“…Hidden Markov models have been used to model occurrence [ Hughes et al ., ] as well as intensity [ Ailliot et al ., ; Charles et al ., ]. Other methods include nearest‐neighbors resampling [ Apipattanavis et al ., ; Buishand and Brandsma , ; Rajagopalan and Lall , ], generalized chain‐dependent processes [ Zheng and Katz , ; Zheng et al ., ], power transformation to normality [ Sansó and Guenni , ; Yang et al ., ], artificial neural network methods [ Cannon , ], copula‐based approaches [ Bárdossy and Pegram , ], and multifractal rainfall models [ Gupta and Waymire , ; Lovejoy and Schertzer , ; Menabde et al ., ; Deidda , ; Veneziano and Langousis , ; Langousis and Veneziano , ; Veneziano et al ., ; Veneziano and Langousis , ]. Current approaches to this problem typically seek the assistance of latent multivariate normals, sometimes including a transformation, to generate the occurrence/intensity values over a spatial domain.…”