Reconciling Disparate Trends in High Elevation Streamflow and Low Elevation Precipitation

advertisement
Reconciling Disparate
Trends in High Elevation
Streamflow and Low
Elevation Precipitation
Charles Luce, US Forest Service
Zachary Holden, US Forest Service
John Abatzoglou, University of Idaho
Greg Pederson, US Geological Survey
Viviana Lopez-Burgos, US Forest Service
Annual Runoff compared to 1980-1999
(SRES A1B 2090-2099)
IPCC AR 4 SYR 3.5
Reconciling Trends
•
•
•
•
•
•
The Dilemma – “Watching P’s and Q’s”
Is it Evapotranspiration?
Limited Observational Support for P
Related discrepancies
Why the discussion is important!
Informing projections
Declining PNW
Streamflows
Fu et al., 2010
Luce and Holden, 2009
Significant at
Not Significant at
0.1
0.1
-4.3 % to 4.3 %
-12.9 % to -4.3 %
-21.5 % to -12.9 %
-30.1 % to -21.5 %
-38.7 % to -30.1 %
-47.3 % to -38.7 %
Clark , 2010
Trends in HCN2 Precipitation (% change in 25th %ile)
27.3 % to 35 %
19.6 % to 27.3 %
12 % to 19.6 %
4.3 % to 12 %
-3.4 % to 4.3 %
-11.1 % to -3.4 %
-18.7 % to -11.1 %
-26.4 % to -18.7 %
-34.1 % to -26.4 %
-41.8 % to -34.1 %
Reconciling Trends
• The Dilemma – Watching P and Q
• Is it Evapotranspiration?
– Discussion of trend patterns
– Discussion of trend magnitudes
– An “experiment”
•
•
•
•
Limited Observational Support for P
Related discrepancies
Why the discussion is important!
Informing projections
700
400
500
600
75th %ile
50th %ile
200
300
Annual Flow (mm)
800
BOISE
NR TWIN
SPRINGS
ID Yield Quantiles
Middle
ForkRBoise
– Trend
in Water
25th %ile
P.25=0.02
1950
Luce and Holden, 2009
1960
1970
1980
1990
2000
a)
c)
25thth
%ile
25
– Dry Years
b)
75thth
%ile
75
– Wet Years
d)
50th %ile
50th
Mean
Mean
Luce and Holden, 2009
E=P
1
Q/P
E/P
0.8
0.6
E=Rn
0.4
0.2
0
0
0.5
Wet
Budyko, 1974
1
1.5
2
Rn/P
2.5
3
Dry
3.5
0
Low
Sensitivity
Changes in Net Radiation
most strongly affect flows in
wet places and wet years
dQ/dRn
-0.2
-0.4
-0.6
-0.8
High
Sensitivity
-1
0
Wet
0.2
0.4
0.6
E/P
Milly and Dunne, 2002
0.8
Dry
1
Implied Form of ET Driven Trend
Convergent
S75 /S25 > 1
Q
S75
S25
Year
8
6
4
0:8
0
2
Fraction of Decline Potentially
Explained by Evaporation
S75/S25
>1
.8 – 1 S25
.6 - .8 S25
.4 - .6 S25
.2 - .4 S25
0 - .2 S25
S75 > 0
-1.0
-0.5
1.8 W/m2
of Evaporation
-1.5
Change in 25th %ile Runoff (mm/day)
0.0
Energetics of Streamflow Change
>35% decline
0
1000
2000
3000
Average Annual Runoff (mm)
4000
Middle Fork
Boise R. –
45% Burned
South Fork
Boise R.
Contrasting fire related flow
increase to long term flow decrease
40
*
*
.
20
3
Change in Monthly Yield (MM m )
*
*
*
*
0
*
-20
-40
-60
Climate Change Only
Fire Only (45% of basin)
-80
-100
*
*
-120
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Reconciling Trends
• The Dilemma – Watching P and Q
• Is it Evapotranspiration?
• Limited Observational Support for P
– Most obs at lower elevations
– Long-term high-elev stations rare
– Examine short-term sensitivity
• Related discrepancies
• Why the discussion is important!
