QnAs with Gary A. Glatzmaier

In the interiors of stars and most planets, global magnetic fields are generated by thermal convection of an electrically conducting fluid under the influence of rotation. Planetary scientist Gary A. Glatzmaier, Professor Emeritus of Earth and Planetary Sciences at the University of California, Santa Cruz and a member of the National Academy of Sciences, explores the dynamics of planetary and stellar interiors. He produces 3D time-dependent simulations of thermal convection and magnetic field generation. Glatzmaier has modeled the solar dynamo, the geodynamo, and recently Jupiter’s dynamo, just as NASA’s Juno mission is returning high-fidelity data of Jupiter’s near-surface magnetic and gravity fields. His results, reported in his Inaugural Article (1), describe how certain banded patterns in Jupiter’s near-surface fields, if detected by Juno, would help answer the longstanding question about the depth to which the jet streams observed on Jupiter’s surface extend below its surface. Glatzmaier recently spoke to PNAS about his findings.

Gary A. Glatzmaier. Image courtesy of Gary A. Glatzmaier.

PNAS: How did you become interested in modeling the convection of celestial bodies?

Glatzmaier: I was a graduate student in physics at the University of Colorado, Boulder, when I became interested in solar physics, in particular studying convection and magnetic field generation deep within the interior of the Sun. The approach I chose was to design and write a 3D computer code that I used to investigate the structure of convection in a rotating density-stratified body like the Sun. Since then I have been developing more sophisticated codes that include magnetic field generation, that is, convective dynamo models.

PNAS: What data are needed to model the interior of a planet?

Glatzmaier: The coupled set of nonlinear equations that are solved are based on the conservation of mass, momentum, and energy. The details depend on the particular planet or star that is being simulated.

The input data are estimated quantities like the total mass of the body, its radius, average rotation rate, and spherically symmetric profiles in radius of the reference state density, pressure, and temperature. Fluid properties like the viscous, thermal, and magnetic diffusivities are also prescribed as functions of radius. Then the computer solves the set of equations during each numerical time step to update the 3D perturbations in the density, pressure, temperature, and the three vector components of the fluid velocity and magnetic field.

Typically, for a single computer simulation, my code updates all of these variables on several hundred million grid points within the 3D spherical model during each of several million numerical time steps. Such a simulation can take more than a year of supercomputer time running on 500 parallel processors.

For Jupiter, I compare the simulated fluid flow, heat flow, and magnetic field at the model surface to observations on Jupiter’s surface to judge how realistic the simulation may be. When there is relatively good qualitative agreement, we begin to have confidence in the dynamics the model is simulating in the deep interior and can begin to understand the physics of Jupiter’s convective dynamo by analyzing the simulation.

PNAS: How do your Jupiter simulations relate to preliminary results from NASA’s Juno mission that were recently published in Nature (2)?

Glatzmaier: The Juno gravity measurements from two near-surface orbital passes in December 2016 and May 2017 are very exciting. Besides detecting the slightly oblate shape of Jupiter’s gravity field, which was expected because of the planet’s slight flattening due to its rotation, gravity variations not symmetric with respect to Jupiter’s equator were detected. This strongly suggests that the banded zonal wind jets, long observed on Jupiter’s surface, extend well below Jupiter’s shallow gaseous atmosphere. These preliminary results are exciting because the depth of these winds has been debated by researchers for decades. The majority of papers published on this issue have come from the climate community, which uses models and approximations that are valid for the Earth’s shallow atmosphere; they have argued or assumed that the winds observed on Jupiter’s surface are also maintained just within its shallow gaseous atmosphere. But Jupiter has a deep liquid interior below its atmosphere, not a solid surface.

Deep convection models, on the other hand, naturally maintain banded zonal winds at the surface with a strong eastward-directed jet at the equator, like Jupiter’s and unlike the Earth’s. The zonal winds simulated by these deep interior models extend well below the shallow atmosphere. There is still a question of just how deep into the interior they exist in Jupiter, which is the focus of my Inaugural Article (1). Although the recent results from Juno do seem to support deep zonal winds, I think it is clear to most researchers in this field that the smaller-scale flows observed on Jupiter’s surface are likely maintained by processes within its shallow atmosphere.

PNAS: How are the shallow surface clouds of Jupiter connected to the planet’s rocky core?

Glatzmaier: Well, that’s hard to know. The clouds observed on Jupiter’s surface are transported by the banded eastward and westward jet streams, relative to Jupiter’s basic rotation rate. The clouds are maintained by moist convection in Jupiter’s shallow atmosphere, which is far above Jupiter’s relatively small rocky core.

However, one of my Jupiter simulations produced a pair of east–west oscillating zonal wind jets at midlatitude in both hemispheres, in addition to other nonoscillating jets at other latitudes. The oscillating jets are a manifestation of an inertial wave, driven by Coriolis restoring forces, that extends from the surface down to the model’s small rocky core. The latitudes at which the wave intersects the surface and oscillation frequency of this wave are directly related to the size of the model’s core. If this were observed on Jupiter it would provide new evidence of the existence and size of a solid core. It would be really exciting if such an observation were made. However, this feature in the computer simulation comes and goes; so not observing such a wave on Jupiter’s surface would not necessarily rule out the existence of a rocky core. Also, the computer model is far from being a perfect tool for predicting such processes.