ̊ 48.563'45 ̊ 26.572' 4556072854649 13 25 ̊ 48.371'45 ̊ 26.776' 4556062854295 14 25 ̊ 48.298'45 ̊̊̊ 26.916' 4550312854162 15 25 ̊ 47.930'45 ̊ 27.306' 4543772853485

16. Locations of tested wells 16

17. Preliminarily Statistical Analysis Mean, Range, Extremes, S.Dev. 17

18. Summary Statistics for WQP 18 Count = 15 Average = 363.013 Variance = 454.537 Standard deviation = 21.3199 Minimum = 333.6 Maximum = 385.7 Stnd. skewness = -0.232746 Stnd. kurtosis = -1.49015 In this case, the standardized skewness value is within the range expected for data from a normal distribution. The standardized kurtosis value is within the range expected for data from a normal distribution. Summary Statistics for “EC” Percentiles for “EC” 1.0% = 333.6 5.0% = 333.6 10.0% = 334.2 25.0% = 341.8 50.0% = 356.3 75.0% = 383.5 90.0% = 385.1 95.0% = 385.7 99.0% = 385.7

19. 19

20. Summary Statistics of the Wells 20 Preliminary statistics lead to exclude the following parameters from the multivariate statistics: F, S, Cr, Color and Ba as they have abnormal distribution which appear from the normal probability distribution and the Box-and- Whisker plot.

21. Comparison with Standards SAS, WHO, EPA 21

22. Comparing the WQP with the Standards 22 NoVariable Av. Measured values SAS standards EPA standards WHO standards Notes 1 pH8.1536.5-8.56.5 – 8.56.5- 8.5 2 Zinc (Zn)0.072 mg/l5000 µg/l5 mg/l3 mg/l 3 TDS232.3 mg/l1500 mg/l500 mg/l1000 mg/l 4 Sulfate (S)2.53 mg/l400 mg/l250 mg/l400 mg/l 5 Silver (Ag)0.027 mg/l-0.1 mg/l 6 Nitrite (NO3)0.018 mg/l-1.0 mg/l 7 Mercury (Hg)0.373 µg/l1 µg/l0.002 mg/l0.001 mg/l 8 Lead (Pb)10.67 µg /l10 µg/l0.015 mg/l0.01 mg/l > MPL 9 Iron (Fe) 0.130 mg/l 1.0 mg/l0.3 mg/l 10 Cyanide (Cn)0.008 mg/l70 µg/l0.2 mg/l0.1 mg/l 11 Copper (Cu)0.200 mg/l1000 µg/l1-1.13 mg/l2.0 mg/l 12 Chromium (Cr)0.031 mg/l0.05 13 Chloride (Cl)0.433 mg/l600 mg/l250 mg/l 14 Cadmium (Cd)4.72 µg /l3.0 µg/l0.005 mg/l0.003 mg/l> MPL 15 Aluminum (Al)0.05 mg/l 100-200 µg/l0.05 -0.2 mg/l0.2 mg/l 16 Dissolved Oxygen (DO) 5.53 mg/l --- 17 Calcium (Ca)0.33 mg/l -200 mg/l*75 mg/l 18 Magnesium (Mg)0.143 mg/l -50 mg/l*50 mg/l 19 EC363.0 µS /mm --- 20 Color103.80 TCU 15 TCU > MPL

23. Multivariate Statistical Methods 23

24. Multivariate Statistical Methods PCA method can be used to:  reduce number of variables and  detect relationships between them. The large set of water quality parameters can be further studied using this method to determine the interrelationship between the parameters. 24

25. Why Multivariate Statistics? 1.To identify the hidden dimensions or constructs that may not be apparent from direct analysis. 2.To identify relationships between variables, it helps in data reduction. 3. It helps the researcher to cluster the product and population being analysed. 25

26. 26 Correlation Matrix- Excel Assumptions: 1) Correlation values between (0.25 – 0.50) indicate weak correlation. 2) Correlation values between (0.50 – 0.74) indicate good correlation. 3) Correlation values > 0.75 indicate strong correlation. Table : Correlation matrix controlling all variables, using Excel

