OMWG meeting June 2019 Impact of climate change on the Gulf Stream and US East coast warming in CESM: An Update
Justin Small Dan Whitt, LuAnne Thompson John Truesdale, Nan Rosenbloom Susan Bates, Gary Strand, Jerry Meehl, Don Wuebbles
Acknowledging: Mike Alexander, Adam Phillips NW Atlantic SST: observations of warming • Observed warming off US East coast in recent decades – Burrows et al 2011, Pershing et al 2015 –link to Gulf Stream and PDO – Thomas et al 2017, Gulf of Maine lengthening of warm season, partly related to Gulf Stream index (more northern) • Effect on fisheries – E.g. Cod stock reduction (Pershing et al 2015 First high-resolution modelling paper to describe East coast warming: Saba et al 2016 SST difference (averaged over years 60-80) between 1% per
year CO2 increase run and preindustrial control. Bottom right is high-resolution model CM2.6. Model resolutions are labelled. Note enhanced warming in NW Atlantic in CM26.
Bottom temperature change, CM26
From Enhanced warming of the Northwest Atlantic Ocean under climate change Vincent S. Saba1,*, Stephen M. Griffies2, Whit G. Anderson2, Michael Winton2, Michael A. Alexander3, Thomas L. Delworth2, Jonathan A. Hare4, Matthew J. Harrison2, Anthony Rosati2, Gabriel A. Vecchi2 andRong Zhang2 2016, JGR Models and Interpretation Warming of western boundary currents Lixin Wu et al 2012- WBCs warm, shift in current climate Yang et al. 2016, WBCs warm , intensify, shift poleward, except Gulf Stream (weakens, AMOC influence) Saba et al 2016 – Gulf Stream northward shift Labrador Current retraction AMOC weakening Warming notable at canyons Caesar et al 2018, Sanchez-Franks and Zhang 2015 East coast warming-strong link to AMOC Use temperature-based Gulf Stream path Weak AMOC – northward GS path. See Zhang and Vallis 2007, Claret et al 2018 Rapid deoxygenization of Northwest Atlantic shelf Linked to Labrador Current retreat Alexander, M. A., S. Shin, J. D. Scott, E. Curchitser, C. Stock, 2019: The Response of the Northwest Atlantic Ocean to Climate Change J. Climate , submitted. ROMS downscaling of CMIP global models Investigate “delta” anomalies of RCP8.5 (2070-2099) relative to present day 3 Simulations: large spread in downscaled solution Climate change scenario with High- resolution CESM • NSF allocation: High Resolution Earth System Modeling for International Climate Assessment Using Blue Waters Capabilities • PI: Don Wuebbles (University of Illinois) – The NCAR collaborators and co-PIs: Drs. Warren Washington, Jerry Meehl, Justin Small, and Joseph Tribbia, Susan Bates, Dan Whitt – External collaborators: LuAnne Thompson, Miguel Jiminez-Urias, Mike Alexander, Adam Phillips • High-resolution CESM – CAM5.2 SE, 0.25deg – POP2, 0.1deg – CESM1.0.4, Small et al 2014, CESM1.3beta17 (Meehl, Bates et al 2019) • Branching off from year 50 of the previous high resolution control run (year 2000 conditions), – ran with historical conditions for years 2001-2005, – then with the RCP8.5 scenario for 2006-2097. – Control run extended to 135 years (85 years to compare with RCP8.5) • Run on Blue Waters supercomputer (University of Illinois) – Accompanying simulations include 0.25deg atmosphere, 1deg ocean Aims
• Quantify US East coast warming in high resolution CESM and CESM-LE • Global models, continuous simulations – RCP8.5 • Processes – Gulf Stream shifts? – Role of surface heat fluxes vs ocean dynamics – Role of Labrador Current/AMOC Recap/update Comparison with CESM-LE CESM-Large Ensemble (LE) – Kay et al. 2015, Deser et al. Long multi-century 1850 control 1850-1920 (1 simulation) 1920- 2100+ (40+ ensembles) • CESM-HR • Small et al 2014, Abernathey et al • Century-long year 2000 control (100+ years) • Branched from year 50 to an RCP8.5 • 1 ensemble member • We compare beginning to end of 21st C CESM-LE
