increases in puget sound estuarine flood risk under climate change

INCREASES IN PUGET SOUND ESTUARINE FLOOD RISK UNDER CLIMATE CHANGE
Joe Hamman , Alan F. Hamlet
1
1,2
(1) Department of Civil and Environmental Engineering, University of Washington, (2) Climate Impacts Group, University of Washington
Washington
Idaho
Nisqually River
Oregon
Figure. Map showing the two study locations.
PROJECT GOALS
1. Incorporate Regional Climate Model (RCM) projections into ongoing climate impacts studies.
2. Use RCM projections to develop temporally consistent storm responses (future floods and storm surges).
3. Compare the RCM results to those of previous studies.
4. Use projected river flows and storm surges to evaluate physical flood impacts using hydrodynamic models.
METHODS
PROJECT OUTLINE
HYDROLOGY MODEL: The Variable Infiltration Capacity (VIC) model was run at 1/16th degree
resolution. All input variables (Tmin, Tmax, Wind,
and Precip) were obtained from the bias corrected
RCM outputs or the statistically downscaled GCM
outputs. Routed VIC outputs were also bias corrected using a quantile mapping approach with USGS
daily flows.
RESERVOIR MODELS: The reservoir models
used in this study were constructed following the
methodology outlined in Lee et al. (2011). In short,
the models simulate reservoir operating policies at a
daily timestep while meeting prescribed hydropower demands, minimum flow requirements and flood
control targets.
This project relies heavily on work previously completed by others, especially
in regards to the complex climate and hydrologic modeling. The figure below
outlines the process of developing local scale flood projections beginning with
a global climate model.
GLOBAL CLIMATE MODEL
1.5 DEG - ECHAM5 - SRES A1B
Tide Anomaly (FT)
1
0.5
0
-0.5
-1
-0.04
HYDROLOGIC VARIABLES
PRECIP, WIND, TEMP
TIDE / STORM SURGE LEVELS
DAILY FLOWS
BIAS CORRECTION
RESERVOIR OPERATIONS MODEL
FLOOD & HYDROPOWER
DAILY FLOWS
0 H.D. -­‐ ECHAM5 -­‐ 2080s WRF – ECHAM5 ECHAM5-­‐
2020s WRF -­‐ Reanalysis 43 WRF -­‐ Reanalysis 97 WRF – ECHAM5 0 WRF – ECHAM5 WRF – ECHAM5 RESULTS
Observations 20 WRF – ECHAM5 WRF – ECHAM5 55 • Pairing peak monthly tidal anomaly (surge) with corresponding daily flow, the
joint probability of exceedence is calculated.
• The figure below shows the joint probability distribution for p = 0.10 and p =
0.01 (Note that these figures are peak monthly values, as opposed to peak annual as shown earlier).
Hybrid Delta 2040s using WRF reanalysis as time consistent storm surge forcings. Hybrid Delta 2080s using WRF reanalysis as time consistent storm surge forcings. WRF -­‐ ECHAM5 20th century climate run. STORM SURGE / SEA LEVEL RISE:
• Minimal change in actual storm surge.
• Good match between ECHAM5 1980s
and Historical 1980s time periods.
• Change in peak annual storm surge is
dominated by sea level rise.
WRF -­‐ ECHAM5 SRES A1B, 2041 – 2069. Probability of Exceedence
1
1
0.8
0.6
0.4
0.2
PROBABILITY OF EXCEEDANCE, p
0
5
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
x 10
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
3.5
1
3
2.5
2
-0.02
0
0.02
0.04
0.06
SEA LEVEL RISE & STORM SURGE:
• The principle harmonic constituents were fit to the
measured hourly water levels using a least-squares
approach.
• These constants were then used to provide a time
series of predicted hourly tides.
• The differences from the predicted and measured
tides were summarized at a daily timestep.
•-0.04A regression approach was used for each month to
estimate future water levels given a set of atmospheric variables from the GCM.
