1. Late 20th Century Precipitation Variability in the Western U. S
Late 20th Century Precipitation Variability in the Western U.S. in the Context of Long-Term Climate Variability and Global Change
Alan F. Hamlet
Anthony L. Westerling
Tim P. Barnett
Dennis P. Lettenmaier
JISAO/CSES Climate Impacts Group
Dept. of Civil and Environmental Engineering
University of Washington
Scripps Institute of Oceanography
School of Engineering, University of California, Merced

2. Cool Season Climate of the Western U.S.
PNW GB CA
CRB
DJF Temp (°C)
NDJFM Precip (mm)

3. Schematic of VIC Hydrologic Model and Energy Balance Snow Model
PNW CA
CRB GB
Snow Model

4. Evaluation of Streamflow Simulations of the Colorado River at Lee’s Ferry, AZ

5. Cool Season Precipitation Explains Most of the Variability
in Annual Flow in the PNW and CA
R2 = 0.83
Relationship Between Annual Flow and
Cool Season Precip.
Columbia River
R2 = 0.91
Relationship Between Annual Flow and
Cool Season Precip.
Sacramento River

6. Cool Season Precip Explains Most of the Variability in Annual Flow in the CRB, but the Summer Monsoon Also Plays a Role
R2 = 0.56
Relationship Between Annual Flow and
Cool Season Precip.
Colorado River
R2 = 0.18
Relationship Between Annual Flow and
Warm Season Precip.
Colorado River

7. Consensus Forecasts of Temperature and Precipitation Changes from IPCC AR4 GCMs

8. Observed 20th century variability
Curves are fits to ln(CO2) for A2 (solid) and B1 (dashed)
Warming ranges are shown for 2020s, 2040s and 2090s relative to 1990s. Central estimates: 0.7C by 2020s, 1.7C by 2040s, 3.2C by 2090s. Pink box shows +/- 2 sigma for annual average temperature (sigma=0.6C). Red lines show previous generation of change scenarios.
Pacific Northwest

9. Observed 20th century variability%
+6%
+2%
+1%
Curves are fits to ln(CO2) for A2 (solid) and B1 (dashed)
Precip changes are shown for 2020s, 2040s and 2090s relative to 1990s. Central estimates: 1% by 2020s, 3% by 2040s, 6% by 2090s. Pink bar shows +/- 2 sigma for PNW annual precip.
Observed 20th century variability
-1 to +3%
-1 to +9%
-2 to +21%
Pacific Northwest

10. Regionally Averaged Cool Season Precipitation Anomalies

11. Summary Statistics for Regionally Averaged Cool Season

12. Regionally Averaged Warm Season Precipitation Anomalies

13. Simulated Changes in System Wide Energy Production
in the Western U.S.
Correlation:
CRB-SSJ = 0.07
CRB-PNW = 0.08
SSJ-PNW = 0.36
Correlation:
CRB-SSJ = 0.14
CRB-PNW = -0.14
SSJ-PNW = 0.06
Correlation:
CRB-SSJ = 0.73
CRB-PNW = 0.51
SSJ-PNW = 0.65
Fig 7 Annual hydropower production has followed suit. These results suggest reduced opportunity for conjunctive management of the West’s hydropower resources in recent decades.

14. 20-year Flood for “ ” Compared to “ ” for a Consistent Late 20th Century Temperature Regime
DJF Avg Temp (C)
X20 ’73-’03 / X20 ’16-’03
Hamlet A.F., Lettenmaier D.P., 2007: Effects of 20th Century Warming and Climate Variability on Flood Risk in the Western U.S., Water Resour. Res., 43, W06427
X20 ’73-’03 / X20 ’16-’03

15. Are the changes in variability that have been observed in the last third of the 20th century consistent with normal patterns of variability?

16. Long-Term Comparison of Annual Flow Records from
Observations and Paleo Reconstructions
PNW:
Observed (naturalized) annual flow in the Columbia River at The Dalles, OR
(reconstructed from observed peak river stage)
(naturalized from observed monthly records)
CA:
Reconstructed combined annual flow in the Sacramento/San Joaquin basin from tree-ring records.
(Overlapping period )
(Meko, D.M., 2001: Reconstructed Sacramento River System Runoff From Tree Rings, Report prepared for the California Department of Water Resources, July)
Colorado River Basin:
Reconstructed annual flow in the Colorado River at Lees Ferry, AZ from tree ring records.
(Woodhouse, C.A., S.T. Gray, and D.M. Meko, 2006: Updated Streamflow Reconstructions for the Upper Colorado River Basin, Water Resources Research, Vol. 42, W05415)