27. 27 Correlation Matrix- SPSS Table : Correlation matrix controlling all variables using SPSS pHZnTDSAgNO3HgPbFeCnCuClCdAlDOCaMgEc pH 1 Zn.0601 TDS -.033.2211 Ag -.119.645.2451 NO3.062.835.360.7671 Hg -.095.657.238.921.7701 Pb.119.840.411.664.899.7051 Fe.230.664.335.519.762.547.8891 Cn.093.773.267.627.799.656.804.6611 Cu.154.794.276.567.747.597.841.728.9001 Cl.281.677.252.498.696.492.781.754.784.8151 Cd.266.808.217.544.767.574.834.735.893.923.8391 Al.335.706.194.448.683.456.778.773.816.853.9341 DO.294-.109-.414-.145-.107-.222-.168-.106-.117-.057.031.007.1081 Ca.192.286-.110.145.225.135.266.221.231.309.321.280.257.1951 Mg.072.217.083.098.188.090.235.207.225.248.129.185.200-.088.6501 Ec -.033-.2211.000.245.360.238.411.335.267.276.252.217.194-.414-.110.0831

28. Results of PCA 28 Data input: observations Number of complete cases: 45 Standardized: yes Number of components extracted: 4 Component Number Eigenvalue Percent of Variance Cumulative Percentage 1 8.7918951.717 2 2.3668013.92265.639 3 1.496458.80374.442 4 1.387978.16582.607 5 0.720774.24086.846 6 0.587683.45790.303 7 0.413712.43492.737 8 0.334591.96894.705 9 0.303561.78696.491 10 0.156020.91897.408 11 0.142400.83898.246 12 0.120770.71098.957 13 0.068930.40599.362 14 0.057970.34199.703 15 0.036540.21599.918 16 0.013910.082100.000 17 0.000030.000100.000

29. Component Weights of PCA 29 Component (1) Component (2) Component (3) Component (4) Ag0.2458-0.11550.34870.2098 Al0.29000.1774-0.0765-0.2171 Ca0.10570.3283-0.32220.4832 Cd0.31010.1327-0.0011-0.1385 Cl0.28870.1226-0.0677-0.1803 Cn0.30560.03640.0684-0.0140 Cu0.30830.0756-0.0141-0.0370 DO-0.04260.44810.0318-0.2175 EC0.1322-0.4898-0.3725-0.1096 Fe0.28540.0106-0.0820-0.1072 Hg0.2521-0.12330.34980.2028 Mg0.08970.1348-0.43910.5829 NO30.3066-0.05440.11920.0529 Pb0.3213-0.0438-0.0076-0.0005 pH0.05510.2949-0.3630-0.3868 TDS0.1321-0.4904-0.3717-0.1089 Zn0.29360.03540.11590.0786

30. 30 The previous table shows the equations of the principal components. For example, the first principal component has the equation. 0.24584*Ag + 0.289998*Al + 0.105667*Ca + 0.310126*Cd + 0.288733*Cl + 0.305632*Cn + 0.308319*Cu - 0.042581*DO + 0.132234*EC + 0.285392*Fe + 0.252125*Hg + 0.0897126*Mg + 0.306629*NO3 + 0.321259*Pb + 0.0551099*pH + 0.132119*TDS + 0.293593*Zn

31. Plots of PCA 31 Scree Plot using STAT Graph Scree Plot using SPSS

32. Plots of PCA 32 Plot of Component in rotated space using SPSS Plot of Component Weights using STAT Graph

33. Interpretation of Multivariate Statistics Results The values obtained from data reduction using PCA method reveal that the first component (factors) involves: Cd, Cn, Cu, NO3 and Pb in one group. The variables: Ca, DO, EC and TDS are the main variable in group two. Also, group three consists of seven variables that might have interrelationship among them, those are: Ca, Ag, EC, Mg, pH, and TDS. The fourth group has two main variables: M g and pH. 33

34. Geostatistics Techniques Contour Maps 34

35. Geostatistics and Kriging Techniques Geostatistics is the statistics of spatially or temporally correlated data. The technique has been used to be a practical approach to the problems of ore reserve estimation and mine planning. It has been also used for other applications concerned with petroleum and gas resources estimation. Kriging is the most famous geostatisics technique that is being use now for several applications. In this project, Kriging is performed using SURFER software. 35

