Best applicable geostatistical model for interpolating

Best applicable geostatistical model for interpolating groundwater-levels in El-Obour groundwater-levels groundwater ElEl city, Egypt Dina Mostafa Elleithy (Presenter)

8th International Conference on Sustainable Water Resources Management

Water Resources Management 2015

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

15 - 17 June 2015 A Coruña, Spain

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

1. 2. 3. 4. 5. 6. 7.

Aim of the study Case Study Objectives Methodology Results and Analysis Conclusions and Further studies Delimitations

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

• General Methodology for identifying the most suitable geostatistical model for interpolating groundwater-level data. • using Geographical information systems (GIS) • Identify the best applicable geostatistical model (Applying the methodology)

• District VI, El-Obour city

• Generate Groundwater-level map with good accuracy • designing and/or optimizing the existing monitoring network Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

District VI, El-Obour City

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

• El-Obour City • 816 Wells • 54.3 km2

• District VI • 313 Wells • 3.1 km2 • (6% % of El-Obour)

Aim of the study Case Study

Topography

Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Catchment “Heliopolis basin”

Aim of the study Case Study Objectives Methodology

Lithological Data Lithology Model Side-View using GMS

Results & Analysis Conclusions Delimitations

N Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Lithological Data Sample of Cross-Section

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aquifer A

Aquifer A

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

• 154 Wells • 1.4 km2

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions

• Most suitable geostatistical model • Study Area (Aquifer A)

• Best accuracy for generated maps • Groundwater-level surface

Delimitations

i.e. Best usage of available data

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions

• Built all available geostatistical models

• using Hierarchy Stepwise criteria (HS)

• Compare them

• Cross-Validation

Delimitations

• Choose the best suitable model for Interpolation Groundwaterlevels • Study Area (Aquifer A)

• Generate Groundwater-level Surface Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis

Selection criteria • Based on Cross-validation statistical parameters n

• Mean Prediction Error

∑ | (Z

MPE =

i =1

Delimitations

• Root Mean Square

∑ (Z

RMS =

i =1

• Mean Standardized

*

( si ) − Z ( si )) 2 n

n

MS =

( si ) − Z ( si )) | n

n

Conclusions

*

∑ | (Z

*

i =1

( si ) − Z ( si )) | / σ * ( si ) n

n

• Root Mean Square Standardized

( Z * ( si ) − Z ( si ))  * ∑  σ ( si ) i =1  n

RMSS = n

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

• Average Standardized Error

ASE =

∑σ i =1

*

n

( si )

2

Aim of the study

Selection criteria

Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

• criteria used in comparing models are: • Absolute Standardized Error (ASE) is close to Root Mean Square Standardized (RMSE) Validity

• smallest Root Mean Square (RMS) Prediction ≈ Measured

• Mean Standardized (MS) nearest to zero Unbiased

• RMSS nearest to 1 Validity

• smallest Mean Prediction Error (MPE) Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Validity

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

First Step: Different Methods

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study

Second Step: Fitting Functions

Case Study Objectives Methodology Results & Analysis Conclusions

Cross-Validation Statistical Parameters

Delimitations Models Comparison

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Parameter MPE RMS MS RMSS ASE RMS-ASE RMS MS≈0 RMSS≈1 MPE≈0

Units m (*10^-3) m (*10^-3) m m (*10^-2) m (*10^-3) m(*10^-3)

C 0. 504 0.607 -0.876 0.902 0.671 6.373 0.607 0.876 0.098 0. 504

Sph 2.158 0.605 1.688 0.888 0.680 7.409 0.605 1.688 0.112 2.158

T-Sph 2.206 0.605 1.776 0.882 0.684 7.861 0.605 1.776 0.118 2.206

Ordinary kriging fitting functions P-Sph Exp G R-Q H-E 2.192 15.265 2.162 17.148 3.617 0.605 0.606 0.591 0.602 0.581 1.733 20.403 1.816 22.537 4.890 0.880 0.894 0.987 0.863 1.014 0.685 0.682 0.597 0.701 0.573 8.035 7.553 0.661 9.916 0.875 0.605 0.606 0.591 0.602 0.581 1.733 20.403 1.816 22.537 4.890 0.120 0.106 0.013 0.137 0.014 2.192 15.265 2.162 17.148 3.617

