Understanding some weather phenomena behavior, especially temperature, is very important to many human activities. For this reason the aim of this work is to model monthly temperature in Valle del Cauca between 1971 and 2002 using Weighted Smoothing Splines. Two stratums (valley and mountain), in terms of temperature were considered during the process, since Valle del Cauca is a region located in different thermal floors, that affect its behavior. A voronoi diagram was used to determine the area of influence of each weather station located in the department and assign its modeling weight.