Introduction

The variation in time of environmental quantities can be studied using the rich branch of statistics known as time series analysis. A discrete (as opposed to continuous) time series^9.1is a sequence of observed values $\{x_1,x_2,\ldots,x_n\}$ measured at discrete times $\{t_1,t_2,\ldots,t_n\}$. Climatological time series are most often sampled at regular intervals $t_k=k\tau$ where $\tau$ is the sampling period.

The main aims of time series analysis are to explore and extract signals (patterns) contained in time series, to make forecasts (i.e. future predictions in time), and to use this knowledge to optimally control processes.

The two main approaches used in time series analysis are time domain and spectral (frequency) domain. The time domain approach represents time series directly as functions of time, whereas the spectral domain approach represents time series as spectral expansions of either Fourier modes or wavelets.