The process of surface modeling involves estimating values for a continuous single-valued variable, z(x,y), as the spatial attributes (coordinates) x and y vary across a study area. The estimation process used will depend upon several factors, including prior evidence of what general type of surface is being modeled and for what purpose the modeling is being carried out. For example, some surfaces are typically very smooth over large areas, or exhibit smooth wave-like variations (e.g. air pressure at a given altitude), whilst others are extremely jagged with sharp variations over small regions. For many datasets the samples themselves show substantial variation that it may be possible to model statistically, reflecting the uncertainty associated with the information available. The various approaches to modeling currently applied are described in the following subsections.