17. Changes in Streamflow Variability from Long-Term Observations and Paleo Reconstructions (1858-1977)

18. Changes in Streamflow Variability from VIC Simulations of Annual Flow (1916-2003)

19. All three metrics high together
Changes in Streamflow Variability from Combined Paleo Reconstructions and VIC Simulations of Annual Flow ( )
All three metrics high together

20. What about changes in ENSO and PDO as possible explanations?

21. Natural Flow Columbia River at The Dalles
Patterns of ENSO Related Variability About a Shifting Long-Term Mean Seem to be Robust in the 20th Century
Natural Flow Columbia River at The Dalles

22. Could ENSO explain the lag1 and interregional metrics being anti-correlated?
What about the most recent behavior?
In periods of especially strong (weak) controls on cool season storm track behavior associated with ENSO (i.e. strong or weak NW/SW bipole), both interregional and lag1 autocorrelation would tend to be LOW (HIGH) at the same time.
The data, however, show that typically lag1 autocorrelation and interregional correlation are anti-correlated for the West as a whole. So it would seem that variations in the strength of the ENSO related NW/SW dipole does not provide an explanation of the typical behavior over most the record.
Coupled with the fact that there is little compelling evidence to suggest a systematic change in ENSO telleconnections, it seems that both the explanation for the typical behavior and the most recent changes in variability must lie elsewhere.

23. Cool Season Precipitation Anomalies Compared to the PDO
(Pattern is not robust)
-0.845
-0.264
-0.438
-0.053
(Regional to PDO
Correlation R2 )

24. A working hypothesis:
The most recent changes suggest:
Increasingly unstable storm track in cool season (increased interregional correlation)
Increased lag1 autocorrelation and variation in storm intensity at the scale of the Pacific Rim

25. Are the changes in cool season precipitation variability in the 20th century consistent with GCM projections for the PNW?

26. Overview of GCM Data Processing
IPCC AR4 “A2”
GCM
Simulations
CDFs Match Observations for the Training Period
Large-Scale
Bias Correction at
GCM Grid
Simple Aggregation of GCM cells over the PNW
Upscaling
To PNW

27. *Signal to noise ratio is high*
Change in Pacific Northwest winter temperatures for HadCM3 between and
Change = C
*Signal to noise ratio is high*
-1.64
-2.5

28. *Signal to noise ratio is low*
Change in Pacific Northwest winter precipitation for HadCM3 between and
Change = - 2%
*Signal to noise ratio is low*
628
614

29. *Signal to noise ratio is low*
Change in Pacific Northwest winter precipitation for ECHAM5 between and
Change = + 5.4%
*Signal to noise ratio is low*
571
602

30. Lessons Learned from 20th Century Observations

31. Comparison of 20th century winter temperature observations and 10 bias-corrected IPCC AR4 GCM simulations for the Pacific Northwest
A2 Emissions Scenarios

32. Comparison of 20th century winter precipitation observations and 10 bias-corrected IPCC AR4 GCM simulations for the Pacific Northwest
A2 Emissions Scenarios

33. Evaluating Precipitation Changes Using a GCM Super Ensemble Approach

34. A super ensemble approach applied to nine GCM simulations of PNW winter precipitation for two different 30-year periods.
270 years
270 years

35. Super ensemble CDFs of PNW winter precipitation for four 30 year time slices from nine GCM simulations
Sample Size = 270 years

36. Conclusions
Cool season precipitation is a major driver of annual river flow, hydropower production, and flood risk in the West.
Substantial and persistent changes in cool season precipitation variability have emerged over the West since about 1975, including increased CV, within-region persistence, and inter-regional correlation.
Long-term streamflow reconstructions show that the current changes in variability are very unusual in the context of natural variations over the last 150 years or so, and the changes are broadly consistent with GCM projections of cool season precipitation in the PNW.
Are these systematic changes? Can they be related to changes in circulation associated with greenhouse-forced warming?

37. Thoughts on Planning Implications:
Even if the current precipitation regime in the West is not a systematic change, it is clear that this is something that can emerge suddenly and persist for a long time. I.e. we can expect that there may be analogous periods in the 21st century that we should be prepared to cope with.
Given the relative performance of GCMs in predicting precipitation and the inherently greater noise that is present in precipitation records, I think it is doubtful that we will have any conclusive information about whether these observed changes are related to greenhouse forcing or not.
This suggests to me that flexible approaches based on monitoring may be the only workable options.