Layered Mantle Convection and Magma Production on Mars: Effects of Dense

Introduction: Geologically recent volcanism on Mars [1-4] can be well explained with mantle convection and plumes [5-7]. Most of these studies have assumed single layer convection in Martian mantle. However, a layered Martian mantle has been suggested in the model of magma ocean and mantle overturn [8]. The possible effects of such stratification on the early volcanism, magnetic dynamo, and thermal history of Mars were explored before [9-11]. We did preliminary exploration of the effect of a deep dense layer on mantle convection and magma production of Mars by fixing the core-mantle boundary temperature [12]. Here, we explore the effects of dense layer properties on mantle convection and magma production by fixing the layer interface temperature. Our focus is on the viscosity and radioactivity contrast between the two layers. Computational Approach: We use the spherical axi-symmetric version of CitCOM [13, 14] to simulate mantle convection on Mars. A particle ratio method [15] is used in the code to model thermochemical convection. The non-dimensional model domain (θ = 0-π/4, R = 1-2) is meshed with a 128×160 grid that has a typical mesh resolution of 13 km in the upper layer and 4 km in the lower layer. Each element is assigned with 16 particles to trace chemical composition. The upper and lower boundaries are constant in temperature. The side boundaries are thermally insulated. All four boundaries are free-slip. The mantle viscosity is temperature dependent, obeying the Arrhenius viscosity law [16]. The activation energy is 160 kJ/mole [5]. Half of the radioactive elements [17] are differentiated into the crust. The buoyancy number, B, is the ratio between the chemical density difference between the layers and the thermal buoyancy [18]. We assume B=1, corresponding to a density difference of about 200 kg m between the two layers. The actual density difference could be larger [8], but further increase in B does not modify the fluid dynamics [18]. The dimensional model parameters are the same as in [5]. An initial thermal perturbation is applied to generate a plume at the center of the model. The model has been run to reach a statistically steady-state for each model case. We use Katz’s melt fraction calculation formula, in which melt fraction is a function of solidus, liquidus and the mantle temperature [19]. The Katz’s dry solidus model coincides well with the experimental results of martian analog composition [20]. Magma production is calculated using the formalism of Kiefer [6]. In model parameters input and model results analysis, the core-mantle boundary temperature is set to dimensionalize the temperature fields. In our previous studies [12], we keep the core-mantle boundary temperature fixed among model cases. Since the upper layer behaves similar to a single layer convection when B>1 [21], we choose to fix the layer interface temperature here. This is accomplished through iteration. First, we give a try core-mantle boundary temperature. Second, we obtain a horizontally averaged interface temperature through model simulation. Third, we update the core-mantle boundary temperature to scale the layer interface temperature to be 1600 °C. These steps are iterated for 2-3 times to let the interface temperature be within 1600 °C ± 20°C among model cases. Results: The thickness of the dense layer is not well constrained and model dependent [22, 23], so we consider a range of bottom layer thickness (40-180 km). We also consider a range of convective vigor (thermal Rayleigh number of the upper layer defined using the volume average viscosity varies from 2.3×10 to 1.3×10), and viscosity and radioactivity contrast between the upper and lower layers. An example thermal field is shown in Figure 1. Model parameters for this example case are: thickness of the chemical layer is 180 km; the volume-averaged thermal Ra of the upper layer is 6.5×10; activation energy is 160 kJ/mole; half of the radioactivity is in the crust; and mantle viscosity and radioactivity are the same between the two layers.