Multisite, multivariate weather generation based on generalised linear models

Abstract This paper describes a methodology for constructing and simulating from models of daily weather time series at multiple locations, incorporating potential nonstationarities and suitable for use in those studies of climate impacts and adaptation where a detailed representation of local weather is required. The approach is based on generalised linear models (GLMs) and aims to allow for realistic representations of local weather structures including spatial, temporal and inter-variable dependencies. The theory is implemented in a software tool, Rglimclim , that runs in the R programming environment; and is illustrated using a case study involving generation of daily precipitation and temperature at 26 locations in northern Iberia.