1. Climate and Energy in California David W. Pierce Tim P. Barnett Eric Alfaro Alexander Gershunov Climate Research Division Scripps Institution of Oceanography La Jolla, CA

2. How we got started: a typical climate change result IPCC, 2001 What does this mean to us?

3. Effect of Climate Change on Western U.S. Large and growing population in a semi-arid region How will it impact water resources? Use an “end-to-end” approach

4. Project overview Tim Barnett, SIO; R. Malone, LANL; W. Pennell, PNNL; A. Semtner, NPS; D. Stammer, SIO; W. Washington, NCAR

5. Step 1 Begin with current state of global oceans

6. Why initialize the oceans? That’s where the heat has gone Data from Levitus et al, Science, 2001

8. Step 2 Estimate climate change due to emissions

9. Global Climate Change Simulation Parallel Climate Model (PCM) Business as Usual Scenario (BAU) 1995-2100 5 ensemble members

10. How well does the PCM work over the Western United States? Dec-Jan-Feb total precipitation (cm)

11. Step 3 Downscaling and impacts

12. Why downscale? Global model (orange dots) vs. Regional model grid (green dots)

13. How good is downscaling? El Nino rainfall simulation ObservationsDownscaled modelStandard reanalysis Ruby Leung, PNNL

14. Change in California snowpack

15. Projected change by 2050

16. River flow earlier in the year

17. Runoff already coming earlier

18. Columbia Basin Options Hydropower Or Salmon

19. Los Angeles water shortage Christensen et al., Climatic Change, to appear

20. Miss water treaty obligations to Mexico Christensen et al., Climatic Change, to appear

21. More wildfires 100% more acres burned in 2100

22. Less time for Salmon to reproduce Now: Lance Vail, PNNL Future:

24. A reduction of winter snowpack. Precipitation more likely to fall as rain, and what snow there is melts earlier in the year. River flow then comes more in winter/spring than in spring/summer – implications for wildfires, agriculture, recreation, and how reservoirs are managed. Will affect fish whose life cycle depends on the timing of water temperature and spring melt. Will also change salinities in the San Francisco bay. Climate change conclusions

25. More heat waves Dan Cayan and Mike Dettinger, Scripps Inst. Oceanography

26. August daily high temperature, Sacramento, CA On a warm summer afternoon, 40% of all electricity in California goes to air conditioning

27. California Energy Project Objective: Determine the economic value of climate and weather forecasts to the energy sector

28. Climate & weather affect energy demand Source: www.caiso.com/docs/0900ea6080/22/c9/09003a608022c993.pdf

29. …and also energy supply Green et al., COAPS Report 97-1 Typical effects of El Nino: CA hydro

30. Project Overview Scripps Inst. Oceanography University of Washington Georgia Inst. Tech California Energy Commission California ISO PacifiCorp San Diego Gas & Elec. SAIC Academia State Partners Industrial Partners

31. Why aren’t climate forecasts used? Climate forecasts are probabilistic in nature – sometimes unfamiliar to the user

32. What climate forecasts mean

33. Why aren’t climate forecasts used? Climate forecasts are probabilistic in nature – sometimes unfamiliar to the user Lack of understanding of climate forecasts and their benefits Language and format of climate forecasts is hard to understand – need to be translated for end-users Aversion to change – easier to do things the traditional way

34. 1. California "Delta Breeze" An important source of forecast load error (CalISO) Big events can change load by 500 MW (>1% of total) Direct cost of this power: $250K/breeze day (~40 days/year: ~$10M/year) Indirect costs: pushing stressed system past capacity when forecast is missed!

35. NO delta Breeze Sep 25, 2002: No delta breeze; winds carrying hot air down California Central valley. Power consumption high.

36. Delta Breeze Sep 26, 2002: Delta breeze starts up; power consumption drops >500 MW compared to the day before!

37. Weather forecasts of Delta Breeze 1-day ahead prediction of delta breeze wind speed from ensemble average of NCEP MRF, vs observed.

38. Statistical forecast of Delta Breeze (Also uses large- scale weather information) By 7am, can make a determination with >95% certainty, 50% of the time

39. Delta Breeze summary Using climate information can do better than dynamic weather forecasts Possible savings of 10 to 20% in costs due to weather forecast error. Depending on size of utility, will be in range of high 100,000s to low millions of dollars/year.

