Climate risks: Temporal trends and heteroscedasticity

摘要

The idea is developed that the consideration of the climate change even in the form of the linear trend can provoke the dramatic nonlinear increase in the probability of adverse weather events. Considered is a possibility of using the concept of heteroscedasticity (the heterogeneity of the variance of the random error of the regression model) for studying climatic time series. Discussed are the cases of temporal and factor heteroscedasticity. Introduced is a concept of the trend-type climate risk. Considered are the issues related to the risk assessment in global climate models.