1. LUSciD-LLNL UCSD/SIO Scientific Data Project:
Climate Studies
SIO LLNL SDSC
Tim Barnett Doug Rotman Reagan Moore
David Pierce Dave Bader Leesa Brieger
Dan Cayan Ben Santer Amit Chourasia
Hugo Hidalgo Peter Gleckler
Mary Tyree Krishna AcutaRao

2. Objective
Can we detect a global warming signal in main hydrological features of the western United States?

3. Program Elements
Control run: Natural variability CCSM3 from NCAR on Thunder. (approx 4.5 TB)
Downscaling: 12 km grid over west for spatial resolution (control+anthro; another 5 TB)
Hydrological modeling: The downscaled data on rainfall, temperature, terrain, etc. force a hydrological model for time histories of steam flows and snow pack evolution in the western US (control+anthro: another 5 TB).
Detection and attribution (D&A) analysis.

4. Deliverables
Year 1
Complete a long GCM control run and begin statistical downscaling for selected geographic regions.…..DONE
Begin VIC simulations with both downscaled data and PCM forced realizations.……………………….....…..DONE
Implement a data grid linking resources between LLNL and SDSC. The data grid will be used to manage the simulation output that is generated.………………….…..DONE (1st order)

5. Deliverables (con’t) Year 2 Complete downscaling of Control.
Complete VIC run on downscaled Control run.
Prepare paper on downscaling intercomparisons
Begin preliminary D&A analysis.
Develop a digital library for publishing results, and integrating with PCMDI
Year 3
Complete D&A analysis.
Write a paper describing the results.

6. Change in California Snowfall

7. Change in Snow Water Equivalent
Observed,
Courtesy P. Mote

8. River flow earlier in the year
Insert Dettinger Merced river response slide

9. Runoff already coming earlier

10. Projected change by 2050

11. Key Question
Do the signals we see happen naturally or are they human-induced?
To answer, we need to know the levels and scales of natural variability in the western hydrological cycle.

12. ACCOMPLISHMENTS: Year 1

13. Long GCM control run CCSM3 with finite volume dynamical core (“-FV”)
Atmospheric resolution is 1.25ox1o with 26 vertical levels
Ocean resolution is 320x384 stretched grid with 40 levels (so-called “gx1v3” grid; averages 1 1/8ox0.5o)
760 years of a long pre-industrial control run transferred to SDSC

14. CCSM3-FV: Temperature, DJF

15. CCSM3-FV: Temperature, JJA

16. CCSM3-FV: Precipitation, DJF

17. CCSM3-FV: Precipitation, JJA

18. CCSM3-FV: Precip Variablity, DJF

19. CCSM3-FV: Precip Variablity, JJA

20. CCSM3-FV and El Nino

21. CCSM3-FV and La Nina

22. Next steps with CCSM3-FV
Dynamical downscaling
Provisional plan is to use COAMPS model
First tests underway with 20-yr segment of CCSM3-FV

23. Statistical downscaling
Uses “analogue” technique:
Start with daily CCSM3-FV data on coarse grid, and daily obs. data on fine grid (Mauer et al. 2002; PRISM data disaggregated to daily level using daily obs)
Coarsen obs to model grid
Compare model field to coarsened obs
30 closest matches (least RMSE) and optimal weights found
Weights applied to obs on original fine grid
Hidalgo et. al 2006, J. Climate, submitted

24. CCSM3-FV downscaling: Examples

25. CCSM3-FV downscaling: Precipitation monthly EOFs

26. CCSM3-FV downscaled: T-max EOFs vs. Obs

27. Runoff applied to river flow model

28. Sacramento River at Sacramento
Columbia River at the Dalles
Colorado River at Lees Ferry

29. Next steps with statistical downscaling
Have processed ~100 yrs of the 760 yrs available
Process rest of CCSM3-FV control run
Evaluate observed changes in hydrology against this estimate of unforced variability

30. PCM/VIC runs (Andy Wood, UW)
Historical simulations with estimated GHG and sulfate aerosols
4 ensemble members covering

31. PCM/VIC: Trend in Snow Water Equivalent

32. PCM/VIC: Trend in T-air

33. PCM/VIC: River flow

34. Amit Chourasia, SDSC, for the LUSciD project

35. PCM/VIC: River flow Amplitude shows strong decadal variability
Phase shows flow earlier in the year for some, but not all, rivers

36. Next steps for PCM/VIC
Process other ensemble members to reduce natural internal decadal variability
Is the forced change statistically significant?
How does it compare to observations?

