| Popis: |
A new inverse method for groundwater flow is presented which can optimize large scale inverse problems efficiently. All parameters, initial and boundary conditions and driving forces are allowed to vary in space, while boundary conditions and driving forces can vary in time as well. The objective function, which is minimized, is a weighted least squares function with prior terms. The method minimizes the objective function by directly solving the Euler-Lagrange equations in an iterative way by using a basis function expansion with representers. The method was applied successfully in a 2 dimensional transient problem where it had to optimize 9225 unknowns. |
