1. Forecasting the Long-Range Implications of the 2001 Drought for the Columbia River Basin
(April, 2001-September, 2002)
JISAO Climate Impacts Group and the
Department of Civil Engineering
University of Washington
October, 2001
Alan F. Hamlet
Andy Wood
Dennis P. Lettenmaier

2. Hydrologic Characteristics of the Columbia Basin
Avg Naturalized Flow
The Dalles
Elevation (m)

3. Pacific Decadal Oscillation El Niño Southern Oscillation
A history of the PDO
A history of ENSO
warm
warm
cool

4. Effects of the PDO and ENSO on Columbia River
Summer Streamflows
Cool
Cool
Warm
Warm

5. Overview of Long-Range Forecasting Methods
and Long-Range Forecasts for Water Year 2002

6. Overview of Modeling Linkages
Observed Meteor.
Data
Optimization or
Simulation
Global Climate
Models
Water
Resources
Models
Hydrology
Models
streamflow
temp
precip
wind
downscaling
Regional Climate
Models
water demand

7. Variable Infiltration Capacity (VIC) Model

8. Time Line for 12-month Lead Time Retrospective Forecasts
Climate
Forecast
Estimated
Initial Conditions
via observed met. data
Forecast
Ensemble
Lead time = 12 months

9. Experimental Forecasts for 2002 Based on Resampled Observed Driving Data
2002 Forecast Construction Details:
Hindcast hydrologic state for August 1, 2001 based on observed weather data
VIC driving data from August 1- Sept 30 is taken from 15 observed water years from associated with winter ENSO neutral conditions in the tropics.
This produces 15 forecast ensembles from Oct, 2001 to September, 2002

10. Raw streamflow forecast ensemble at The Dalles compared to 50-year “observed” min, max and mean monthly simulated streamflow. All ENSO neutral.
Highest Observed
Observed Long Term Mean
Lowest Observed

11. Summary of raw streamflow forecast ensemble compared to 50-year “observed” min, max and mean monthly simulated streamflow {whiskers show 90th and 10th percentiles of forecast}
Natural Flow at The Dalles

12. Cool PDO/ENSO Neutral Raw Forecast

13. Experimental Climate Model Driven Forecasts for 2002
2002 Forecast Construction Details:
Hindcast hydrologic state for August 1, 2001 based on observed weather data
VIC driving data from August 1, February 28, 2002 are taken from 20 climate global climate model ensembles.
VIC driving data from March 1- September 30, 2002 are based on median 1953 meteorological data
This produces 20 forecast ensembles from Oct, 2001 to September, 2002

14. Climate Model-Based 20-member streamflow forecast at The Dalles
1953 driving data for March-Sept

15. Summary of raw streamflow forecast ensemble compared to 50-year “observed” min, max and mean monthly simulated streamflow {whiskers show 90th and 10th percentiles of forecast}

16. Effects to System Storage
Retrospective Forecast Method

17. and System of Dams and Reservoirs Included in ColSim Model
Columbia River Basin
and System of Dams and Reservoirs Included in ColSim Model
Storage Reservoirs
Run-of-River Dams
ColSim includes about 95% of the basin’s hydropower, and simulates simplified reservoir operations at 1995 level of development.

18. Simulated System Refill for Several Historic Drought Years
and Observed Refill for 2001
1977 streamflow sequence
Expected Storage
following 1977
drought
Observed Storage, 2001

19. Trends in Regulated Peak Flow at The Dalles
Completion of Major Dams

20. Forecast System Storage: Oct, 2001- Sept, 2002
Normal Winter Energy is met in 3 out of 15 simulations (20% reliability)
full
Observed System Storage Sept 1, 2001
Obs
refill sequence
Observed System Storage April 1, 2001
empty

21. Summary of System Storage Simulation
{whiskers are 10th and 90th percentiles}
reference refill

22. Ensemble Average Energy Shortfalls from ColSim
Associated with Streamflow Forecast

23. Conclusions
Drought conditions in 2001 will increase the likelihood of below average streamflows in 2002 due to reduced soil moistures.
Limited summer refill in 2001 will produce relatively high vulnerability to capacity-related hydro system shortfalls in winter 2002, and will likely also prevent complete reservoir refill in summer Normal winter energy production from the hydro system in the ColSim reservoir model is shown to be about 20% reliable in the simulations, with most shortfalls occurring in Jan and Feb.
Complete reservoir refill is unlikely in summer 2002.
The aftermath of the 2001 drought and the west coast power crisis puts an interesting emphasis on winter weather conditions in the context of meeting PNW winter power demand in Best case would be warm, wet conditions in mid winter (low power demand, high winter runoff). Worst case would be very cold conditions, which would produce low runoff (even for wet conditions) and high power demand.