INTRODUCTION

Some of the problems that most interest the scientific community studying the atmosphere involve the description and analysis of climate variability. Rainfall is a major factor in agriculture and in recent years interest has increased in learning about precipitation variability and predictability for periods of months to years. We are concerned particularly about the threat of desertification throughout the Iberian Peninsula (hereafter IP). The precipitation time series appear to fluctuate randomly; however, a closer examination implies some regularity, and based on this behaviour we try to detect and estimate the interannual deterministic components or ‘signals’ within the observations in order to characterise some of the precipitation variations.

Krepper et al. (1989) have analysed monthly and seasonal rainfall in Argentina and determined the spatial and temporal variability using empirical orthogonal functions analysis; they also obtained the variance contribution of the spectral peaks. Recently, Rowell et al. (1995) used spectral analysis on the rainfall data in the north of Africa and obtained peaks at 3–۸ years in Sudan and at 2–۳ years in Guinea. Climatic fluctuations with several quasi-periods were found by Hogg (1995) in a study of time series of Canadian extreme rainfalls. Other authors have reported oscillations for different climate series: Vautard et al. (1992) applied singular spectral analysis (SSA) to global surface air temperature and found oscillations at 4.6, 5.2, 10, 7–۸, ۱۵ and 26 years; Labitzke and van Loon (1995) obtained the ten to twelve (TTO) year oscillation in the tropospheric temperature.