Statistical analysis of monthly precipitation in Sweden using the Tweedie distribution

Abstract Statistical models for precipitation is of importance in several applications in climatology and agriculture. In this paper, we study a statistical distribution, the Tweedie distribution, to analyse monthly precipitation. Results are found in the literature for sites found in Australia. Here we investigate the amount of precipitation at several sites in Sweden, where various climate zones are found according to the Köppen–Geiger climate classification. The Tweedie distribution was fitted by maximum-likelihood techniques and the behaviour of the so-called index parameter was assessed on a monthly basis for the various sites. Moreover, a clustering procedure was applied to possibly cluster the sites according to the index parameter. We found similar features as for the Australian studies concerning magnitude of the estimated index parameter and influence of coastlines. The results indicate that the Tweedie distribution is a relevant model also for sites in Northern Europe.