Hugo Hidalgo1, Tapash Das1, Dan Cayan1,2, David Pierce1, Tim Barnett1, Govindasamy Bala3, Art Mirin3, Andrew Wood4, Celine Bonfils3, Ben Santer3.

1) Scripps Institution of Oceanography2) United States Geological Survey3) Lawrence Livermore National Laboratory4) University of Washington

Detection and attribution of Climate Change in Streamflow Seasonality of the Western US

Acknowledgments

This research is supported by grants from the Lawrence

Livermore National Laboratory and the California

Energy Commission

Warming already has driven observable hydroclimatic change

Knowles et al.,2006

-2.2 std devsLESS as snowfall

+1 std devMORE as snowfall

Observed: Less snow/more rain

Observed: Less spring snowpack

Observed: Earliergreenup dates

Observed: Observed: Earlier Earlier snowmelt runoffsnowmelt runoff

Slide courtesy of M. Dettinger

TRENDS (1950-97) in April 1 snow-water content at

western snow courses

Mote 2003

Cayan et al. 2001

Stewart et al. 2005

Optimal detection & attribution

● Detection of climate change is the process of identifying if an observed change is significantly different from what would be expected from natural internal climate variability (Hegerl et al. 2006).

● Attribution of anthropogenic climate change is the process of identifying if the observed change is: a) consistent with the type of changes obtained from climate simulations that include external anthropogenic forcings and internal variability and b) inconsistent with other explanations of climate change (Hegerl et al. 2006).

Optimal detection & attribution

● Given a variable for detection, the basic idea is to reduce the problem of multiples dimensions (n) to a univariate or low-dimensional problem (Hegerl et al. 1996).

● In this low-dimensional space, a detection vector is used to compare the observations with the expected climate change pattern.

Previous D&A studies● Ocean heat content (Barnet et al. 2001; Levitus et

al. 2001; 2005; Reicher et al. 2002; Wong et al. 2001),

● ocean salinity (Wong et al. 1999; Curry et al. 2003), ● sea level pressure (Gillet et al. 2003), ● changes in sea ice extent (Tett et al. 1996), ● precipitation (Hegerl et al. 2006; Hoerling et al.

2006), ● tropopause height (Santer et al 2003) ● temperature (Santer et al. 1996a, 1996b; Hegerl et

al. 1996; 1997; 2000; 2006; 2007; Barnet et al. 1999; Tett et al. 1999; 2001; Mitchel and Karoly 2001; Stott et al. 2001; Stott 2003; Zwiers and Zhang 2003; Kalroy et al. 2003; Kalroy and Braganza 2005; Stone et al. 2007).

Modeling● Downscaled to 1/8 degree resolution using

method of constructed analogues (CA) or bias correction followed by spatial disaggregation (BCSD)

● Precipitation, tmax and tmin used as input to the variable infiltration capacity model (VIC; Liang et al. 1994)

● The VIC runoff and baseflow were routed using a computer program by Lohmann et al. (1996) to obtain daily streamflow data for the rivers

● The JFM streamflow fractions were computed

● Fingerprint was computed● The d parameters were computed for all

model runs (d parameter: trends projected into fingerprint).

Defining characteristics of VIC are the probabilistic treatment of sub-grid soil moisture capacity distribution, the parameterization of baseflow as a nonlinear recession from the lower soil layer, and that the unsaturated hydraulic conductivity at each particular time step is a function of the degree of saturation of the soil (Sheffield et al. 2004; Campbell 1974; Liang et al. 1994)

Modeled vs. “Observed” monthly streamflow

SACRAM ENTO R IVER SAN JO AQ UIN R IVER

CO LO RADO R IVER CO LUM BIA R IVER

F igu re 2 . S ca tte rp lo ts o f m o n th ly m o d e led v e rsu s n a tu ra liz ed s tream flo w fo r b as in s in th e w es te rn U S .

“Observed” trends JFM streamflow fraction

PCM anthro trends JFM streamflow fraction

PCM solar volcanic trends JFM streamflow

fraction

FINGERPRINT: FIRST PRINCIPAL COMPONENT OF THE AVERAGE

JFM FRACTIONS FROM THE ANTHRO RUNS

DETECTION PLOT (JFM streamflow fraction)

CONTROL CCSM3-FV (CA)

ANTHROPCM (BCSD)

VIC

NATURALIZED

CONTROL PCM (BCSD)

SOLAR & VOLCANICPCM (CA)

ENSO and PDO out

FINGERPRINT: FIRST PRINCIPAL COMPONENT OF THE AVERAGE

JFM FRACTIONS FROM THE ANTHRO RUNS

DETECTION PLOT (JFM streamflow fraction)

CONTROL CCSM3-FV

ANTHROPCM VIC

NATURALIZED

CONTROL PCM

Center timing of streamflow

CT detection

CONTROL CCSM3-FV (CA)

ANTHROPCM (BCSD) VIC

CONTROL PCM (BCSD)

SOLAR & VOLCANICPCM (CA)

Columbia Out

Detection plot (JFM streamflow fraction)

Columbia out

CONTROL CCSM3-FV (CA)

ANTHROPCM (BCSD) VIC

NATURALIZED

CONTROL PCM (BCSD)

SOLAR & VOLCANICPCM (CA)

MIROC runs

Detection plot (JFM streamflow fraction)

CONTROL CCSM3-FV

ANTHROPCM VIC

NATURALIZED

CONTROL PCM

SOLAR & VOLCANIC

PCM

ANTHROMIROC

CALI

COLO

COLU

MIROC, RUN 8

Summer detection plot

CONTROL CCSM3-FV

ANTHROPCM

VIC

NATURALIZED

CONTROL PCM

SOLAR & VOLCANIC

PCM

ANTHROMIROC

CONCLUSIONS

● Detection of climate change was found for the JFM streamflow fractions over the second half of the 20th century.

● There was a positive detection even in the case when the ENSO and PDO signals were regressed-out.

● Attribution of climate change for JFM streamflow turned out to be model dependant: PCM implies it, but for MIROC the jury is out.

Downscaling methods

• CA– Daily variability

is conserved– Tends to

produce textured weaker trends

• BCSD– Daily variability

is not conserved

– Tends to produce smooth stronger trends

T R E N D S IN JF M T A V G ( C /P E R D E C A D E )