36. 36 pH - Contour Map Contour map shows the distribution of “pH” at the study area

37. 37 TDS - Contour Map Contour map shows the distribution of “TDS” at the study area

38. 38 Contour map shows the distribution of “EC” at the study area EC - Contour Map

39. 39 Contour Maps - Comparison pH TDS EC Zn

40. 3D representation of “pH” at the study area 40 pH – Representation

41. 41 TDS – Representation 3D representation of “TDS” at the study area

42. 42 EC – Representation 3D representation of “EC” at the study area

43. 43 3D- Comparison pH TDS EC Zn

44. Conclusions 44

45. 45 Conclusions Water quality parameters of groundwater wells at part of Majmaah city has been characterised using intensive descriptive statistics and multivariate statistical analysis. 15 groundwater wells in a farming area near Majmaah. Where 3 samples from each well were gathered and sent to environmental engineering lab for chemical analysis. GPS instruments were used to determine the X, Y and coordinates of the tested wells, as there were no coordinates available for those points.

46. 46 Conclusions Laboratory analysis has been conducted over the three samples for 22 water quality parameters, and results were sorted with the determined GPS data in one database for further statistical and Geostatistics analysis. Results of the preliminarily statistical analysis reveal that only 17 variables are suitable for carrying out multivariate statistical analysis to reduce the number of measured variables. A comparison study between WQP and SAS, WHO and EPA is introduce in tables, where most of the recorded parameters were below the standards except Lead ( P b) and Cadmium ( Cd ).

47. 47 Conclusions An attempted to interoperate the resultant groups were made, despite the physical interpretation is not deep. X, Y and measured values of each WQP were used with Surfer software to generate contour maps and 3D representation of each variable within the study area. Nevertheless, the study has been carried out over a small portion, the steps and introduced procedure can be easily applied elsewhere for similar purpose.

48. Recommendations 48

49. Recommendations The project examined the calculation methodology over an area of 21 sq.km with 15 water wells, It might be better if more wells are included, for carrying out trustable geostatistics studies. WQP such as: temp., Turbidity, K, BOD, Na, P, etc.. should be investigated. Grouping WQP as: 1) field parameters and 2) laboratory parameters, then conduct PCA or FA to correlated the two groups and find the inter- correlations between them. Such study needs intensive data and reliable WQP analysis. 49

50. References [1] EPA, (2001), “Parameters of Water Quality – Interpretation and Standards” Report, 133P, ISBN 1- 84096-015-3. [2]Mohammed Al-Saud et.al, (2011), “Challenges for an Integrated Groundwater Management in the Kingdom of Saudi Arabia” International Journal of Water Resources and Arid Environments 1(1): 65-70, 2011. [3]Walid Abdelrahamn, (2006) “Groundwater Resources Management in Saudi Arabia” Special presentation at water conservation workshop, Khoper, KSA, December 2006. [4]FAO, (2009), “Groundwater Management in Saudi Arabia”, A Report by FAO, 14P. [5 ] كتاب :”مرقب جبل منيخ بمحافظة المجمعة” التقرير التاريخي والأثري الصادر من وكالة الآثار والمتاحف. [6]Sameh S Ahmed, (2014), “Surveying 1” Lecture notes of surveying course at Civil and Environmental Engineering Department, MU, KSA, 2014. [7]Gregory T. French (1997), “Understanding the GPS, An Introduction to the Global Positioning System" First edition, April 1997. [8]GPS Coordinate Converter, Maps and Info, http://boulter.com/gps/http://boulter.com/gps/ [9]Excel 2010, “Microsoft Office 2010" [10]STATGRAPHICS plus, (1996): Statistical Graphics Corp. [11]EPA, (2012), “Drinking Water Standards and Health Advisories”, 2012 Edition, EPA 822-s-12-001, Office of water, U.S Environmental Protection Agency. April 2012. [ 12] World Health Organization, (2004),”Guidelines for Drinking-water Quality”, Vol. 1, 3rd Edition, Geneva, ISBN: 9241546387. 15]SPSS 16.0 for windows, (Release 16.0.0, Sept 2007) http://www.winwarp.comhttp://www.winwarp.com 50

51. Thanks for your attention