Decadal predictability of tropical Indo-Pacific Ocean temperature trends due to anthropogenic forcing CCSM3 DJF A1B Surface Temperature Trend 2018-2062 Amy Solomon and Matt Newman CIRES/University of Colorado & NOAA/ESRL 2011 CESM Workshop, Breckenridge, Colorado 1
Outline 1) CCSM3 Large Ensemble (40 members, 60 years, A1B) 1) Unfiltered trend patterns 2) Defining natural variability (ENSO) 3) Isolating a predictable signal 2) Application to CCSM4 RCP8.5 (5 members, 2005-2100) 2011 CESM Workshop, Breckenridge, Colorado 2
50-year Tropical Indo-Pacific 10m Temperature Trends from 8 A1B CCSM3 Ensemble Members 2011 CESM Workshop, Breckenridge, Colorado 3
Motivation For decadal climate forecasts to be useful they must provide verifiable regional skill on 10-30 year time scales CCSM3 A1B Ocean Temperature Trend 10 meters Since on these time scales natural variability and the response to external forcing are of the same order a strategy is needed to Assess the skill of decadal forecasts in a way that will provide insight into the contributions (and potential interactions) of forced and natural variability Projection on 40 Ensemble Members Compare this skill across models 2011 CESM Workshop, Breckenridge, Colorado 4
Assumptions of Analysis 1) Skill on decadal time scales comes from the forced response of the climate system to steadily increasing greenhouse gases CCSM3 A1B Ocean Temperature Trend 10 meters 2) Skill on decadal time scales is limited by natural variability much of which acts as climate noise Projection on 40 Ensemble Members 2011 CESM Workshop, Breckenridge, Colorado 5
Details of Analysis With a Focus on---- Wintertime (DJF) variability CCSM3 A1B Ocean Temperature Trend 10 meters Using--- 1) CCSM3 A1B Large Ensemble --40 60-year integrations --same Jan 1, 2000 ocean state --A1B+Commitment runs Ocean temperature down to 300m In the --- Projection on 40 Ensemble Members Tropical Indo-Pacific Basin 2011 CESM Workshop, Breckenridge, Colorado 6
Is the A1B Trend El Niño-like? A1B Ocean Temperature Trend (domain mean removed) Maximum warming in the central equatorial Pacific --- El Niño-like Asymmetric North-South warming --- not El Niño-like 2011 CESM Workshop, Breckenridge, Colorado 7
Is the A1B Trend El Niño-like? A1B Ocean Temperature Trend (domain mean removed) Maximum warming in the central equatorial Pacific --- El Niño-like Asymmetric North-South warming --- not El Niño-like Sharpening of the equatorial thermocline --- not El Niño-like 2011 CESM Workshop, Breckenridge, Colorado 8
Is the A1B Trend El Niño-like? A1B Ocean Temperature Trend (domain mean removed) Maximum warming in the central equatorial Pacific --- El Niño-like Asymmetric North-South warming --- not El Niño-like Sharpening of the equatorial thermocline --- not El Niño-like Off-equatorial steepening of the thermocline --- El Niño-like 2011 CESM Workshop, Breckenridge, Colorado 9
Is the Trend Predictable on Decadal Time scales? Maximum warming in the central equatorial Pacific --- El Niño-like Asymmetric North-South warming --- not El Niño-like A1B Ocean Temperature Trend (domain mean removed) Sharpening of the equatorial thermocline --- not El Niño-like Off-equatorial steepening of the thermocline --- El Niño-like Signal > Noise for timescales greater than ~25 years 2011 CESM Workshop, Breckenridge, Colorado 10
CCSM3 ENSO Structure Natural patterns of variability isolated in control (fixed forcing) simulations prevents aliasing of trend onto natural patterns Useful when working with small forced ensembles 10 meters EQ Using 3D EOFS of Ocean Temperature With a focus on ENSO variability as the dominant source of noise in the IndoPacific Projection on 40 A1B Ensemble Members 150 W 2011 CESM Workshop, Breckenridge, Colorado 11