37. Cooperative project #1:
Ocean Heat Content
SIO LLNL
Tim Barnett Krishna AchutaRao
David Pierce Peter Gleckler
Ben Santer
Karl Taylor

38. Motivation
Can GHG and sulfate aerosol forcing explain the warming signal in the world’s oceans?
YES!
(surprisingly well)

39. PCM Signal Strength & Noise

40. Inter-Model Comparison: N. Hem PCM signal + HadCM3 noise

41. What about other models?
38 realizations of 20th century climate from 15 coupled models in the IPCC AR4 archive are being analyzed.
Work in progress
Krishna AchutaRao; David Pierce; Peter Gleckler; Tim Barnett

42. MRI CGCM 2.3a
NCAR CCSM 3.0

43. Preliminary findings
Most models show a detectable warming signal in all the ocean basins with some exceptions
NCAR CCSM 3.0 shows large natural variability in the North Atlantic
Details of signal penetration in some ocean basins vary
More complex picture than the previous study (Barnett et al. 2005) that considered two models
Does the fidelity of model heat uptake relate to climate sensitivity?

44. Heat uptake vs. climate sensitivity
Note: Plot shows only a subset of the 15 models analyzed.

45. Volcanic Eruptions and Heat Content
P. Gleckler1, K. AchutaRao1, T. Barnett2, D. Pierce2, B.D. Santer1 , K. Taylor1, J. Gregory3, and T. Wigley4
(1PCMDI 2UCSD/SIO, 3U.Reading, 4NCAR)
How do volcanic eruptions affect ocean heat content?
Can this give insight into how ocean heat content anomalies are formed and propagate?

46. Background
Volcanic eruptions substantially reduced 20th Century ocean warming and thermal expansion.
Recovery from Krakatoa (1883) takes decades.
Effect of Pinatubo is much weaker than Krakatoa because it occurs against backdrop of substantial ocean warming.
Models including V forcing agree more closely with late 20th Century observations than those without V
Gleckler et al., Nature, 2006
Heat Content (1022 J)
Krakatoa
Pinatubo
Heat Content
Depth (m)
Temperature

47. Cooperative project #2:
Atmospheric water vapor
SIO LLNL JPL
Tim Barnett Peter Gleckler Eric Fetzer
David Pierce Ben Santer

48. Water vapor a key greenhouse gas
How well do models simulate it?
New 3-D satellite data set available
Compare to AR-4 model fields in PCMDI database

49. Annual mean: models too moist

50. Fractional errors greater with height

51. Summary
Goal: can we detect a global warming signal in main hydrological features of the western United States?
Long CCSM3-FV control run for estimate of natural internal variability is done
CCSM3-FV simulation comparable to other NCAR-series models
Statistical downscaling to give river flow is progressing
PCM runs give another estimate of natural varaibility, and also in this case of the forced signal
Other cooperative projects (ocean heat content, atmospheric water vapor) progressing well

53. Sierra snow pack
Now and ………………….………….future?

54. Downscaling: the motivation
Global model (orange dots) vs. Regional model grid (green dots)
Insert regions/multiple grid slide

55. CCSM3-FV: Temperature, MAM

56. CCSM3-FV: Temperature, SON

57. CCSM3-FV: Precipitation, MAM

58. CCSM3-FV: Precipitation, SON

59. CCSM3-FV: Precip Variablity, MAM

60. CCSM3-FV: Precip Variablity, SON

61. CCSM3-FV downscaling: T-max monthly EOFs

62. CCSM3-FV downscaling: T-min monthly EOFs

63. CCSM3-FV downscaled: Precipitation EOFs vs. Obs

64. CCSM3-FV downscaled: T-min EOFs vs. Obs

65. Spectrum of T-max PC#1 Observed M.Y. 240-289 M.Y. 290-339
(x axis is cycles per month)

66. Spectrum of T-min PC#1 Observed M.Y. 240-289 M.Y. 290-339
(x axis is cycles per month)

67. HadCM3 Signal Strength & Noise

68. Seasonal cycle in North Pacific
Blue = 10 yrs of model
Red = 3 yrs of AIRS

69. Work in progress: Evolution of the Krakatoa anomaly
GISS-HYCOM
NCAR CCSM3
6 realizations of CCSM3, GISS-HYCOM (and GFDL2.1)
Large inter-model differences
Uncertainties associated with external forcings, model physics and initial conditions
S/N analysis (in-progress) should help isolate model responses to eruptions - necessary to evaluate realism
Decadal average ocean heat content anomalies: zonally integrated cross-sections
Heat content (1016 Joules/m)

70. Spectrum of Precipitation PC#1
Observed M.Y M.Y
(x axis is cycles per month)