On the use of cross-correlation analysis in studies of patterns of rainfall variability

Cross-correlation structures based on daily rainfall data recorded during a period of 5 years at 67 stations throughout Tunisia are used to examine spatial rainfall variability. Basic properties of the correlation-distance function such as cumulating period, exclusion versus inclusion of dry periods, and data volume are analyzed. Analysis of the correlation patterns for each month shows that two periods (February-May, and September-November) are delimited by a period of great instability (December-January) and by the dry period (June-August). Three main rainfall regions can be distinguished, delimited and governed mainly by the topography and the coastal influence