Extreme value theory for space-time processes with heavy-tailed distributions (Q2476290)

The authors study the asymptotic theory for point processes and extremes of filtered processes of the form \[ X_t(\mathbf s )=\sum_{i=0}^{\infty} \psi_i(\mathbf s)Z_{t-i}(\mathbf s), \, \mathbf s \in [0,1]^d, \] where \((Z_t)_{t \in \mathbb Z}\) is an iid sequence of regularly varying random fields on \([0,1]^d\) with values in the Skorokhod space \(\mathbb D([0,1]^d)\). The coefficients \(\psi_i\) are deterministic real valued fields. First, they apply the notion of regular variation on \(\mathbb D\) to max-stable fields and obtain a representation of a max-stable random field in terms of a homogeneous Poisson process and iid fields. Next, regular variation of linear combinations of random fields is studied and the obtained results are used for showing point process convergence of the scaled linear process \(X_t\) towards a compound Poisson process. Some examples and applications are given.