Anisotropic Radial Basis Function Methods for Continental Size Ice Sheet Simulations

Gong Cheng,Victor Shcherbakov

DOI: https://doi.org/10.1016/j.jcp.2018.06.020

2017-11-28

Abstract:In this paper we develop and implement anisotropic radial basis function methods for simulating the dynamics of ice sheets and glaciers. We test the methods on two problems: the well-known benchmark ISMIP-HOM B that corresponds to a glacier size ice and a synthetic ice sheet whose geometry is inspired by the EISMINT benchmark that corresponds to a continental size ice sheet. We illustrate the advantages of the radial basis function methods over a standard finite element method. We also show how the use of anisotropic radial basis functions allows for accurate simulation of the velocities on a large ice sheet, which was not possible with standard isotropic radial basis function methods due to a large aspect ratio between the ice length and the ice thickness. Additionally, we implement a partition of unity method in order to improve the computational efficiency of the radial basis function methods.

Numerical Analysis,Fluid Dynamics

What problem does this paper attempt to address?

The problems that this paper attempts to solve are: to develop and implement the anisotropic radial basis function (RBF) method to simulate the dynamic processes of glaciers and continental - scale ice sheets. Specifically, the paper addresses the following key issues:

1. **Limitations of traditional numerical methods**:
- The traditional finite element method (FEM) requires frequent remeshing when dealing with complex geometries with moving boundaries, resulting in low computational efficiency.
- For continental - scale ice sheet simulations, due to the large aspect ratio between the length and thickness of the ice body, the standard isotropic RBF method is difficult to provide accurate velocity field simulations.

2. **Improving simulation accuracy and efficiency**:
- The paper proposes using the anisotropic RBF method to improve the simulation accuracy of the ice sheet velocity field, especially in continental - scale ice sheets where the standard isotropic RBF method fails due to the large aspect ratio.
- Introduces the Partition of Unity Method based on anisotropic patches to improve computational efficiency while maintaining high accuracy.

3. **Verifying the effectiveness of the method**:
- The paper verifies the advantages of the proposed method through two test cases: one is the ISMIP - HOM B benchmark test, corresponding to glacier - scale ice bodies; the other is a synthetic ice sheet inspired by the EISMINT benchmark, corresponding to continental - scale ice sheets.
- Through these tests, the advantages of the anisotropic RBF method over the standard finite element method in terms of time and accuracy are demonstrated.

### Summary of mathematical formulas

- **Full Stokes equation**:
\[
\nabla\cdot(\eta(\nabla\mathbf{v}+(\nabla\mathbf{v})^T))-\nabla p+\rho g = 0,
\]
\[
\nabla\cdot\mathbf{v}=0,
\]
where \(\mathbf{v}\) is the velocity vector, \(\eta\) is the viscosity, \(p\) is the pressure, \(\rho\) is the density of ice, and \(g\) is the acceleration due to gravity.

- **Glen's flow rate law**:
\[
D = A(T')f(\sigma)T_D,
\]
where \(D\) is the strain - rate tensor, \(A(T')\) is the rate factor, \(f(\sigma)=\sigma^{n - 1}\), and \(\sigma\) is the effective stress.

- **Anisotropic RBF distance definition**:
\[
\|x - y\|_a=\sqrt{a_1^2(x_1 - y_1)^2+\cdots+a_d^2(x_d - y_d)^2},
\]
where \(a_i\) is the aspect ratio of domain discretization.

Through these improvements, the paper aims to provide a more efficient and accurate numerical method for the dynamic simulation of continental - scale ice sheets.