HI-RES
The change in SST from 2010s to 2090s from CESM-LE (above) and CESM-HR (right) CESM-LE
HI-RES
The change in SST from 2010s to 2090s from CESM-LE (above) and CESM-HR (right) However – Low-res starts off much warmer than Observations
LOW-RES OCEAN SST BIAS LR minus OBS
SST bias, CESM Similar with 0.25deg improvement in atmosphere, 1deg North-west ocean. Relative to Pacific/Kuroshio Reynolds (2007). and in Southern Annual mean Ocean/ACC
CHANGE DUE TO HIGH-RES OCEAN LR minus HR OCEAN
SST difference, Sign convention CESM with – matching 0.25deg colors (top and atmosphere: 1deg. bottom) implies Ocean minus improvement 0.1deg ocean. with resolution. . Processes
• Surface heat flux – Warming or cooling? • Gulf Stream shifts (?) – Gulf Stream is located by fitting an error function to SSH. – Note, we are not using a temperature-based metric as the ocean warms with climate change • Changes to Labrador Current and AMOC Focus on Sub-polar gyre: 2010s to 2090s 2010-2019 to 2090-2097 CESM1: SST change CESM1: SSH change
cm Excepting SSH, background contours are the 2010-2019 values Surface heat flux change. Positive=ocean gains heat CESM1: SSS change
W/m2 PSU Circulation changes
Zhang&Vallis 2007
• Bottom vortex stretching/bottom torque important term in vorticity equation at western boundary- – Zhang and Vallis 2007, – Strong DWBC interacts with topography causes downwelling – Downwelling induces vorticity – Strong northern recirculation gyre – Keeps Gulf Stream separating off Hatteras , in southerly position – Note: other theories exist for Gulf Stream separation – See also Yeager & Jochum 2009, Yeager 2015 • Climate change, AMOC weakening – Saba et al 2016, Caesar et al 2017 – Weakening sub-polar gyre – Lab Current weakens – Weakening northern recirculation gyre of Gulf Stream – Gulf Stream shifts north? Circulation changes
10-year means These plots are for three decadal averages of sea surface height 2010-2019 average, 2040-2049 Average and 2090-2097 Average Position and strength found by fitting the SSH to an error function
Height Differences across the stream: weakening over time particularly downstream Position of the Gulf Stream: little change except in the large meander after separation
2010-2019 to 2090-2097 CESM1: SST change CESM1: SSH change
cm Excepting SSH, background contours are Location of GS from SSH fit to error Location of Gulf Stream from Max function Geostrophic Velocity
Note that the Gulf Stream moves slightly north in 2040 from both metrics on left, but then moves back to the 2010 position in 2090 Location where 15C isotherm crosses 200m shows that in 2090, most of the water in the region is above 15C 200m. Even in 2040 this metric shows a movement of several degrees to the North, while the surface express does not show this change
mean latitude of the 15 °C isotherm at a 200-m depth in the Northwest Atlantic, between 75° W and 55° W
2010 2040 2090 Changes to Labrador Current, Gulf Stream
From Saba et al
Labrador current has less intrusion into East coast waters CONTROL CO2 doubling
Saba et al. hypothesise that weakening of AMOC (later slides) leads to retreat of Labrador current, northward shift of Gulf Stream, replacement of cold Labrador slope water by warm Atlantic slope waters (Saba et al. 2016) Blow up near the NE corner of the current coming around the grand banks.
