1. University of Texas at Austin
Title: Space-Time Analysis of Water Availability Model Data in GIS
Statistical Analysis of Naturalized Flows in Texas’ Water Availability Model
Considerations on the impact of trends in the naturalized flows of Texas’ Water Availability Model (WAM), as evidenced by statistical analysis.
Clark Siler, 28 Apr 2009
CE 397 – Statistics in Water Resources
University of Texas at Austin

2. Presentation Outline Introduction Correlation and Trend Analysis
What is a WAM
What are Naturalized Flows
Correlation and Trend Analysis
Pearson’s r
Kendall’s Tau
Results
Conclusions and Possible Future Work
Presentation Outline

3. Water Availability Model (WAM)
Texas’ Water Availability Model consists of:
Water Rights Analysis Package
Suite of programs to digitally manage water rights
Based on probabilistic and statistical analyses
Developed by Dr. Ralph Wurbs at Texas A&M
Input datasets for Texas’ 23 River Basins
This includes the naturalized flow datasets
Wurbs (2005)

4. Uses of WAM Texas Commission on Environmental Quality (TCEQ)
Assigning new or adjusting existing water right permits
Determine if sufficient water to meet proposed use
Evaluate impact of use on other water users
Consultants and Water Management Entities
Local and regional planning studies
Preparing permit applications
Texas Water Development Board (TWDB)
Statewide planning
Wurbs (2005)

5. Naturalized Flows (Analytical)
NF naturalized flow
GF gaged flow
D water supply diversions upstream
RF return flow upstream
EP reservoir evaporation minus precipitation
DS change in storage in upstream reservoirs
Goal is homogeneous flow dataset
NF = GF + SDi – SRFi + SEPi + SDSi NF GF D RF EP DS

6. Naturalized Flows ?

7. Purpose of Study Goal of naturalized flows is homogeneity
Trends may result from many causes
Climate change
El Niño-Southern Oscillation (ENSO)
Incorrect calculations (not considered)
Trended data are not homogeneous
WRAP does not handle trended data
Results from such are not considered representative

8. Neches Basin
Naturalized Flows:
WRAP

9. Correlation (Pearson’s)
to WRAP
WRAP to WRAP
WRAP

10. Trend Analysis: Kendall’s Tau, t
Non-parametric Statistic (no assumptions on distribution)
Based on rank; tests monotonicity
H0: no correlation between time and flow
Rejected (or not) based on comparison with standard normal calculations
Can address seasonality through careful data separation and processing

11. Overall Kendall Results
Regression using time
Only one location rejected H0
WRAP to time

12. Seasonal Kendall Results
Despite many individual seasons having H0 rejected, only one additional location is rejected overall
WRAP to time

13. Conclusions Kendall’s Tau provides non-parametric statistic
Failed to reject H0 for majority of Neches gages
Suggests homogeneity of dataset
Kendall’s test is for monotonicity; other trends may still exist (e.g. longer-term seasonality, ENSO)

14. Possible Future Work
Working to detect cycles (such as ENSO) following process outlined by Wurbs
Comparing Spearman to Kendall rank correlation
Exploring applications of Pearson (parametric) correlation to Neches datasets
Represent results effectively in GIS
Analyzing other Texas basins

15. Personal Information Clark Siler Graduate Student
University of Texas at Austin Environmental Water Resources Engineering (EWRE) Center for Research in Water Resources BS – Brigham Young University Civil Engineering
MS – University of Texas at Austin EWRE
Personal Information (Presenter)
Nov 2008

16. Additional Information

17. San Diego MSL

18. Codes

19. Tables
WRAP to time
WRAP to time