Hi Hippie.
I'm glad you appreciate my posts. Apologies if I sounded a bit frustrated in the last one, but I think there is some serious talking at cross-purposes going on here, which is always frustrating. I will try to do better this time.
The first question is, can convection occur in the outer core?
There's all this smoke and noise about viscosity, and the wide range of estimates. This is because it's actually a very hard thing to measure.
This paper does indeed give a wide range of estimates, covering a wide range of different methods for indirectly estimating outer core viscosity. This indirectness is why there's so much uncertainty. But does it matter? As previously stated, values of the individual properties by themselves are not diagnostic all that is required is that the Rayleigh number of the system is above the critical value of ~1000. So, lets do some calculations.
The Rayleigh number Ra = g.a.p.T.d^3 / k.h
Here's some estimates of all the parameters in the above equation other than viscosity (h):
g = acceleration due to gravity = 10 ms-2
a = thermal expansion coefficient = 6e-6 K-1
k = thermal diffusivity = 1e-5 m2s-1
T = temperature difference = 1000 to 2000K
p = density = 1e7 to 1e9 kgm-3
d = thickness of layer = 2300 km (2,300,000 m)
I doubt you'll dispute g. The next three parameters are derived from lab experiments which try to measure these properties at conditions similar to the core. There is some uncertainty here, particularly in T, because we don't know the exact composition of core material. The last two have been derived from seismological data, and are pretty reliable.
Plug these into the equation and you get:
Ra = 7.3e30 / h
So with a range of viscosity from 10e-3 to 10e11 Pas (coverted to SI from the figures in the paper above: 1 poise = 0.1 Pa S) we can estimate that the Rayleigh number for the outer core ranges from:
7.3e19 < Ra < 7.3e33
So, yes, there's a large range of possible values, but even the lower bound is still 16 orders of magnitude above the critical number, so convection is no only likely, but will be extremely vigorous. We'd have to be a long long way out on our numbers for this not to be the case.
(I'll just add here that the viscosity of the mantle is approximately 10e19 Pa.s, and it still convects, because the Rayleigh number is high enough).
The next issue is, can this convection generate a magnetic field? This seems to be where the real confusion is (on my side as well as yours, perhaps). I'll try and explain it as I understand it, perhaps you can show where you're unhappy.
A conductor moving in a magnetic field generates a current; therefore a convection current involving liquid metal, moving in the present magnetic field, will generate a current.
This current generates a magnetic field.
In the outer core, these processes feed back into each other - the currents generate magnetic fields which maintain the currents. So all that was needed was a small 'seed' magnetic field to set the whole thing going, at some time in the past.
As you correctly point out, there will be energy loss in this system due to resistive dissipation. But the system is constantly being supplied with heat to maintain convection. I suppose the term 'self-exciting dynamo' is potentially misleading. Perhaps 'self-organising' would be better.
So on to the models. Running computer models of this process could have one of two aims:
- proving the general valdity of this mechanism (i.e can this sort of process actually generate a stable field)?
- Trying to produce a field with behaviour that matches our observations of the Earth's.
Only the second category strictly requires we exactly model based on the parameters of the Earth's outer core. And you're right in that we don't seem to do that: A brief trawl reveals that these models are generally run with much smaller Rayleigh number (of the order of 10 times the critical value). It seems that modelling a system with the ultra-large Rayleigh numbers I derived above is a bit of a computing challenge - I'm not sure why but I'd suspect its something to do with turbulent flow becoming a big issue.
So it was a bit of a surprise when these models started producing fields which *did* look a lot like the Earth's, including the occurence of reversals. This is what is driving the current (ho ho) thinking about the primary importance of the Earth's rotation on the behaviour of the field.
Just in closing, bear in mind that we have *observed* a dominantly dipolar field which reverses every few hundred thousands years. Our models try to explain this behaviour. Invalidating the model does not invalidate the observations. And, let me emphasise: REVERSALS HAPPEN. I STUDY THEM. I have gone to a sedimentary section. I have drilled and oriented samples from different levels. I have measured and analysed the samples in the lab, and I have found polarity reversals through the section. So, I'm wondering whether to be insulted that you don't rate my competence, or amused that you think I'm going to all this effort as part of the Grand Plate Tectonic Conspiracy. They don't pay me that well, you know!
Please don't take that the wrong way - you're a very polite poster, and I appreciate that.
This message has been edited by gengar, 01-21-2005 11:18 AM
™ Version 4.2
