Alan F. Hamlet Ingrid Tohver Se-Yeun Lee JISAO/CSES Climate Impacts Group Dept. of Civil and Environmental Engineering University of Washington Quantifying the Effects of Climate Variability and Change on Hydrologic Extremes in the Pacific Northwest
CBCCSP Research Team Lara Whitely Binder Pablo Carrasco Jeff Deems Marketa McGuire Elsner Alan F. Hamlet Carrie Lee Se-Yeun Lee Dennis P. Lettenmaier Jeremy Littell Guillaume Mauger Nate Mantua Ed Miles Kristian Mickelson Philip W. Mote Rob Norheim Erin Rogers Eric Salathé Amy Snover Ingrid Tohver Andy Wood chap1_intro_final.pdf
The Myth of Stationarity: 1) Climate Risks are stationary in time. 2) Observed streamflow records are the best estimate of future variability. 3) Systems and operational paradigms that are robust to past variability are robust to future variability.
Image Credit: National Snow and Ice Data Center, W. O. Field, B. F. Molnia Aug, 13, 1941Aug, 31, 2004 The Myth of Stationarity Meets the Death of Stationarity Muir Glacier in Alaska
Why a Focus on Hydrologic Extremes? Many human and natural systems are quite robust under “normal” conditions, but have the potential to be profoundly impacted by hydrologic extreme events.
Floods
Drought Evacuated Reservoir During the 2001 PNW Drought
Wildfire
Low Flow and Temperature Impacts to Fish Temperature/ Disease Related Fish Kill in the Klamath River in 2002
Dissolved Gas Management Tailrace below Bonneville Dam
Dam Safety Aftermath of the Johnstown Flood 1889
Dilution Flows for Industrial Pollutants
Stormwater Management
Sediment Transport and Mudslides
Nuts and Bolts: Traditional Methods for Estimating Hydrologic Extremes
Step 1: Select Extreme Event from Each Historical Year Streamflow (cfs) Day of the Water Year (1 = Oct 1)
Step 2: Rank Extreme Events for All Years and Estimate Quantiles Streamflow (cfs) Probability of Exceedance 1999
Step 3: Fit a Probability Distribution to the Data Examples of Commonly Used Probability Distributions: Extreme Value Type 1 (EV 1) Log Normal (LN) Log Pearson Generalized Extreme Value (GEV) For climate change experiments, GEV is a good choice since the true nature of the future probability distributions is essentially unknown. However it turns out that the choice of distribution is not very critical in terms of the evaluating the sensitivity to warming and/or precipitation change.
Step 4: Estimate Extremes Associated with Return Intervals Site NameRet. Int.Flow (cfs) SNOMO : SNOMO : SNOMO : Note that any return interval can be estimated. E.g. one could provide an estimate of the “5000 year flood”.
Step 5 (Optional) : “Regionalize” the Results In order to avoid the inherent “noise” that comes with using imperfect site specific data, a common approach is to “regionalize” the results. The idea is to pool as many sites as possible that have common hydroclimatic features (e.g. sites in western WA), and express the flood statistics as a simple ratio to the mean annual flood (MAF) averaged over many different basins. E.g. Q 100 = 2.7 * MAF This approach is used by Ecology in providing estimates of extreme events for the Dam Safety Program, for example.
Low flow analysis is essentially the same except we select the extreme low flow event from each year. 7Q10, for example, extracts the lowest 7- day running mean flow from each historical year, fits a probability distribution to the sequence of extremes, and selects the 90% exceedance value (i.e. a 10% probability of being at or below this extreme value)
Historical Perspectives: Changing Flood Risk in the 20 th Century
References: Niemann, PJ, LJ Schick, FM Ralph, M Hughes, GA Wick, 2010: Flooding in Western Washington: The Connection to Atmospheric Rivers, J. of Hydrometeorology, (in review) Hamlet AF, Lettenmaier DP (2007) Effects of 20th century warming and climatevariability on flood risk in the western U.S. Water Resour Res, 43:W06427.doi: /2006WR005099
Observed Characteristics of Extreme Precipitation Events
Evidence of Changing Flood Statistics
Role of Atmospheric Rivers in Flooding (Nov 7, 2006) Niemann, PJ, LJ Schick, FM Ralph, M Hughes, GA Wick, 2010: Flooding in Western Washington: The Connection to Atmospheric Rivers, J. of Hydrometeorology, (in review)
Role of Atmospheric Rivers in Flooding (Oct 20, 2003)
Niemann, PJ, LJ Schick, FM Ralph, M Hughes, GA Wick, 2010: Flooding in Western Washington: The Connection to Atmospheric Rivers, J. of Hydrometeorology, (in review)
