Transcript Slide 1

Susceptibility of Colorado River basin to megadroughts in a warming climate
Dennis Lettenmaier
Project Overview
Soil Moisture Conditions in November 2007 (expressed as percentiles relative to 1915-2003
climatology) show widespread drought, particularly in the Colorado River basin (blue circle).
1.6
AVG_PRECIP
1.4
1.2
(mm/day)
EVAP
1
0.8
P- E
0.6
0.4
Task 1: Preliminary Results
The Colorado River, a critical water resource to the southwestern United
States, has been experiencing dry conditions that have now persisted for
almost a decade. The ongoing drought has the inferred severity of the
“megadrought” of the 1100s, thought to be the most severe drought within
the last 1300 years, and corresponds to a 25-year mean that is 85 percent of
the naturalized flow of the 100+ years of the stream gauge record, and
persisted for several decades. IPCC AR4 GCM studies suggest, averaged
across many models, that Colorado River flows by the middle of this century
will be of order 70 percent of the 20th Century mean, and will further decline
by the end of the century. If correct, these simulations suggest that the
Colorado River basin will enter a state of perpetual drought more severe
than the megadrought of the 1100s. However, other approaches using IPCC
AR4 climate scenarios, but with statistical downscaling to prescribe off-line
forcing to a land hydrology model, concluded that decreases in Colorado
River runoff by mid-century will be as little as 6 percent. When adjusted for
methodological differences, this resulted in inferred runoff decreases of
about 13 percent in the annual average. By using a regional climate
modeling framework, we address shortcomings that underlie both of the
earlier approaches, and further assess:
0.2
.
2097
2091
2085
2079
2073
2067
2061
2055
2049
2043
2037
2031
2025
2019
2013
2007
2001
0
YEAR
Multimodel averages over the Colorado River basin from 19 IPCC AR4 GCMs, A2 emission
scenarios. Note drastic decline in P-E by 2100. Model Output courtesy Richard Seager.
UW Civil
and
Environmental
Engineering
Will the Colorado River basin, and the U.S. Southwest more
generally transition to a permanent megadrought state over the
next century, and what role will land-atmosphere feedbacks play
in those changes relative to remote climate forcings?
Complementary Science Questions
1. How are interseasonal, interdecadal, and future projected climate change in the land-atmosphere dynamics of the Colorado River
basin affected by feedbacks associated with a transition to early snowmelt, and hence occurrence of maximum soil moisture earlier
in the year when evaporative demand is lower?
2. How might perpetual “climate change drought” associated with reduced moisture flux convergence over the Colorado River basin
compare with pre-instrumental megadroughts inferred to have occurred in the paleo record of the last 1000 years?
3. How do the atmospheric dynamics and land-atmosphere feedbacks of recent drought years in the Colorado basin compare with
projected long-term changes over the next century, i.e., to what extent can recent drought years be considered a harbinger of
future conditions?
Precipitation Elasticity and Temp Sensitivity
Spatial Distribution of Temperature Sensitivities
In the upper Colorado River
basin, most increases in
temperature
result
in
decreases in runoff. There
are,
however
specific
locations that have runoff
increases. The figure on the
left indicates temperature
sensitivities in VIC model
simulations where Tmin and
Tmax were both increased
by equal amounts. Note the
spatial distribution of runoff
increases
(red),
the
mechanisms behind these
increases will be explored in
future work.
TempTempPrecipitation- sensitivity
Model
sensitivity
Elasticity
(Tmin &
0C
(
Tmax)
%/
Tmax ) %/ 0C
VIC
NOAH
SAC
1.9
1.81
1.77
-2.2
-2.85
-2.65
-3.3
-3.93
-4.10
Precipitation
elasticities
and
temperature
sensitivities (for annual discharge at Lees Ferry) for
VIC, NOAH, and SAC models. While precipitation
elasticities
are
quite
similar,
temperature
sensitivities vary greatly depending on whether the
change in temperature is applied to the entire range
or just to Tmax.
Seasonal Runoff
Composite seasonal water balance for the 25% of the
Colorado River basin that produces most (about 75%)
of its runoff, by quartile of the temperature sensitivity
distribution (lowest quartile has the smallest (most
negative) temperature sensitivities. Black line is for
runoff, and shows that those grid cells with smallest
(most negative) temperature sensitivities tend to
produce the most runoff, contrary to the Dettinger
Hypothesis. Results are for the VIC model, but results
for NOAH and SAC (Sacramento) models are
qualitatively similar.
Methods
Next Steps
In previous studies, the GCM approach was unable to resolve highly skewed (towards high elevation areas) source
areas for runoff like the western U.S, whereas the land hydrology model approach decoupled the hydrologic forcing
with the atmospheric circulation. To overcome these shortcomings, we use a regional climate model (Weather
Research and Forecasting model) coupled with a land hydrology model (Variable Infiltration Capacity model).
Through collaboration with Dr. Rudy Leung of the Pacific Northwest National Laboratory, we will continue our efforts to
understand how increasing global greenhouse gas concentrations over the next century will change the susceptibility
of the Colorado River basin to megadroughts.
Driving data: Historical conditions (North
American Regional Reanalysis and/or
CMIP 20th century model runs) for
boundary conditions as well as boundary
conditions taken from 21st Century (IPCC
AR4) model simulations. Sea surface
(temperature) boundary conditions will be
as in the North American Climate Change
Application
Project
(NARCCAP).
NARCCAP domain is indicated (above),
colors show vegetation types.
Our next step is to implement the coupled WRF-VIC model over the NARCCAP domain (continuation from Task 1).
WRF-VIC will then be implemented for the 20th century GCM simulations and for future simulations. We will compare
differences between the 21st and 20th century simulations and will particularly compare differences between the
coupled WRF-VIC simulations of runoff, as well as snow and soil moisture storage, precipitation, and
evapotranspiration between the regional climate runs and the GCM from which the boundary conditions will be taken.
Our working hypothesis is that evapotranspiration will be less, runoff greater, and possibly that precipitation will be
greater in the regional model runs than in the GCMs. Therefore, we further hypothesize that the magnitude of
projected 21st century hydrologic change over the Colorado basin will be less in the higher resolution regional climate
model runs than in the GCMs. The next step will be performing diagnostic studies of the nature of land hydrologic
feedbacks as they are predicted to affect runoff response to climate change in the Colorado River Basin. Finally, we
will evaluate the nature and implications of land surface feedbacks in recent drought years, focusing on water years
(Oct-Sep) 2002 and 2004.
Land surface model:
The
Variable Infiltration Capacity
(VIC) is a macroscale energy
and water balance model
+
Regional climate model:
The Weather
Research and Forecasting (WRF) is a
mesoscale numerical weather and climate
prediction system.
The primary mechanism for dissemination of results will be through publication in the refereed literature. We will
maintain an archive of model output, which will be made available to the broader community.
Research funding provided by Department of Energy