1. Long-term Salinity Prediction with Uncertainty Analysis
Jim Prairie
CADSWES
(Masters’ Thesis Status)
September 11, 2000

2. Law of the River - Water Quality
Mexico Treaty Minute No. 242
assured water received by Mexico have an average salinity of no more than 115 ppm +/- 30 ppm above the average annual salinity at Imperial Dam
Colorado River Basin Salinity Control Act of 1974
meet United States obligation to Mexico under Minute No. 242
authorized construction of desalting plant and additional salinity control projects

3. Salinity Control Forum
Created by Basin States in response to Federal Water Pollution Control Act Amendments of 1972
Developed numerical salinity standards
723 mg/L below Hoover Dam
747 mg/L below Parker Dam
879 mg/L at Imperial Dam
review standards on 3 year intervals
Develop basin wide plan for salinity control

4. Salinity Control Forum
Salinity Control efforts in place removed 621,400 tons from the system in This accounted for 9% of the salt mass at Imperial Dam
total expenditure through $426,020,000
Projected 1,476,600 tons of salinity controls needed by 2015
projects additional expediture through $169,698,000
(data taken from Quality of Water, Progress Report 19)

5. Colorado River Simulation System - Fortran
First implemented in the early 1980’s
Basin wide model for water and salinity modeling
CRSS model was an essential tool for decision support
model used to determine required salt removal to maintain standards

6. Colorado River Simulation System - RiverWare
CRSS was replaced by a poilcy model in RiverWare in 1996 (RW-CRSS)
Recent attempts to verify RW-CRSS against historic data from 1970 to 1990 failed

7. USBR and Research Objectives
Investigating the salinity methodologies currently used and improving them as necessary for future predictions
Verify the data and calibrating the current model for both water quantity and water quality (total dissolved solids, or TDS)
Development of a long-term, basin-scale salinity model for the Colorado River Basin, with the following features:
model flow weighted salinity concentrations with uncertainties at locations with salinity standards
develop methodology for operational and long-term policy analysis
Funding provided by the USBR salinity group for USBR and Research Objectives

8. Current Salinity Modeling Methodology
Separation of gauged flow and natural flow
necessary for predictions
gauged flow influenced by changing human factors
natural and human-induced salt
Salt can enter the system from two sources
natural flows
salt loading (predominately agriculture)
Salt modeled as a conservative substance
Reservoirs modeled fully mixed
Monthly timestep

9. Review of Current Model Problems
Inconsistent data and methodologies
status reports outlined new methods to develop the natural flow database and store model inputs and outputs
Calibration problems
quantified the over-prediction throughout the basin
Low-flow problems
Limited uncertainty analysis

10. Calibration Problems

11. Low-Flow Problems

14. Limited Uncertainty analysis
Natural variability of flows
Index sequential modeling
generates synthetic streamflow that exactly match the historical record, shifted in time

17. Research Objectives
Development of a long-term, basin-scale salinity model for the Colorado River Basin, with the following features:
model flow weighted salinity concentrations with uncertainties at locations with salinity standards
develop methodology for operational and long-term policy analysis

18. Literature Review
Water Resource, GEOBASE and GeoRef databases were searched
Multiple keyword searches found a total of approximately 700 records
Articles discussing groundwater, estuary, or salt intrusion models were eliminated leaving approximately 100 articles
Review of these abstracts found 2 papers directly related to our research

19. Directly Related Articles
Basin Scale Water Quality Models for the Colorado River Basin
Stochastic Analysis of Water Quality. Malone, R., Bowles, D., Grenney, W., Windham, M., Utah Water Research Laboratory.
Considered uncertainty with salt loading
A Stochastic Model of River Water Quality: Application to Salinity in the Colorado River. Lee, D., Howitt, R., Marino, M., Water Resources Research.
Considered uncertainty due to natural variability of flow

20. Additional Papers Reviewed
Water Quality Models with Uncertainty
Modeling Salt Processes
Modeling Methods for Uncertainty

21. Uncertainty in Water Quality
greatest
uncertainty
least
uncertainty

22. Uncertainty in Water Quantity
greatest
uncertainty
least
uncertainty

23. Separation of Natural Flow from Gauged Flow
Need separation for predictions ?
Can historical gauged flow be used for predictions ?
Introduces additional uncertainty
Need to separate natural salt load from gauged salt load
neither natural nor human-induced salt loading can be directly measured

24. Representation of Stochastic Hydrology
Natural variability
Sustained low-flow periods
Time series analysis
Stochastic Analysis, Modeling and Simulation (SAMS)
generates synthetic streamflow sequences that have not occurred in the past, but are statistically possible
critical low-flow analysis
Could another type of analysis be used?

25. Detailed Reach Study Upper Colorado and Gunnison River
most complete water quantity dataset
calculation of salinity pickup in the Grand Valley
Research a process to describe salt loading into the system and associated uncertainty

26. Calibration and Use of Model for Projections
Current calibration is only a patch
removes over-predicted salt with calibration nodes
Development of a final calibration will not be complete until a dataset for the basin is developed
Projections
SAMS to represent natural variability
land use as a parameter in predictive model

27. Calibrated Model

30. Research Issues to Discuss
SAMS for representation of natural variability
Separating natural salt load from gauged salt load
neither natural nor human-induced salt loading can be directly measured
difficult to produce good prediction or quantified uncertainty
Developing a process to describe salt loading into the system and associated uncertainty
Modeling changes in salt loading due to future land use changes