Modeling Studies of Changing 20 th Century Flood Risk in the West
Snow Model Schematic of VIC Hydrologic Model Sophisticated, fully distributed, physically based hydrologic model Widely used globally in climate change applications 1/16 Degree Resolution (~5km x 6km or ~ 3mi x 4mi) General Model Schematic
Avg WY Date of Flooding VIC Avg WY Date of Flooding OBS Ln (X 100 / X mean ) OBS Ln (X 100 / X mean ) VIC Evaluating the Hydrologic Model Simulations in the Context of Reproducing Flood Characteristics Red = PNW, Blue = CA, Green = Colo, Black = GB
Zp X 100 GEV flood/mean flood Red = VIC Blue = OBS 5-yr 20-yr 10-yr 50-yr 100-yr
Tmin Tmax PNW CA CRB GB Regionally Averaged Temperature Trends Over the Western U.S
Temperature Historic temperature trend in each calendar month Detrended Temperature Driving Data for Flood Risk Experiments “Pivot 2003” Data Set “Pivot 1915” Data Set
X / X DJF Avg Temp (C) Simulated Changes in the 20-year Flood Associated with 20 th Century Warming X / X
Freezing Level Snow Schematic of a Cool Climate Flood Precipitation Produces Snow Precipitation Produces Snow Precipitation Produces Runoff Snow Melt
Freezing Level Snow Schematic of a Warm Climate Flood Precipitation Produces Snow Precipitation Produces Snow Precipitation Produces Runoff Snow Melt
Regionally Averaged Cool Season Precipitation Anomalies PRECIP
DJF Avg Temp (C) 20-year Flood for “ ” Compared to “ ” for a Constant Late 20 th Century Temperature Regime X 20 ’73-’03 / X 20 ’16-’03
Summary of Flooding Impacts Rain Dominant Basins: Increases in flooding due to increased precipitation intensity, but no significant change from warming alone. Mixed Rain and Snow Basins Along the Coast: Strong increases due to warming and increased precipitation intensity (both effects increase flood risk) Inland Snowmelt Dominant Basins: Relatively small overall changes because effects of warming (decreased risks) and increased precipitation intensity (increased risks) are typically in the opposite directions.
Effects of ENSO and PDO on Flood Risk
DJF Avg Temp (C) X 100 nENSO / X X 100 cENSO / X X 100 wENSO / X X 100 nENSO / X X 100 cENSO / X X 100 wENSO / X
DJF Avg Temp (C) X 100 nPDO / X X 100 cPDO / X X 100 wPDO / X X 100 nPDO / X X 100 cPDO / X X 100 wPDO / X
Scenarios of Flood Risk in the 21 th Century
Mote, P.W. and E. P. Salathe Jr., 2010: Future climate in the Pacific Northwest, Climatic Change, DOI: /s z 21 st Century Climate Impacts for the Pacific Northwest Region
Seasonal Precipitation Changes for the Pacific Northwest Mote, P.W. and E. P. Salathe Jr., 2010: Future climate in the Pacific Northwest, Climatic Change, DOI: /s z
Smaller basins down to ~500 km 2 Monthly and daily streamflow time series Assessment of hydrologic extremes (e.g. Q100 and 7Q10) Columbia Basin Climate Change Scenarios Project 297 Sites
Available PNW Scenarios 2020s – mean ; 2040s – mean ; 2080s – mean Downscaling Approach A1B Emissions Scenario B1 Emissions Scenario Hybrid Delta hadcm cnrm_cm ccsm3 echam5 echo_g cgcm3.1_t4 7 pcm1 miroc_3.2 ipsl_cm4 hadgem1 2020s s s109 Transient BCSD hadcm cnrm_cm ccsm3 echam5 echo_g cgcm3.1_t4 7 pcm Delta Method composite of s s s11
Hybrid Downscaling Method Performed for each VIC grid cell: Hist. Daily Timeseries Hist. Monthly Timeseries Historic Monthly CDF Bias Corrected Future Monthly CDF Projected Daily Timeseries yr window “Base Case”
Spatial Variability of Temperature and Precipitation Changes
Daily Precipitation (mm) Day of Month Monthly to Daily Precipitation Scaling SeaTac. Feb, 1996, hypothetical 30% Increase
Snow Model Schematic of VIC Hydrologic Model Sophisticated, fully distributed, physically based hydrologic model Widely used globally in climate change applications 1/16 Degree Resolution (~5km x 6km or ~ 3mi x 4mi) General Model Schematic
Watershed Classifications: Transformation From Snow to Rain Map: Rob Norheim
Flood Analysis: What’s In? What’s Out? Issue Affecting AnalysisYesNo Based on explicit daily time step simulations of streamflow? Yes Changing freezing elevation? Yes Rain on snow captured?Yes Increases/decreases in storm intensity? Yes (monthly statistics only) Changes in tails of probability distributions affecting extreme daily precipitation ? No Changes in size and sequencing of storms? No Changes in small scale thunder storms? No Includes water management effects? No
