Tightly linked zonal and meridional sea surface temperature gradients over the past five million years

Transcription

1 SUPPLEMENTARY INFORMATION DOI: /NGEO2577 Tightly linked zonal and meridional sea surface temperature gradients over the past five million years Alexey V. Fedorov 1*, Natalie J. Burls 1,4, Kira T. Lawrence 2, and Laura C. Peterson 3 1 Department of Geology and Geophysics, Yale University, New Haven, CT Department of Geology and Environmental Geosciences, Lafayette College, Easton, PA Department of Chemistry and Environmental Studies Program, Luther College, Decorah, IA Center for Ocean-Land-Atmosphere Studies, Dept. of Atmospheric Oceanic and Earth Sciences, George Mason University, Fairfax, VA * To whom correspondence should be addressed. alexey.fedorov@yale.edu NATURE GEOSCIENCE 1

2 Figure S1 The same as in Fig. 2, but a linear trend of 0.2 o C/Myr was added to the temperature change in panel (c) resulting in warm pool SSTs being about 1.5 o C warmer during the Pliocene. This additional linear trend has been added only to the mean trend (red line), not to the original records. This modification is then included when computing changes in the meridional and zonal SST gradients in the bottom panel. 2

3 Figure S2 A scatter plot based on the changes in zonal and meridional SST gradients shown in Fig. 2d. Each data point is obtained by averaging over 400Ky worth of trend data. Negative values indicate reductions in the two SST gradients. The gray line shows the best linear fit (the slope of this relationship is 0.9±0.2 and correlation coefficient 0.95). 3

4 Figure S3 SST trends from the timeseries used in this study (lines) contrasted to the recently published TEX 86 data [7] superimposed (dots). Note the pronounced cold bias in the TEX 86 temperature estimates in the eastern equatorial Pacific over the entire 5 Myr time interval and in the western equatorial Pacific over the last 2Myr. 4

5 Figure S4 Examples of the regional extent of the imposed modifications in cloud albedo: These are the regions where the atmospheric water path was systematically modified to produce the results shown in Figs. 3 and 4. Dark (light) shading indicates the regions over which cloud albedo has been systematically reduced (increased). For further details of the approach see ref. [27]. 5

6 Figure S5 The close agreement between the SST gradients as defined for Fig. 3 and those based only on SST values within numerical experiments at the site locations shown in Fig. 1. Gray solid lines indicate a slope of 1. There is an offset of about 1 o C between the differently defined zonal gradients because of a cold bias in the model along the equator, but this offset does not affect the slope of the relationship between the zonal and meridional SST gradients. This plot confirms that the sparse sampling in Fig. 1a is sufficient to reconstruct the relationship between the two temperature gradients. 6

7 Figure S6 A schematic of the idealized analytical model, adapted from [24] and [27]. The upper ocean is divided into three boxes the Warm Pool box, the Cold Tongue box, and the Extra-tropical box. q indicates the ocean volume transport into the eastern equatorial box and is a function of the atmospheric zonal (Walker) and meridional (Hadley) circulations which are in turn coupled to the respective meridional ( Tuo merdional ) and zonal ( Tuo zonal ) upper ocean temperature gradients (within the top 50m). The temperature of each box is set by the ocean heat transport (parameterized as a function of q and the upstream temperature gradient) and the surface heat flux (parameterized as a restoring towards a local-equilibrium SST at a specified restoring timescale) into the box. T em and T et, represent the local-equilibrium SST (defined here as temperatures that the ocean would have in the absence of oceanic heat transport) for the mid-latitudes and tropics, respectively. For further details we refer the reader to Methods and refs. [24] and [27]. 7

8 Figure S7 Variations in key box model parameters estimated from the GCM simulations. Note that all experiments remain in the saturation regime (Q >1). 8

9 Figure S8 The dependence of the ratio of the merdional and zonal gradients, Tuo zonal / Tuo meridional, on parameters Q and, as computed from the analytical model (see Equation 4). The blue box indicates the upper and lower values for these parameters found within the range of numerical simulations conducted. Note that above saturation (Q >1), the ratio of the gradients depends only weakly on both Q and. 9

