Transcript lettenmaier_colorado_flow_reconciliation_nov08

RECENT FINDINGS ON RUNOFF
SENSITIVITY OF COLORADO RIVER
DISCHARGE TO CLIMATE CHANGE
Dennis P. Lettenmaier*
Department of Civil and Environmental Engineering
University of Washington
Reconciling Colorado River Flows Stakeholder Meeting
Southern Nevada Water Authority
Las Vegas, NV
November 14, 2008
*with contributions from Hugo Hidalgo (SIO), Tazebe Beyenne (UW), and Kostas Andreadis (UW)
Outline of this talk
1. The history of conflicting estimates
2. Understanding the hydrologic sensitivities
3. Unanswered questions
Postmortem: Christensen and Lettenmaier (HESSD,
2007) – multimodel ensemble analysis with 11 IPCC
AR4 models (downscaled as in C&L, 2004)
PCM Projected Colorado R. Temperature
PCM Projected Colorado R. Precipitation
Question: Why such a large discrepancy in
projected Colorado River flow changes?
• ~6=7% annual flow reduction in Christensen and
Lettenmaier (2007)
• 10-25% by Milly et al (2005)
• > 35% by Seager et al (2007)
Magnitude and Consistency of Model-Projected Changes
in Annual Runoff by Water Resources Region, 2041-2060
Median change in annual runoff from 24 numerical experiments (color scale)
and fraction of 24 experiments producing common direction of change (inset numerical values).
58%
+10%
67%
62%
58%
96%
+2%
62%
62%
71%
87%
-2%
75%
100%
67%
67%
67%
-5%
-10%
-25%
(After Milly, P.C.D., K.A. Dunne, A.V. Vecchia, Global pattern of trends in streamflow and
water availability in a changing climate, Nature, 438, 347-350, 2005.)
Decrease
87%
+5%
Increase
+25%
from Seager et al, Science, 2007
Diagnosis

Wood et al (2002; 2004) downscaling method
removes bias by mapping from PDF of GCM
output to PDF of observations on a monthly
basis

PDFs are estimated for each grid cell and
month of the year

This same mapping is then applied to the
future climate run.

The method does not attempt to preserve
GCM inferred differences in precipitation.

There is in general no reason to assume that
the GCM precipitation changes are applicable
to higher spatial resolutions
CL2007 Re-runs
All precipitation values were rescaled so as to match GCM
changes on an annual basis
This resulted in a change (reduction) in mean annual
precipitation for 2040-2070 from 1.9% (CL2007) to 2.6% for
A2 emissions scenario (closest to A1B used in M2005 and
S2007)
The associated annual mean runoff reduction (Imperial
Dam, averaged over 11 GCMs) changed from 5.9 to 10.0%
This is within (although at the lower end of) the range
reported in M2005
Note that M2005 and S2007 use the A1B IPCC emissions
scenario, vs A2 scenario used by CL2007
M2007 and S2007 use (partially) different GCM runs and
procedures (M2005 count multiple ensembles from a single
GCM as separate runs
Understanding the hydrologic sensitivities
Dooge (1992; 1999):
where ΨP is elasticity of runoff with respect to precipitation
For temperature, it’s more convenient to
think in terms of sensitivity (v. elasticity)
Inferred runoff elasticities wrt precipitation for major Colorado River
tributaries, using method of Sankarasubramanian and Vogel (2001)
Visual courtesy Hugo Hidalgo, Scripps Institution of Oceanography
Unconditional histograms of 1/8 degree grid cell
precipitation elasticities from model runs of three
land surface models (VIC, NOAH, and SAC) over
Colorado Basin for 20 years, 1985-2005
VIC
NOAH
SAC
Summary of precipitation elasticities and
temperatures sensitivities for Colorado River at
Lees Ferry for VIC, NOAH, and SAC models
Model
Precipitation Temp-Elasticity
sensitivity
(Tmin &
Tmax ) %/
0C
Tempsensitivity
( Tmax)
%/ 0C
Flow @
Lees
Ferry
(MACF)
VIC
1.9
-2.2
-3.3
15.43
NOAH 1.81
-2.85
-3.93
17.43
SAC
-2.65
-4.10
15.76
1.77
VIC
Precipitation
elasticity
histograms, all
grid cells and
25% of grid cells
producing most
(~73%) of
runoff
Spatial distribution
of precipitation
elasticities
Censored
spatial
distribution of
annual runoff
Composite seasonal water cycle, by quartile
of the runoff elasticity distribution
Temperature
sensitivity (equal
change in Tmin
and Tmax)
histograms, all
grid cells and 25%
of grid cells
producing most
(~73%) of runoff
Censored spatial
distribution of
annual runoff
Spatial distribution of
temperature
sensitivities (equal
changes in Tmin and
Tmax)
Composite seasonal water cycle, by quartile of the temperature
sensitivity (equal change in Tmin and Tmax) distribution
Temperature
sensitivity (Tmin
fixed) histograms,
all grid cells and
25% of grid cells
producing most
(~73%) of runoff
Censored spatial
distribution of
annual runoff
Spatial distribution
of temperature
sensitivities (Tmin
fixed)
Composite seasonal water cycle, by quartile of the
temperature sensitivity (fixed Tmin) distribution
So is there, or is there not, a dichotomy between
the various estimates of mid-century Colorado
River runoff changes?
Replotted from Seager et al (2007)
a) Lowest mid-century estimate (Christensen and
Lettenmaier, 2007) is based on a precipitation
downscaling method that yields smaller midcentury precipitation changes. Adjusting for this
difference nearly doubles the projected change to
around 10% by mid century – not far from Milly et
al (2005), but still well below Seager et al (2007)
b) On the other hand, from Seager et al (2007), very
roughly, mid-century ΔP  -18%, so for
= 1.5-1.9,
and temperature sensitivity  -0.02 - -0.03, and ΔT 
2 oC, ΔQ  35% (vs > 50% + from GCM multimodel
average)
More important, though, is the question: In the
context of hydrologic sensitivities to (global) climate
change, does the land surface hydrology matter, or
does it just passively respond to changes in the
atmospheric circulation?
i.e., in the long-term mean, VIMFC  P-E  Q, so do we
really need to know anything about the land surface to
determine the runoff sensitivity (from coupled models)?
OR is the coupled system sensitive to the spatial variability in
the processes that control runoff generation (and hence ET),
and in turn, are there critical controls on the hydrologic
sensitivities that are not (and cannot, due to resolution
constraints) be represented in current coupled models?
The answer …
… Probably lies in high resolution, coupled landatmosphere simulations, that resolve areas
producing most runoff, and their role in modulating
(or exacerbating) regional scale sensitivities.