• An iterative process was used to determine which
variables best described tidal anomalies experienced in the Puget Sound. Local pressure, the
1st and 3rd EOF signals, and ENSO were the only
variables found to be significantly impactful on
the anomaly.
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 !"#$ = 𝑓𝑓(𝑃𝑃, 𝑃𝑃!!"# , 𝑃𝑃!"! , 𝑃𝑃!"! , 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸) HYDRODYNAMIC MODELS
2D FLOOD ROUTING
• Regression during stormy months yields R2 >0.8.
• Sea level rise was uniformly added to the Tideanom
for each time period based on projections outlined
by Mote et al. (2008).
0
1
1
0.8
0.6
0.4
0.2
PROBABILITY OF EXCEEDANCE, p
0
0
1
0.8
0.6
0.4
0.2
PROBABILITY OF EXCEEDANCE, p
2
x 10
Estimated Daily Flow at Mount Vernon,
Skagit River 2040-2069
1.5
1
0.5
0
2040
2
2045
2050
2055
2060
2065
2070
Estimated Daily Tidal Anomaly at Seattle,
2040-2069
1
0
-1
-2
2040
1
5
1
2
3
4
TIDAL ANOMALY - ABOVE PREDICTED LEVEL (FT), h
NISQUALLY RIVER
10000
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
8000
5000
4000
7000
5000
4000
3000
2000
2000
5
1% Joint Probability of Exceedence
6000
3000
1
2
3
4
TIDAL ANOMALY - ABOVE PREDICTED LEVEL (FT), h
1000
0
1
2
3
4
TIDAL ANOMALY - ABOVE PREDICTED LEVEL (FT), h
2050
2055
5
0
1. Use hydrodynamic models to evaluate impacts of increased flood risk.
2. Develop methods to estimate storm surge closer to estuary.
3. Expand methodology to other Puget Sound estuaries.
1. Hamlet, A.F., P. Carrasco, J. Deems, M.M. Elsner, T. Kamstra, C. Lee,
S-Y Lee, G. Mauger, E. P. Salathe, I. Tohver, L. Whitely Binder, 2010, Final
Project Report for the Columbia Basin Climate Change Scenarios Project,
http://www.hydro.washington.edu/2860/report/.
2. Lee, Se-Yeun, A.F. Hamlet, 2011: Skagit River Basin Climate Science
Report, a summary report prepared for Skagit County and the Envision Skagit
Project by the Department of Civil and Environmental Engineering and The
Climate Impacts Group at the University of Washington.
3. Mote, P.W., A. Peterson, H. Shipman, W.S. Reeder, and L. Whitely Binder,
2008: Sea level rise in the coastal waters of Washington. Report for the
Climate Impacts Group, University of Washington, Seattle.
4. Salathé, E.P., Leung, L.R., Qian, Y., and Zhang, Y., 2010. Regional climate
model projections for the State of Washington. Climatic Change, dot:10.1007/
s10584-010-9849-y.
Corresponding Author: Alan Hamlet: [email protected]
2045
5
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
9000
6000
0
0
REFERENCES
TEMPORALLY CONSISTENT PROJECTIONS:
5
3
FUTURE WORK
1.5
0.5
4
• Using regional climate model results, we demonstrate a process for developing and implementing temporally consistent climate change projections
into future flood risk studies.
• A monthly regression approach was successfully developed to estimate
storm surge.
• The modest increases in Skagit flood peaks produced by WRF requires further study.
• The development of joint probability functions facilitates informed scenario
selection for hydrodynamic modeling studies.
4
2
5
DISCUSSION
Probability of Exceedence
4
4.5
1.5
1
2
3
4
TIDAL ANOMALY - ABOVE PREDICTED LEVEL (FT), h
10% Joint Probability of Exceedence
7000
1000
NISQUALLY RIVER
Probability of Exceedence
2.5
0
8000
1.5
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
2
9000
2
SKAGIT RIVER
x 10
3
10000
2.5
10% Joint Probability of Exceedence
4
x 10
6
4
3
PEAK ANNUAL FLOW:
• Skagit River: WRF shows modest increases in flood magnitude while the
H.D. shows increases up to 20%.
