*2.7. Experimental Design*

In designing the simulations presented in this paper, the author had several goals: (1) to test the GLOM's ability to reproduce the gross circulation and stratification structure of the world ocean; (2) to illustrate several unique capabilities of the GLOM that stem from its Lagrangian nature; and (3) to make a contribution to the field of physical oceanography. After reflecting on these goals, he chose to use the GLOM to extend the results of Haertel and Fedorov [15] (HF12) to include realistic topography and a global domain. Briefly, HF12 addressed the question of to what extent ocean circulation and stratification depend on interior mixing. They used a predecessor to the GLOM to simulate circulations in an idealized ocean with the scale of the Atlantic, and which included a circumpolar channel. They compared a simulation with a moderate amount of tracer mixing to one in which the tracer diffusivity was set to zero. They found that the leading order solution for ocean circulation, stratification, and heat transport could be reproduced with zero tracer diffusivity, and that interior mixing essentially contributed first-order perturbations to this solution. Accordingly, in this paper, we apply a surface forcing much like that used by HF12, but instead of using an idealized ocean basin, we use a global ocean with a smoothed version of actual bathymetry.

The temperature restoring function (Figure 5a, solid black line) is a piecewise linear function with a range set so that zonal average model SST (Figure 5a, red dashed line) has similar minima and maxima to that of zonal average SST in nature (Figure 5a, blue dotted line). Note that the restoring temperature is slightly lower in the Antarctic than in the Artic, which leads to Antarctic Bottom Water being more dense than North Atlantic Deep Water. The idealized zonal wind stress forcing (Figure 5b) is the same as that used by HF12; it includes strong westerlies at midlatitudes and weaker easterlies in the tropics.

**Figure 5.** Surface forcing. (**a**) restoring temperature (solid black line), average SST in the simulation with mixing (red dotted line), and observed average SST (blue dotted line), which is from the NCEP Optimally Interpolated weekly SST and Sea Ice datasets for the years 1998–2009. (**b**) zonal wind stress.

The forcing was applied to an initially isothermal ocean (T = 0 ◦C) for a period of 1400 years. In one case, there was a moderately high vertical tracer diffusivity (1 cm<sup>2</sup> s<sup>−</sup>1), and in the other case the diffusivity was zero. One minor adjustment was made after 500 years of model integration–the temperature difference between the restoring temperature in the Artic and Antartic was reduced from 1 ◦C to 0.5 ◦C to deepen the Atlantic Meridional Overturning Circulation (AMOC). By the year 1400, the simulation with interior mixing was in an approximately steady state in the upper ocean, with very weak cooling in the abyss. In the simulation without interior mixing, weak cooling continued in the

deep ocean and abyss. However, the temperature here had reached within a few tenths of a degree of the minimum restoring temperature, limiting the amplitude of possible future temperature changes.