K-B 2.280 0.594 1.971 1.009 0.587 0.652 0.594 1.971 0.009 2.280

J-B 2.859 0.584 3.343 1.003 0.581 0.273 0.584 3.343 0.003 2.859

S 2.453 0.592 2.245 1.001 0.590 0.221 0.592 2.245 0.001 2.453

Aim of the study Case Study

Advanced Comparison

Objectives

1

Methodology Results & Analysis Conclusions

I m = 100 * (1 −

∑r n

m

( p) − n

n * ( R − 1)

)

Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Where: rm (P): Model Rank in each parameter n : Total No. of caparisoning Parameters R : Total No. of Models Im : Model Index

Aim of the study

Second Step: Fitting Functions

Case Study Objectives Methodology Results & Analysis Conclusions

Cross-Validation Statistical Parameters

Delimitations Models Comparison

Parameter Ranking Advanced Comparison Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Parameter Units MPE m (*10^-3) RMS m MS (*10^-3) RMSS ASE m RMS-ASE m (*10^-2) RMS m MS≈0 (*10^-3) RMSS≈1 MPE≈0 m(*10^-3) RMS-ASE RMS MS ≈ 0 RMSS ≈ 1 MPE ≈ 0 Index (%)

C 0. 504 0.607 -0.876 0.902 0.671 6.373 0.607 0.876 0.098 0. 504 6 11 1 6 1 60

Sph 2.158 0.605 1.688 0.888 0.680 7.409 0.605 1.688 0.112 2.158 7 9 2 8 2 54

T-Sph 2.206 0.605 1.776 0.882 0.684 7.861 0.605 1.776 0.118 2.206 9 8 4 9 5 40

Ordinary kriging fitting functions P-Sph Exp R-Q H-E G 2.192 15.265 2.162 17.148 3.617 0.605 0.606 0.591 0.602 0.581 1.733 20.403 1.816 22.537 4.890 0.880 0.894 0.987 0.863 1.014 0.685 0.682 0.597 0.701 0.573 8.035 7.553 0.661 9.916 0.875 0.605 0.606 0.591 0.602 0.581 1.733 20.403 1.816 22.537 4.890 0.120 0.106 0.013 0.137 0.014 2.192 15.265 2.162 17.148 3.617 10 8 3 11 5 7 10 3 6 1 3 10 5 11 9 10 7 4 11 5 4 10 3 11 9 42 20 74 10 52

K-B 2.280 0.594 1.971 1.009 0.587 0.652 0.594 1.971 0.009 2.280 4 5 6 3 6 62

J-B 2.859 0.584 3.343 1.003 0.581 0.273 0.584 3.343 0.003 2.859 2 2 8 2 8 66

S 2.453 0.592 2.245 1.001 0.590 0.221 0.592 2.245 0.001 2.453 1 4 7 1 7 70

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Third Step: a) Data Normalization

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Third Step: b) Data trend Removing

Aim of the study Case Study

Generated Map

Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Groundwater-Level Surface isoline map

Aim of the study Case Study

Generated Map

Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Groundwater-Level Surface map

Aim of the study Case Study

Final Geostatistical Model

Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

• Ordinary Kriging • Gaussian Fitting function • Transformation of Data Using Box-Cox: Power Parameter = 2 • No External Trend affecting the Data

Aim of the study

Further Studies

Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

•

Advanced comparison assumes equal weights for all the parameters

•

Apply bivariate data, Implement co-kriging models

• •

Check the change in the hierarchy [steps order] Reapply the same methodology on different case studies compare the results

• •

• •

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

change in the weights will enrich and gives more reliability correlating the groundwater-level data to the topography

validate the methodology highlights its limitations

Aim of the study Case Study Objectives Methodology Results & Analysis Conclusions Delimitations

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

Delimitations • The data is considered to be a uni-variate. • The spatial interpolation methods are only those commonly used or cited in environmental studies. • Output results from this thesis are limited to the case study of District VI, El-Obour City, Egypt on which the thesis methodology was applied. • Software products were Microsoft Excel, Groundwater Simulation Model GMS, and ArcGIS 10 geostatistical Analyst tool.

Best applicable geostatistical model for interpolating groundwater-levels in El-Obour groundwater-levels groundwater ElEl city, Egypt

Thank You 8th International Conference on Sustainable Water Resources Management

Water Resources Management 2015

Irrigation and Hydraulics Department Ain Shams University

Dina M. Elleithy

15 - 17 June 2015 A Coruña, Spain