• Informing projections
3000
1000
2000
Watershed
Max
Mean
Min
0
Elev (m)
Precipitation
Gage
-124
-122
-120
-118
Longitude
-116
-114
2000
1500
Elevation (m)
1000
500
0
-124
-122
-120
-118
Longitude
-116
-114
-112
900
800
600
700
75th %ile
500
50th %ile
400
Water Year Precipitation (mm)
Idaho City Annual Precipitation (~1500 m)
25th %ile
1950
1960
1970
1980
1990
2000
5.0
4.5
4.0
• 30 Station pairs ID & MT
• Ratio is non-constant
• Ratio higher in wet years
Nonlinear relationship of
low-elev to high-elev P
3.5
Mean OP ratio (NDJFM)
5.5
Orographic
Precipitation
Ratio vs.
ENSO
Amplification of low-elev
trends
25
26
Wetter
27
28
Nino3.4 (NDJFM)
Holden et al., in prep, also see Dettinger et al., 2004
29
Drier
Change in pptn
per SD change in
Multivariate
ENSO Index
Inc. ENSO =>
Increased P at
low elev
Decreased P at
high elev
Image courtesy of John Abatzoglou
Reconciling Trends
•
•
•
•
The Dilemma – Watching P and Q
Is it Evapotranspiration?
Limited Observational Support for P
Related discrepancies
– April 1 SWE - Trend v. Sensitivity
– Streamflow Timing – Trend, Sensitivity, Stats
• Why the discussion is important!
• Informing projections
Historical April 1 Snowpack (SWE)
April 1 SWE (cm)
(1950-1999)
- +
Regonda et al, 2005
see also Mote et al, 2005
Sensitivity of April 1 SWE
Mote et al., 2005
See also Mote, 2006
Relative Apr 1 SWE Sensitivity for temperature and precipitation
Pale blue indicate stronger precipitation sensitivity
Darker red indicates stronger temperature sensitivity
Image courtesy of V. Lopez-Burgos
Streamflow Timing
Stewart et al., 2005
16 Jun
Center of Timing
01 Jun
16 May
01 May
16 Apr
01 Apr
Salmon R nr Salmon
16 Mar
100
Luce and Holden, 2009
150
200
250
Annual Streamflow (mm)
Reconciling Trends
•
•
•
•
•
•
The Dilemma – Watching P and Q
Is it Evapotranspiration?
Limited Observational Support for P
Related discrepancies
Why the discussion is important!
Informing projections
SWE
Regonda et al.,2005
Stewart et al.,
2005
60.00
Change in Monthly Yield (millions m3)
40.00
20.00
0.00
-20.00
-40.00
Modeled Flow Changes
1990s-2080s
-60.00
Observed Flow Changes
1950s-1990s
-80.00
-100.00
-120.00
Oct
Nov Dec
Jan
Feb Mar Apr May Jun
Jul
Aug
Sep
1972-2003 large forest fires
For years with early snowmelt
Westerling et al, 2006
Burned Area Sensitivity
Fire Extent
Temperature
Runoff
Timing
Westerling et al., 2006, 2011
Burned Area Sensitivity
Precipitation
Annual
Runoff
Fire Extent
Temperature
Runoff
Timing
Holden et al., 2011
Burned Area Sensitivity
Precipitation
q = 0.49
Annual
Runoff
Fire Extent
= 0.53
Temperature
Runoff
Timing
q = 0.17
Holden et al., 2012
Reconciling Trends
•
•
•
•
•
•
The Dilemma – Watching P and Q
Is it Evapotranspiration?
Limited Observational Support for P
Related discrepancies
Why the discussion is important!
Informing projections
Annual Runoff compared to 1980-1999
(SRES A1B 2090-2099)
IPCC AR 4 SYR 3.5
Reconstruction of Missouri River at Toston, 1676-1996
Smoothed with a 10-year running average
(the last year of 10 is plotted)
Woodhouse, Pederson et al., 2008
Western US mean Streamflow Trends 1950-99
Regonda et al., 2005
0.3
Average AIC Weight
0.25
0.2
0.15
0.1
0.05
0
Time
ENSO
PDO
T+ENSO
T+PDO ENSo+PDO
Variables in Linear Model for 25th %ile Flow
All 3
-2
400
500
0
300
1
2
1920
1940
1960
Year
1980
2000
PDO
600
-1
200
MF Boise Streamflow (mm)
700
800
25th%ile Streamflow
PDO
Darwin Sea Level Pressure
Matter et al., in revision
Orographic Precipitation
Orographic Precipitation Fall
Velocity
Fall speed
increases may
cause orographic
precipitation to fall
lower on
windward slopes.
