# Some problems concerning the inversion of mantle convection caused by the consideration of conservation of energy and matter

Sir, I would like to study the possible inversion of the direction of mantle convection considering the conservation of energy. May I ask which equations I need to invoke or which instance model from the manual to calculate

Hi Xiyuan,

You can actually already calculate flow back in time with the current code. The only thing you have to do is set the gravitational acceleration (in the Gravity model section of the input parameter file) to the negative of what you want it to be. So for example, if you want it to be 9.81, set it to -9.81 and the flow will be reversed.

Best,
Jacky

• Reversing the gravity vector won’t affect the diffusion of heat. I’d recommend setting all thermal diffusivities to zero and thinking about what you want to do with the boundary layers if you want to try Jacky’s method.
• Even with careful control of boundary conditions, the method is probably only good for about 100 Myr or so.
• Even if ASPECT were able to run a true backwards-in-time simulation, it wouldn’t be a good idea if dissipative processes (thermal diffusion, shear heating etc.) were important. This is because running dissipative processes backwards will amplify numerical noise in temperature and potentially other fields.

See Conrad and Gurnis (2003; https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2001GC000299) for detailed discussions on these and related points.

Dear teacher, I want to use the convection model given in the manual to simulate the inversion of mantle convection caused by energy loss. It is a gradual and periodic process. I would like to ask how to set boundary conditions (I think it is mainly the temperature factor) if I want to realize my experimental idea

Dear Xiyuan,

ASPECT is designed to solve all kinds of problems involved with convection in Earth’s mantle so “the convection model given in the manual” doesn’t give us much of a feeling of your experimental idea. Which convection model? How do you want to adapt it? What is the goal?

In ASPECT, you can set constant temperature boundary conditions that implicitly control energy flux in and out of the system. That isn’t the only option, but it is a common one. The manual should be sufficient to explain to you how you can set that up.

In order to help you best, it would be best if you try to do the following:

• You should write down exactly what you want to study (the goal of your simulation) and how you intend to run the simulation: boundary conditions, initial conditions (temperature, composition, flow laws etc etc), governing physics (radioactivity? shear heating? etc etc).
• Then you should check in the manual if ASPECT can already do all those things.
• Then try adapting an existing prm to do what you want to do.

We are happy to help out during any of these steps - most users contact us for two reasons:

• If they are unclear about whether ASPECT can do all the things they want it to do after reading the relevant parts of the manual. We either direct them to the implementation (if it exists) or give them some pointers to help them implement a new plugin themselves.
• If they have problems with adapting an existing prm. In this case, we ask them to let us know what they have tried, and what their problem is.

Best wishes,
Bob

Dear teacher, when I was studying the basin structures inside the Rift system of Africa in the African plate, I found that the basin structures in the study area developed along with reverse extrusion at the end of each stretch, which aroused my great interest. I considered that such periodic stress inversion phenomenon was caused by mantle convection inversion. Therefore, I want to use ASPECT to conduct relevant research on the earth’s mantle. My guess is that there is a reversal of mantle convection, which is mainly caused by the non-conservation of energy and material. Dear teacher, when I was studying the basin structure in the Central African Rift system in the African plate, I found that the basin structure in the study area was accompanied by inversion extrusion at the end of each tensile stage, which aroused my great interest. I considered that such a periodic stress inversion phenomenon was caused by mantle convection inversion. Therefore, I want to use ASPECT to conduct research on Earth’s mantle. My guess is that there is a reversal of mantle convection, and the reason for this reversal is mainly due to the nonconservation of energy and matter. Therefore, I want to use the simple convection model in ASPECT to verify my conjecture by setting the physical property parameters of the internal material to be consistent with those of the mantle, setting the boundary temperature to be a function of temperature variation with time and energy, and setting the flow of mantle material to be in line with the objective viscosity coefficient. Since I am a beginner, I do not have a comprehensive understanding of the whole control physical parameters of the model, so I would like to ask you how to improve my model, or you can provide some suggestions, such as which existing.PRM is better to use? How do I set the boundary conditions? What should I pay attention to if I need to modify the existing.prm?
Best wishes,
Xi Yuan

Dear Xi Yuan,

Ok, we clearly did not understand your original question. It sounds like you are interested in explaining observations of transient compression in a continental rift setting. That is indeed an interesting observation. Minor reverse faults can develop in grabens as a consequence of fault propagation folding (and these can be modelled with ASPECT), and more significant localised compression can be explained by rotations at the edge of tectonic blocks (see Glerum et al., 2020; Victoria continental microplate dynamics controlled by the lithospheric strength distribution of the East African Rift | Nature Communications). I am unaware of studies discussing transient compression in rifts driven from below.

My guess is that there is a reversal of mantle convection, which is mainly caused by the non-conservation of energy and material

Conservation of mass, energy and momentum are fundamental laws in geodynamics (and indeed all of physics, neglecting mass-energy equivalence), so as stated, your guess is inconsistent with our current understanding of the universe. It is possible for energy and mass to enter or leave a non-isolated system; maybe that’s what you meant?

I want to use the simple convection model in ASPECT to verify my conjecture

I assume you mean test your hypothesis?

Since I am a beginner, I do not have a comprehensive understanding of the whole control physical parameters of the model, so I would like to ask you how to improve my model, or you can provide some suggestions, such as which existing.PRM is better to use? How do I set the boundary conditions? What should I pay attention to if I need to modify the existing .prm?

This is a science question, not a software question. From what you’ve written, I have no idea how your desired simulation would differ from any generic model of mantle convection. For us to help, we need to know details of what you want to put into the model. To give you an idea of the level of detail we need, this prm might be useful: aspect/continental_extension.prm at main · geodynamics/aspect · GitHub. Don’t worry too much about the global parameters, mesh refinement and solver parameters sections (those are software-specific, and we can help with those), but you should have a good conceptual idea about what you want to do with the other sections (what are your temperature and velocity boundary conditions? which material model do you want to use? How many different compositions? Do you want to use a free surface? what is the initial condition? etc etc). The manual will help you with this.

If you’re uncertain how to set up your model I would recommend that you discuss your idea with a colleague / collaborator who specialises in geodynamics, or to find one if you don’t have one. It would probably be most efficient to do this in person or over Zoom/Skype/Teams, as I think you will need to explain your observations and they may need to explain aspects of geodynamic modelling in some detail. Designing good geodynamic studies requires specific knowledge, experience and time, and deserves formal recognition (usually co-I status, authorship etc.).

We will be very happy to help with software / implementation questions once you know exactly what you want to do.

Best wishes,
Bob

Thank you very much, teacher! To understand!