1. Pacific Secular Variation A result of hot lower mantle David Gubbins School of Earth Sciences University of Leeds

2. Thermal Core-Mantle Interaction (hot) (cold)

3. Lateral variations in heat flux boundary condition on spherical rotating convection can: Drive thermal winds “lock” core convection… …and delay drift of convection rolls Produce resonance of length scales… …and secondary resonances Force a lateral scale on the convection Indirectly produce similar scales on the magnetic field

4. The effect of lateral variations is weakened by: Low Prandtl number (inertia) Disparity of length scales between convection and boundary conditions High Rayleigh number (time dependence)

5. Geophysical Input for Core Heat Flux Mantle convection studies suggest large variations in lateral heat flow (100%) …and thermal boundary layer at the base of the mantle (D”) Seismology suggests a boundary layer 250 km thick …with temperature variations of 500 K

6. Observational Evidence of Lateral Variations Modern geomagnetic field Time-average of paleomagnetic field Persistent reversal paths Non-axisymmetric variations in secular variation Low secular variation in Pacific

8. OVERVIEW Evidence for low secular variation in the Pacific - historical and paleomagnetic Lateral heat variations on the core-mantle boundary Simple thermal convection influenced by the boundary Relationship with numerical dynamo simulations and application to the Earth’s core Implications for the thermal state of the core

9. Declination AD 1650

10. Declination AD 1990

11. Declination at Hawaii and Greenwich Meridian

12. Inclination Hawaii and Greenwich meridian

14. Looking for weak Secular Variation Historical record shows little SV in Pacific 400 years is not long enough to be definitive We need 5-50 kyr Big Island, Hawaii, offers 35 kyr with dating

16. Volcanoes of Big Island, Hawaii

19. Mean residual -2.8 o +/- 0.3 o

21. D from flows dated by C 14, Big Island, Hawaii

22. I from flows dated by C 14, Big Island, Hawaii

23. Kilauea East Rift Zone Drilling

24. Hawaiian data last 50 kyr from borehole data and surface flows

25. The Tangent Cylinder

28. Convection with laterally varying heat flux depends on 3 important parameters 1. Ekman number 2. Vertical Rayleigh number where h is the mean surface heat flux 3. Horizontal Rayleigh number where q is the lateral variation of heat flux, average zero

29. 3 LIMITING CASES R v =0: thermal wind R h =0: convection with uniform boundaries R h =0.3R v : convection heated from below and influenced by the boundary variations

30. “Thermal Wind”, R v =0, E=2x10 -4

31. Uniform boundaries E=2x10 -4, R h =0, R v =1.1 R v c

32. Uniform boundaries, equatorial slice

33. Inhomogeneous boundary conditions (periodic solution) surface flow and temperature R h =0.3 R v, E=2x10 -4, R v =1.1 R v c

34. Inhomogeneous boundary conditions R h =0.3 R v, E=2x10 -4, R v =1.1 R v c

35. SUMMARY Boundary heat flux based on shear wave anomalies can inhibit convection at the top of the core below the hot region corresponding to the Pacific… …because the anomaly there is longitudinally broader than in the Atlantic/Africa This convective flow does not generate a magnetic field

36. COMPARISON WITH A GEODYNAMO SIMULATION This convective flow does not generate a magnetic field Bloxham’s geodynamo simulation exhibits a time average that reflects the boundary conditions… …but does not give low Pacific SV or a field that resembles the time average at any instant of time The principle difference is not the magnetic field… It is probably the higher R v in the dynamo simulation

37. APPLICATION TO THE EARTH Resonance with the boundary arises because of similarity in length scales of convection and boundary anomalies Small E (10 -9 ) in the core implies a small scale but magnetic forces increase it A higher supercritical R v is needed for dynamo action, but this produces magnetic fields that are too complex, both spatially and temporally Again, the in the low E regime dynamo action may occur at lower supercritical R v because of its organising effect on the flow

38. IMPLICATIONS FOR CORE HEAT FLUX D’’ slow fast low heat flux high heat flux Difference in V s implies temperature difference 500 K in 250 km

39. HORIZONTAL VS VERTICAL HEAT FLUX Lateral temperature difference 500 K Within D’’ thickness 200 km Thermal conductivity 10 W/m/K Gives heat flux variation 1 TW =… 20% of conventional estimate of vertical heat flux May be larger locally

40. CONCLUSIONS The evidence for weak secular variation in the Pacific is quite strong Simple thermal convection calculations show this can come about from lateral variations in heat flux through the boundary These flows are too simple to generate a magnetic field, and numerical dynamo simulations give magnetic fields that appear more complex than is observed Lateral heat flux variations in D’’ appear to be large enough to cause this effect, provided large scale flow is maintained in the core