• Nisqually River: WRF runs show very large increases in flood magnitudes
(up to 85%).
• Both rivers overall show increases in flood magnitude.
3
5
1
0.5
5
Historical
HD-2040s
HD-2080s
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
3.5
0
7
2
Historical
ECHAM5-1980s
ECHAM5-2020s
ECHAM5-2050s
4
SKAGIT RIVER
1% Joint Probability of Exceedence
6
WRF -­‐ ECHAM5 SRES A1B, 2010 – 2039. 4.5
x 10
7
SEATTLE STORM SURGE
5
4
STREAMFLOW (CFS), q
2040-­‐2069 ECHAM5-­‐
2050s Flow (cfs), q
PSFC - PC1
1.5
SEA SURFACE TEMPERATURES
NINO3.4 INDEX
REGIONAL CLIMATE MODEL
SEA LEVEL RISE / STORM SURGE
1/8 DEG - WRF
REGRESSION MODEL
SURFACE PRESSURE
HYDROLOGY MODEL
1/16 DEG - VIC
1970-­‐1999 ECHAM5-­‐
1980s 2010-­‐2039 Gridded Observations H.D. -­‐ ECHAM5 -­‐ 2040s HD -­‐ 2080s 1970-­‐1999 Description 0.5
2
LARGE SCALE
ATMOSPHERIC FORCINGS
1970-­‐1999 Surge (ft), h
Pacific Ocean
STATISTICAL DOWNSCALING: Statistical
downscaling using the hybrid-delta method was used
to downscale and the raw GCM data. This process
projects the monthly changes indicated by the GCM
onto a historical data set. A complete description
of the methods used in developing the statistically
downscaled climate data can be found in Hamlet et
al. (2010).
HD -­‐ 2040s Sea Level Rise (cm) STREAMFLOW (CFS), q
Skagit River
Historical Hydrologic Source Storm Surge Source STREAMFLOW (CFS), q
British Columbia
1970-­‐1999 JOINT PROBABILITY
STREAMFLOW (CFS), q
REGIONAL CLIMATE MODEL: The Weather
Research Forecast (WRF) model was used to dynamically downscale GCM projections. Because
WRF is run at higher spatial (1/8th degree) and temporal (6 minutes) resolutions that the large scale
GCM, it provides much more realistic weather
events than are otherwise possible. Additionally,
dynamic downscaling allows weather patterns to
change according to first principle relationships.
The RCM outputs were also bias corrected to match
historical statistics. Citation: Salathe et al. (2010).
Time Period Climate Scenario STREAMFLOW (CFS), q
GLOBAL CLIMATE MODEL: The ECHAM5
climate model, SRES A1B, was used by Salathe et
al. (2010) as the large scale forcing for the regional
climate model. Monthly gridded equatorial sea surface temperatures (SSTs) were extracted from this
large scale data set to include ENSO effects.
MODEL RUNS
PEAK ANNUAL STORM SURGE, H
Near coastal environments have been identified as some of the most likely to be
impacted by climate change. Observed changes in Puget Sound sea level and
flood magnitudes are in line with those projected by previous climate change
impacts studies. Current understanding of the combined effects of these changes is relatively low and has promoted us to explore the ways in which these two
influence near coastal ecosystems and infrastructure. Specifically, this project
examines the effects of climate change on estuarine water surface elevations by
exploring the combined effects of sea level rise and riverine flooding. The project utilizes a chain of numerical models to quantify the climate change altered
hydraulic conditions expected in two estuaries in the Puget Sound (the Skagit
and Nisqually Rivers).
DATA
STREAMFLOW (CFS), q
INTRODUCTION
2060
2065
2070
Prepared for the 2012 University of Washington Water Symposium.