Geostatistical Modelling of Precipitation Regimes in Colombia

This paper illustrates the geostatistical procedure followed to model the standard monthly precipitation in Colombia and to characterize the spatial variability of its parameters. Available data consisted of: forty years monthly precipitation series measured in 155 stations and standard monthly precipitation of 450 stations. Both information sets were spread on all the country. Geostatistical processing of the 155 monthly time series shown three time-variability components: two periodic with a period of six and twelve months and a nugget component. It will be shown how percentage of each component and ratio are varying from zone to zone, so that eight type of variogram beaviour have been selected to represent the global time-variability. Then a multivariate geostatistical analysis has been carried out considering the station sites as variables and months as samples. The cross-covariances highlighted a time-delay between the stations for both periodical components. On the base of a such time-analysis results it has been possible to make profitable use of multiannual data. In fact for each station, the 12-months multiannual diagram can be broken into two periodic components of six and twelve months period, each one characterized by an own time-delay (phase). 6-months phase is quite constant, while twelve component phase is varying from zone to zone within a range of six months. In this way by means of only two parameters: ratio between variances of periodic components and 12-months component phase it has been possible to characterize completely pluviometric regimes of Colombia. Besides these parameters shown a strong meso-scale regionalization and they seem to be linked to regional climatic phenomena, as those linked to ITCZ (Inter-Tropical Convergence Zone).