40. 2. Load demand management Induce customers to reduce electrical load on peak electrical load days Prediction challenge: call those 12 days, 3 days in advance Amounts to calling weekdays with greatest "heat index" (temperature/humidity)

41. Why shave peak days? http://www.energy.ca.gov/electricity/wepr/2000-07/index.html

42. Price vs. Demand http://www.energy.ca.gov/electricity/wepr/1999-08/index.html

43. July SundayMondayTuesdayWednesdayThursdayFridaySaturday 1 2990 MW 79 F 2 3031 MW 81 F 3 3389 MW 88 F 4 2958 MW 85 F 5 67 2814 MW 71 F 8 2766 MW 73 F 9 2791 MW 75 F 10 2906 MW 79 F 11 3106 MW 83 F 12 1314 3130 MW 76 F 15 3089 MW 74 F 16 3046 MW 84 F 17 3102 MW 77 F 18 2888 MW 78 F 19 2021 3317 MW 82 F 22 2867 MW 73 F 23 3055 MW 77 F 24 2991 MW 73 F 25 3006 MW 75 F 26 2728 2935 MW 78 F 29 3165 MW 82 F 30 3398 MW 86 F 31 3176 MW 78 F Average = 2916 MW

44. July SundayMondayTuesdayWednesdayThursdayFridaySaturday 1 2990 MW 79 F 2 3031 MW 81 F 3 3389 MW 88 F 4 2958 MW 85 F 5 67 2814 MW 71 F 8 2766 MW 73 F 9 2791 MW 75 F 10 2906 MW 79 F 11 3106 MW 83 F 12 1314 3130 MW 76 F 15 3089 MW 74 F 16 3046 MW 84 F 17 3102 MW 77 F 18 2888 MW 78 F 19 2021 3317 MW 82 F 22 2867 MW 73 F 23 3055 MW 77 F 24 2991 MW 73 F 25 3006 MW 75 F 26 2728 2935 MW 78 F 29 3165 MW 82 F 30 3398 MW 86 F 31 3176 MW 78 F Average = 2916 MWTop days = 3383 MW (16 % more than avg)

45. Peak day electrical load savings If knew electrical loads in advance: 16% With event constraints: 14% (Load is relative to an average summer afternoon)

46. July SundayMondayTuesdayWednesdayThursdayFridaySaturday 1 2990 MW 79 F 2 3031 MW 81 F 3 3389 MW 88 F 4 2958 MW 85 F 5 67 2814 MW 71 F 8 2766 MW 73 F 9 2791 MW 75 F 10 2906 MW 79 F 11 3106 MW 83 F 12 1314 3130 MW 76 F 15 3089 MW 74 F 16 3046 MW 84 F 17 3102 MW 77 F 18 2888 MW 78 F 19 2021 3317 MW 82 F 22 2867 MW 73 F 23 3055 MW 77 F 24 2991 MW 73 F 25 3006 MW 75 F 26 2728 2935 MW 78 F 29 3165 MW 82 F 30 3398 MW 86 F 31 3176 MW 78 F Average = 2916 MW

47. July SundayMondayTuesdayWednesdayThursdayFridaySaturday 1 2990 MW 79 F 2 3031 MW 81 F 3 3389 MW 88 F 4 2958 MW 85 F 5 67 2814 MW 71 F 8 2766 MW 73 F 9 2791 MW 75 F 10 2906 MW 79 F 11 3106 MW 83 F 12 1314 3130 MW 76 F 15 3089 MW 74 F 16 3046 MW 84 F 17 3102 MW 77 F 18 2888 MW 78 F 19 2021 3317 MW 82 F 22 2867 MW 73 F 23 3055 MW 77 F 24 2991 MW 73 F 25 3006 MW 75 F 26 2728 2935 MW 78 F 29 3165 MW 82 F 30 3398 MW 86 F 31 3176 MW 78 F Average = 2916 MWWarm days = 3237 MW (11 % more than avg)

48. Peak day electrical load savings If knew electrical loads in advance: 16% With event constraints: 14% If knew temperature in advance: 11% (Load is relative to an average summer afternoon)

49. What can climate analysis say?

50. Peak day electrical load savings If knew electrical loads in advance: 16% With event constraints: 14% If knew temperature in advance: 11% Super simple scheme (24C, 0.5): 6% (Load is relative to an average summer afternoon)

51. Optimizing the process

52. Peak day summary Might ultimately be a real-time program –Driven by "smart" electric meters –Main benefit would be avoided cost of peaker generation plants ~$12M/yr. Until then, climate prediction: –Far less deployment cost –Cost of avoided procurement ~$1.3M/yr -> Climate analysis can give expected benefits to a program

53. 3. Irrigation pump loads Electricity use in Pacific Northwest strongly driven by irrigation pumps When will the pumps start? What will total seasonal use be?