CCSM3 ENSO Structure Natural patterns of variability isolated in control (fixed forcing) simulations prevents aliasing of trend onto natural patterns Useful when working with small forced ensembles 10 meters EQ Using 3D EOFS of Ocean Temperature ----Initial focus on ENSO variability---- 150 W Projection of ENSO variability on A1B Ensemble has a trend Projection on 40 A1B Ensemble Members that is not predictable on decadal time scales 2011 CESM Workshop, Breckenridge, Colorado 12
CCSM3 A1B Non-ENSO Trend Straightforward strategy to remove the contribution of natural variability (ENSO) to the A1B Trend: domain mean removed 10 meters EQ Remove the projection of the dominant 3DEOFs from the control run from each A1B ensemble member 150 W Projection on 40 A1B Ensemble Members 2011 CESM Workshop, Breckenridge, Colorado 13
CCSM3 A1B Non-ENSO Trend Increase in near-surface equatorial temperature gradient --- consistent with the ocean thermostat hypothesis (Clement et al. 1996) and upwelling of cooler subtropical waters (Seager and Murtugudde 1997) domain mean removed 10 meters EQ Largest trends in the Warm Pool Prominent steepening of the offequatorial thermocline --- in the S. Pacific only --- consistent with increase in southeast trades (Xie et al. 2010) 150 W Significant collapse of ensemble spread! Projection on 40 A1B Ensemble Members 2011 CESM Workshop, Breckenridge, Colorado 14
Perfect Model Skill of Trend Patterns Amplitude of Total trend pattern is predictable after 23 years TOTAL! S!1 Amplitude of ENSO trend pattern is not predictable in these 50-year simulations Amplitude of Non-ENSO trend pattern is predictable after 10 years Average anomaly correlation (ρ(τ)) between individual ensemble members and ensemble mean as a function of forecast lead (τ) ρ(τ) = S(τ) / ( 1 + S(τ) ) 1/2 S= ensemble mean / ensemble spread 2011 CESM Workshop, Breckenridge, Colorado 15
Comparison of A1B vs. RCP Simulations 10 9 RCP8.5 8 Total radiative forcing (W/m 2 ) 7 6 5 4 3 2 A1B IMAGE E1 RCP 2.6 W/m2 RCP 4.5 W/m2 RCP 8.5 W/m2 A1B RCP4.5 A1B SRES (6.05W/m 2 ) E1 1 0 1980 1990 2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100 Year 2011 CESM Workshop, Breckenridge, Colorado 16
CCSM4 RCP8.5 5m 50-Year Trends 2005-2054 2011 CESM Workshop, Breckenridge, Colorado 2050-2099 C/year 17
Evolution of Optimal Singular Vector CCSM4 PICNTRL SODA 1958-2007 2011 CESM Workshop, Breckenridge, Colorado 18
CCSM4 RCP8.5 Projection on PICNTRL EOFS CCSM4 PICNTRL EOFs Projection on RCP8.5 2005 2045 2085 2005 2045 2085 2011 CESM Workshop, Breckenridge, Colorado 19
CCSM4 RCP8.5 S/N-MAX-3DEOF at 5meters Using Unfiltered Data 54.3% 2011 CESM Workshop, Breckenridge, Colorado 20
CCSM4 RCP8.5 S/N-MAX-3DEOF at 5meters Using Unfiltered Data 54.3% Removing Projection on PICNTRL 3DEOF1 52.9% 2011 CESM Workshop, Breckenridge, Colorado 21
Trend Patterns at 5meters CCSM4 RCP8.5 Total Trend Pattern CCSM3 A1B Total Trend Pattern 2011 CESM Workshop, Breckenridge, Colorado 22
Non-ENSO Ensemble Mean 3D-EOF1 at 5-10 meters CCSM3 A1B Non-ENSO Pattern 21.6% CCSM4 RCP8.5 Non-ENSO Pattern 52.9% 2011 CESM Workshop, Breckenridge, Colorado 23
Thank you for your attention! Questions? 2011 CESM Workshop, Breckenridge, Colorado 24
CCSM4 RCP8.5 Skill of Trend Patterns Expected Skill of 3D-EOF1 Patterns Non-ENSO TOTAL ENSO 2011 CESM Workshop, Breckenridge, Colorado 25