A Semi-parametric Estimation for Max-mixture Spatial Process

We propose a semi-parametric estimation procedure in order to estimate the parameters of a max-mixture model as an alternative to composite likelihood estimation. This procedure uses the F-madogram. We propose to minimize the square difference between the theoretical F-madogram and an empirical one. We evaluate the performance of this estimator through a simulation study and we compare our method to composite likelihood estimation. We apply our estimation procedure to daily rainfall data from East Australia.