Pavelsky et al., 2012
Ocean-Land Contrast v. CO2
250
17
350
14
13
CO2
OLC
OLC-loess
1900
1920
Ocean-Land Contrast (C)
300
15
400
CO2 (ppm)
16
12
1940
1960
1980
2000
Year
Karamperidou et al., 2010, See also Braganza et al., 2004
Near-surface specific humidity varies similarly over land
and sea, suggesting that the recent reduction in relative
humidity over land may be due to limited moisture
supply from the oceans, where evaporation has been
limited by sea surface temperatures that have not risen
in concert with temperatures over land.
Simmons et al., 2010
Reconciling Trends
•
•
•
•
The Dilemma – ΔQ ≠ ΔP
Is it Evapotranspiration? – No
Limited Observational Support for P
Flow, timing, and SWE trends consistent
with a decrease in P
• Important consequences!
• Constraining theories
– Does the new information on elevational
differences help discern alternative models?
RMRS GTR 290
At the printers now
http://www.fs.fed.us/rm/pubs/rmrs_gtr290.html
http://www.treesearch.fs.fed.us/pubs/41932
RECONCILING DISPARATE TRENDS IN HIGH-ELEVATION HYDROLOGY AND LOW-ELEVATION PRECIPITATION MEASUREMENTS: IS
PRECIPITATION DECLINING IN PACIFIC NORTHWEST MOUNTAINS?
Luce, Charles H. (1), Holden, Zachary A. (2), Abatzoglou, John (3), and Pederson, Greg (4)
(1) USDA Forest Service R&D, Boise, ID 83702, (2) USDA Forest Service, Region 1, Missoula, MT 59802, (3) University of Idaho,
Moscow, ID 83844, (4) USGS Northern Rocky Mountain Science Center, Bozeman, MT 59715
IPCC reports have repeatedly noted that mountain landscapes are among the most difficult places to predict the effects of climate
change on water resources. Declining snowpacks, earlier streamflow timing, and declining streamflow are a few of the trends that
have been noted in the mountains of the Pacific Northwest over the last 60 years. Efforts to answer questions regarding causality of
the trends are partially hampered by the lack of uninterrupted high elevation precipitation data spanning this period. The large
observational network of precipitation gages primarily located in lower-elevation valleys (USHCNv2) has seen only slight declines over
the same period, suggesting that most of the trends have been related to temperature increases. An important assumption behind
that conclusion, however, is that high elevation precipitation is strongly correlated with low elevation precipitation. We present a
series of interrelated analyses demonstrating that declines in orographic enhancement of precipitation offer an additional mechanism
for describing important geographical and temporal patterns in observed trends and sensitivities. We also offer theory and
observations supporting the idea that trends in orographic precipitation may be a manifestation of warming air temperatures,
potentially representing a second pathway for climate change effects in mountain landscapes. The distinction in process is important.
Many ecological processes are much more sensitive to water balances than temperature, and the combined effects may be more
disruptive than anticipated. Furthermore, choices about adaptation for water resource management are sensitive to mechanisms
causing shifts in timing.
Standardized Regression coefficients for Apr 1 SWE vs Winter Precipitation
SNOTEL Stations
Standardized Regression coefficients for Apr 1 SWE vs Average winter Wet Day Tmin
SNOTEL Stations
6
instrumental mean
reconstructed mean
2 error bars
5
3
3
meter year x 10
9
4
2
1
0
1700
1750
1800
1850
1900
1950
2000
Yellowstone R. flows reconstructed from tree rings
Graumlich et al., 2003
APR 1 SWE
Pederson et al, Science, 2011
MCA
LIA
Modern
Download