10 Figure S9 The linear connection between surface and upper-ocean temperature gradients for (a) meridional and (b) zonal gradients in our comprehensive GCM simulations. The upperocean temperature gradients are an average over the top 50m of the ocean. 10

11 Table S1: Details of ocean sites and proxy data SITE Lat Long Depth (m) Modern SST, o C Average Res. Method Time frame (Myr) 42ºS 178ºW kyr U k' References Present study 41 o N 33ºW kyr U k' [44] 806 0ºN 159ºE kyr Mg/Ca; U k' [10]; [3] 846 3ºS 91ºW kyr U k' [45]; [46] 847 0ºN 95ºW kyr; 13 kyr Mg/Ca; U k' [10]; [14] MD ºN 111ºW kyr U k' [7] 4ºN 43ºW kyr U k' [3] 58N 16ºW kyr U k' [47] 39 N 128 W kyr U k' [48] 43ºS 9ºE kyr U k' [49]; [50] 37N 158ºE kyr U k' 37; U k' [3]; [48] 1ºS 82ºW kyr U k' [51] 2ºN 142ºE kyr Mg/Ca [52] Table S1. Details of the and MD97 sites and the proxy data used in this study. The U K 37 -based SST record from site 1125 has been generated specifically for this study (for details see the Supplement text). Calibrations used are described in ref. [2] and references therein. In addition to the data used in this reference, we have also included U K 37 data for site 850 from [7]. 11

12 Table S2: Experiment details Experiment number CESM version Model resolution Experiment type/ forcing Experiment duration Interval used 1 T31gx3v7 Medium Control 800 yr last 100yr 2-21 T31gx3v7 Medium Albedo changes 800yr 100yr Extended 11,16 T31gx3v7 Medium Albedo changes 3000yr 500yr x1.25gx1v6 High Control 200yr 50yr x1.25gx1v6 High Albedo changes 200yr 50yr T31gx3v7 Medium C0 2 changes 800yr 100yr Table S2. Details of the numerical experiments shown in Fig. 3. Extended experiments are the same experiments as number 11 and 16, but run for 3000 years. Simulations with CESM T31gx3v7 and CESM 0.9x1.25gx1v6 are referred to as T31 and 1-degree experiments, respectively. 12

13 Table S3: Definition of the temperature gradients in numerical experiments Gradient Notation Depth average Definition Zonal SST gradient T zonal 0m Difference between maximum SST within the western equatorial Pacific warm pool (3 o S-3 o N, 140 o E-160 o E) and minimum SST within the eastern Pacific cold tongue (3 o S-3 o N, o E) Meridional SST gradient T meridional 0m Difference between mean equatorial (3 o S-3 o N, o E) and extra-tropical (50-30 o S, o E and o N, o E) SSTs Zonal upperocean temperature gradient Tuo zonal 0-50m Difference between mean western (8 o S-8 o N, o E) and eastern tropical Pacific (8 o S-8 o N, o E) upper ocean temperature Meridional upperocean temperature gradient Tuo meridional 0-50m Difference between mean tropical (8 o S-8 o N, o E) and extra-tropical Pacific (65-25 o S, o N) upper ocean temperature Table S3. Definitions of temperature gradients as used in Fig. 3. Note that the zonal SST gradient is defined between two relatively small regions in the western and eastern equatorial Pacific. Also note that the regions for computing upper-ocean gradients are chosen to be much wider in order to describe the broad subsurface connection between mid and low latitudes and also to be consistent with the idealized model. In fact, this choice of the averaging regions gives the best correlation between changes in the two gradients (correlation coefficient 0.97). 13

14 Table S4: Connection between the meridional and zonal temperature gradient changes The ratio of SST gradient changes, T zonal / T meridional The ratio of upper-ocean temperature gradient changes, Tuo zonal / Tuo meridional Paleo observations 0.9 ± 0.2 N/A GCM simulations 1.1 ± ± 0.07 Idealized model 1.1 ± ±0.15 Table S4. The ratio between changes in the zonal and meridional temperature gradients (surface and upper-ocean) as estimated from the paleo observations, GCM experiments and the idealized box model. The uncertainty ranges are given by the scatter of the data used to generate these estimates. 14