Session A8- Assessing potential climate change impacts to streamflow in the northeast U.S. using regional flow duration models

Transcription

1 University of Massachusetts - Amherst ScholarWorks@UMass Amherst International Conference on Engineering and Ecohydrology for Fish Passage International Conference on Engineering and Ecohydrology for Fish Passage 2011 Jun 29th, 1:35 PM - 1:55 PM Session A8- Assessing potential climate change impacts to streamflow in the northeast U.S. using regional flow duration models Neil Fennessey University of Massachusetts - Dartmouth Follow this and additional works at: Fennessey, Neil, "Session A8- Assessing potential climate change impacts to streamflow in the northeast U.S. using regional flow duration models" (2011). International Conference on Engineering and Ecohydrology for Fish Passage This Event is brought to you for free and open access by the The Fish Passage Community at UMass Amherst at ScholarWorks@UMass Amherst. It has been accepted for inclusion in International Conference on Engineering and Ecohydrology for Fish Passage by an authorized administrator of ScholarWorks@UMass Amherst. For more information, please contact scholarworks@library.umass.edu.

2 Potential Climate Change Impacts to Streamflow in the Northeast U.S. Neil M. Fennessey University of Massachusetts-Dartmouth Fish Passage Engineering and Ecohydrology University of Massachusetts June 29, 2011

3 Study Motivation A growing consensus that climate change is real. Climate change predictions are made using General Circulation Models or GCMs. Applying GCM output results to explore regional issues is called downscaling. downscaling. How do simple gauge record based regional statistical streamflow models compare with sophisticated rainfall-runoff runoff models typically used to explore downscaling?

4 Old vs. New Early downscaling studies focused on potential impacts due to 2xCO2. GCMs were spun up to equilibrium, atm.. CO2 was instantly doubled, simulation continued to a new equilibrium. Grid resolution was O[10k km 2 ] 8x10 degrees. 300 meter slab ocean, vertical fluxes only. Current GCM resolution is 2x4 deg. Coupled ocean-atmospheric atmospheric model. Simulations driven by time varying CO2 concentrations.

5

6

7 USGS HCDN Streamgauge Network HCDN stream gauge network constructed by Slack and Landwher (1988). NMF stream gauge network consists of 166 daily HCDN gauges with POR>25 years. Watersheds range from 1 mi 2 to 5,000 mi 2 NMF network gauges are in Maine, New Hampshire, Vermont, Massachusetts, Connecticut, Rhode Island, New York, New Jersey and Pennsylvania.

8 166 HCDN Gage Network

9 Streamflow Duration Curves FDCs are developed from daily gauge data. Discrete quantiles are estimated for each of the 166 daily POR stream gauges. Quantile estimators are: - Q5: very high flow, exceeded 5% of the POR - Q10: high flow, exceeded 10% of the POR - E[Q] mean daily flow (approx Q27) - Q50: median daily flow, exceeded 50% of POR - Q90: low flow, exceeded 90% of the POR. - Q95: very low flow, exceeded 95% of the POR

10 Q50 Q5 ** * Q10 E[Q] * ** Q90 Q95

11 Climate Data Daily data from 324 NOAA Summary of the Day Climate Observatories. SOD records sampled for: - Mean annual air temperature: T A - Mean annual precipitation: P A

12 324 SOD Climate Stations

13 Regional Model Development Statistical models developed using stepwise OLS multivariate regression. Q5, Q10, E[Q] models depend on both P A and T A. Q50, Q90 and Q95 are only dependent on P A. All model R 2 > Other significant variables include: - main stream channel slope - basin relief - SCS soil moisture retention S/runoff CN

14 Daily Flow and Mean Annual Precipitation 10 Q 5 10 Q p /A (cfs/mi 2 ) Q 10 E[q] Q 50 Q 90 Q Mean Annual Precipitation (inch) 0

15 Daily Flow and Mean Annual Air Temp Mean Annual Temperature ( o F) 16 Q p /A (cfs/mi 2 ) Q 5 Q 10 E[q] Q 50 Q 90 Q

16 Elasticity Elasticity The elasticity is used to quantify how sensitive streamflows are to P A and T A. Easy way to quickly assess potential impacts of climate change on regional streamflow. ( ) Q,T p A = T Q A P Q T p A Elasticity ( ) Q, P p A = P Q A P Q P p A

17 Elasticity(Q P,T A ) Q5: Q10: E[Q]: Q50: 0.0 Q90: 0.0 Q95: 0.0 Elasticity(Q P,P A ) Q5: 0.83 Q10: 1.12 E[Q]: 1.20 Q50: 1.85 Q90: 2.66 Q95: 2.66

18 Example: Sensitivity of E[Q] to T A and P A Elasticity (E [Q], T a ) = A 10% increase in T A will result in a x10% = - 7.5% reduction in the mean daily streamflow, E[Q] Elasticity (E [Q], P A ) = 1.20 A 10% increase in P A will result in a 1.20x10% = 12.0% increase in the mean daily streamflow, E[Q]

19 Downscaling: Applying GCM Simulation Results to Local Issues Kirshen and Fennessey (1995) examined the MWRA water supply system response to 2xCO2 using the Sacramento Soil Moisture Rainfall Runoff Model. Sacramento Soil Moisture Model is very parameter intensive (i.e. expensive). The Ware River is seasonally diverted to the Quabbin Reservoir when flows are above a threshold.

20

21 % Change in Ware River Mean Daily Flow (E[Q]) to changes in T A and P A (Bracket values Kirshen and Fennessey, 1995) P A +3.7 o F (5 o C) +7.2 o F (10 o C) -20% -23% (-40%)( -31% (-46%)( 0% -6% (-9%)( -10% (-16%)( 20% +17% (+26%) +11% (+17%)

22 Summary & Conclusions Streamflow elasticity WRT T A and P A estimated for a wide range of flows observed in northeast US. are Good first order estimates of potential impacts on stream flows due to alternative climate forcing without the need to use sophisticated rainfall-runoff runoff models. Very useful for resolving bar room disputes. PCD Milly (this conference) Climate change model results are uncertain and downscaling is even more uncertain.

23 What will the emissions be?