54. Irrigation pump electrical use

55. Pump start date Eric Alfaro, SIO

56. Total use over summer Idaho Falls, ID

57. Total load affected by soil moisture Eric Alfaro, SIO

58. Irrigation load summary Buying power contracts 2 months ahead of a high-load summer saves $25/MWh (over spot market price) Use: about 100,000 MWh Benefit of 2 month lead time summer load forecast: $2.5 M

59. 4. NPO and winter heating

60. Why the NPO matters Higher than usual pressure associated with the NPO… generates anomalous winds from the north west… …which bring more cold, arctic air into the western U.S. during winter

62. NPO affects summer, too!

63. Summer temperature, NPO above normal in spring Possible benefits: better planning, long term contracts vs. spot market prices

64. 5. Hydropower CalEnergy work done by U.W. hydrology group (Dennis Lettenmaier, Alan Hamlet, Nathalie Voisin)

65. Develop historic precipitation fields…

66. … then apply precipitation to a runoff model

67. Major components of CA model Lake Shasta Lake Trinity Whiskeytown Reservoir Lake Oroville Folsom Lake Pardee/Camanche Resv. New Hogan Reservoir New Melones Reservoir New Don Pedro Res./Lake McClure Millerton/Eastman/Hensley Sacramento-San Joaquin Delta San Luis Reservoir Flood control, navigation, fish conservation Water supply, hydropower, fish conservation Flood control, hydropower Flood control, water supply, hydropower, water quality, environmental conservation Flood control, water supply, hydropower Flood control, water supply Flood control, water supply, water quality, hydropower Flood control, water supply Water supply, recreation Water supply, water quality Water supply, hydropower USBR DWR USBR EBMUD COE USBR TMID, MC USBR, COE USBR, DWR USBR: Bureau of Reclamation DWR: CA Dept Water Resources EBMUD: East Bay Municipal District MC: Merced County TID: Turlock Irrigation District COE: US Army Corp of Engineers Van Rheenen et al., Climatic Change, 2004

68. Finally, make hydropower N. Voisin et al., Univ. Wash., 2004

69. Economic value of climate forecasts to the energy sector 1.Improved bay area and delta breeze forecasts: $100K’s to low $millions/yr 2.Peak day load management: ~$1-10M/yr 3.Pump loads: ~$2M/yr 4.Pacific SSTs: benefits of the information might include risk reduction, improved reliability, and improved planning 5.Hydropower: better water management, reduced costs

71. El Nino/La Nina

72. Why does that affect other places? Global atmospheric pressure pattern “steers” weather Horel and Wallace, 1981

74. Climate change Some of it is straightforward

75. Other parts are harder Clouds have competing effects

76. How good is the Hydrological Model? Andrew Wood, Univ. of Washington

77. Predicted change by 2050

79. Columbia River flow Andrew Wood, Univ. of Washington

80. The problem: Proposal to breach 4 Snake River dams to improve salmon habitat Those dams provide 940 MW of hydropower generation

82. Historical Global Temperatures

83. MSU (microwave sounding unit)

84. A difficult data set…

85. Problem: Orbit decay

86. MSU versus Jones

87. Paleo temperature history Mann et al, 2001

88. Effect of Economic Assumptions IPCC, 2001

89. Natural vs. Human Influences IPCC, 2001

90. Predicting summer temperature based on spring temperature

91. Dennis Gaushell, Cal-ISO

92. Cost of forecast errors

93. NPO and heating degree days Positive NPO Negative NPO Difference is about 150 HDD, or 5% of total HDD

94. Extreme events Same temperature threshold (e.g. 95 °F) => Same percentile threshold (e.g. 95th) =>

95. Spring SST predicting summer temperatures CDD Tmax-95 th percentile

96. Relationship PDO => California Summertime Temperatures Correlations, Mode 1- Tmean, JJA => Correlations, Mode 1-PSST, MAM

97. Contingency Analysis (conditional probabilities):  = 0.01 => ***, 0.05 => **, 0.10 => *

98. Step 3. Verify streamflow Nathalie Voisin et al., Univ. Washington, 2004

99. Step 4. Apply to reservoir model ColSim (Columbia Simulation) for the Pacific Northwest CVmod (Central Valley model) for Sacramento-San Joaquin basin Use realistic operating rules: –Energy content curves (ECC) for allocating hydropower –US Army Corp of Engineers rule curves for flood prevention –Flow for fish habitat under Biological Opinion Operating Plan –Agricultural withdrawal estimated from observations –Recreational use of Grand Coulee Dam reservoir

100. Step 2: Apply to soil/streamflow model Nathalie Voisin et al., Univ. Washington, 2004

101. Strong year to year variability