STOCHASTIC WEATHER TYPE SIMULATION FOR REGIONAL CLIMATE-CHANGE IMPACT ASSESSMENT

A stochastic model is developed for the synthesis of daily precipitation by weather type analysis. Daily rainfall at two sites in southern England is related to the Lamb weather types by using conditional probabilities. Time series of circulation patterns and hence rainfall are then generated using a Markov representation of matrices of transition probabilities between weather types. The model reproduces key aspects of the historic rainfall regime at daily, monthly, and annual temporal resolutions. The general validity of the methodology as a means of disaggregating general circulation model predictions into finer spatial scales for water resource studies is discussed. Several limitations are identified of which the subjectivity introduced by the weather type classification system and the nonstationarity of the precipitation characteristics are the most serious.