Next slide shows cross sections across Labrador Sea Meridional Current Potential Salinity Temperature Labrador Current
Zonal Sections across Labrador Current, 58-60N
In addition to weakening of Labrador Current, it transports warmer water AMOC weakening
Figure courtesy of Frank Bryan, Dan Whitt. Max AMOC reduces by about 10Sv in 21st C in both CESM-LE and CESM-HR. Way Forward
• iHESP simulations to look at natural variability – International laboratory for High resolution Earth System Prediction – Collaboration with TAMU, Qingdao National Marine Laboratory – Small ensembles of CESM-Large-Ensemble-type simulations at high-res – HighResMIP simulations • Heat budget analysis (US North-East coast) – More detail on role of lab Current change, Gulf Stream, surface heat flux Aims Conclusions
• Quantify US East coast warming in high resolution CESM and CESM-LE – 2-4degC in CESM-LargeEnsemble over 21st century in RCP8.5 – 4-8degC in CESM-HighRes – However low-res starts off 4-6degC warmer than HR and 6-8deg warmer than obs • Gulf Stream shift – Some shift close to separation point – But not east of 65W – Temperature-based Gulf Stream metric not useful • Role of surface heat fluxes – Surface heat flux anomaly opposes SST anomaly – Ocean dynamics must be providing warming (& not seen in LR) • Investigate the relationship to Labrador Current, – Labrador Current weakens, retracts – Labrador Current warms – Meanwhile AMOC reduces in CESM-HR at similar rate to CESM-LE • Overall Summary -Large changes to Labrador Current most likely cause of warming on US East coast in High-Res – linked with better resolution of topography, pathways in Gulf of Maine etc Location of maximum and minimum doesn’t change much except in the large meander Minimum SSH across the Gulf Stream reduces down stream, most of the change in SSH across the Gulf Stream occurs North of the Gulf Stream
North of Gulf Stream
South of Gulf Stream
I may drop this Cross stream sea surface height structure (surface geostrophic streamline) for transects as in Dan’s ppt. Note that the GS is getting narrower (and thus the velocity is stronger), but the overall height difference is getting weaker (as in the error function fit). Thus the velocity integrated across the stream is weakening. But total top to bottom transport could be different.
Numbers in box indicate the total height difference (max-min) Surface Geostrophic velocity: The maximum moves North in 2040 after the meander, and then moves south again in 2090. The max velocity is smallest in 2040 of the three time periods Gulf Stream sections – CESMLE Gulf Stream sections – ASD-like RCP8.5 What about climate model drift?
CESM1 RCP8.5: SST change CESM1 Control: SST change
Based on first 50 years of RCP8.5 run vs corresponding 50 years of control run. Control run does have slightly enhanced warming off US East coast but much weaker High-resolution CESM diff 2090s minus 2010s
°C/85 years /80years
Compared to LE ensemble- mean, CESM-High-res has less warming in Southern Hemisphere, more warming off US East coast, similar cooling in Sub-Polar gyre CESM-HR
Trenberth-Fasullo 2017 CESM-HR
Annual mean net surface heat flux: positive cools ocean
CESM-LR CESM-LE
HI-RES The change in surface heat flux from 2010s to 2090s from CESM-LE (above) and CESM-HR (right). Positive=ocean gains heat
Note sub-polar gyre is warmed in all simulations: the main difference is US East coast which is cooled in CESM-HR W/m2 CESM-LE
HI-RES The change in surface heat flux from 2010s to 2090s from CESM-LE (above) and CESM-HR (right). Positive=ocean gains heat
Note sub-polar gyre is warmed in all simulations: the main difference is US East coast which is cooled in CESM-HR W/m2 Timeseries
Gulf of Maine North-west corner
Sea surface Temperature Sea surface Temperature
degC degC
Surface Heat Flux Surface Heat Flux
Positive warms ocean Positive warms ocean
W/m2 W/m2 Implications of surface heating
• Large warm SST associated with large heat loss from ocean – Indicative ocean forcing SST signal • Anomalous warming accompanied by weaker surface heating of ocean – Indicative atmosphere’s role is weak • See Bishop et al 2017,J. Climate Transport is slowing overall Gulf Stream sections – ASD-like RCP8.5 - jet is more localized, stronger, and 10-year means slightly to the north at 63 W - Northward jets are weaker at 31 N
NOTE DIFFERENT COLORBAR S Surface Geostrophic velocity: Note that the Labrador Current does seem to be weakening over time.
Temperature Averaged between 58 and 60N Salinity CESM-large ensemble with 1.0 degree ocean
30+ members
ASD-like RCP8.5 w/ 0.1 degree ocean 1 member, but 10-year means: 2006-2015 2080-2089
Latitude SST change SSH change
cm
2010-2019 to 2090-2097
Note that overall, the North Atlantic subpolar gyre is strongly heated, but see close up of US NE coast next slide. Surface heat flux change. Positive=ocean gains heat
W/m2 Focus on US East coast: 2010s to 2090s 2010-2019 to 2090-2097 CESM1: SST change CESM1: SSH change
cm Excepting SSH, background contours are the 2010-2019 values
Surface heat flux change. Positive=ocean gains heat CESM1: SSS change
W/m2 PSU