This work proposes a linear algorithm which relates precipitable water (PW) with altitude and distance from the coast, on a regional scale. For precipitable water, the ERA-Interim (ECMWF) database for the period 1989-2010 was used, while a digital elevation model was used for the geographical variables. The results obtained indicate statistical significance between the two variables (p< 0.01), with a confidence level of 99%. The model makes it possible to describe the mean monthly space-time variability of precipitable water, stratifying the study area into clearly differentiated zones.