Low Flow Analysis: What’s In? What’s Out? Issue Affecting AnalysisYesNo Based on explicit daily time step simulations of streamflow? Yes Effects of changing snowmelt and soil moisture dynamics? Yes Effects of changing evaporation? Yes, but some potential factors omitted (e.g. changes in cloudiness) Changes in sequencing or duration of drought? No Includes shallow ground water? No, but typically captures relevant affects to low flows anyway (well correlated) Includes deep groundwater?No Includes effects of glaciers?No Includes water management effects? No
Simulate Daily Time Step Streamflow Scenarios Associated with Changes in Climate Fit Probability Distributions To Estimate Flood and Low Flow Risks Compare Flood Risks to Those in the 20 th Century
SNOMO Streamflow (cfs) Probability of Exceedance
2040s Changes in Flood Risk Snohomish at Monroe A1BB1 Historical 10 Member Ensemble Using the Hybrid Delta Downscaling Approach
A1BB1 2040s Changes in 7Q10 Snohomish at Monroe Historical 10 Member Ensemble Using the Hybrid Delta Downscaling Approach
Chehalis at Grand Mound
Relationship Between Change in Q100 and Winter Temp
Changes in High Flows Q 100 values are projected to systematically increase in many areas of the PNW due to increasing precipitation and rising snowlines. _chap7_extremes_final.pdf
7Q10 values are projected to systematically decline in many areas due to loss of snowpack and projected dryer summers Changes in Low Flows _chap7_extremes_final.pdf
Current and Future Research Additional VIC calibration to improve simulations, and comparison with DHSVM models (proposed) Estimate the effects of reservoir management (in progress) Incorporate more realistic effects to extreme precipitation from regional scale climate models (in progress) Incorporate the effects of sea level rise and high flows on inundation using hydrodynamic modeling (proposed)
Regional Climate Modeling at CIG WRF Model (NOAH LSM) 36 to 12 km ECHAM5 forcing CCSM3 forcing (A1B and A2 scenarios) HadRM 25 km HadCM3 forcing
Extreme Precipitation Change from to in the percentage of total precipitation occurring when daily precipitation exceeds the 20 th century 95 th percentile Salathé, E.P., L.R. Leung, Y. Qian, and Y. Zhang Regional climate model projections for the State of Washington. Climatic Change 102(1-2): 51-75, doi: /s y
Snohomish River Near Monroe, WA
Some Implications for Policy Response to Changing Flood Risk
Scenarios not forecasts! The current projections are an initial attempt to provide quantitative estimates of the magnitude and direction of changing hydrologic extremes across the PNW, but there are many missing pieces: More fully integrated modeling studies and summary products are needed to better support many policy and design decisions. Reducing the cost and increasing the frequency of updates will help keep key products and data sets current.
We need to move forward now with the best available information. We almost certainly will not have all of the data and projections that we would like to have before we have to make difficult decisions that materially affect future outcomes. Identifying “No Regrets” strategies may be the best approach for coping with these realities.
Improving Estimates of the 100 ‐ year Flood: Methodology and Applications to the Olympic National Forest USFS Team: Kathy O’Halloran Bill Shelmerdine Luis Santoyo Robin Stoddard Robert P Metzger UW Team: Alan F. Hamlet Ingrid Tohver Se-Yeun Lee Rob Norheim
Mote, P.W. and E. P. Salathe Jr., 2010: Future climate in the Pacific Northwest, Climatic Change, DOI: /s z 21 st Century Climate Impacts for the Pacific Northwest Region
Snow Model Schematic of VIC Hydrologic Model Sophisticated, fully distributed, physically based hydrologic model Widely used globally in climate change applications 1/16 Degree Resolution (~5km x 6km or ~ 3mi x 4mi) General Model Schematic
Intercomparison of USGS and VIC Q100 Estimates
Intercomparison of Change in Q100 from USGS and VIC Models
Hybrid Product Based on USGS Baseline with VIC Change Map
Validation at HCDN Streamflow Sites
VIC
Validation at HCDN Streamflow Sites
Extensions and Next Steps Develop a decision support tool for assessing changing risk at any point or spatial scale (similar to the basic functionality of Streamstats in delineating the basin, etc.) Collaborate with design professionals in the Olympic National Forest to further develop and refine the tool Extend to other PNW National Forests and Parks
1/16 th Degree Changes in Natural Flood Risk