Abstract: Multiple representations of spatial data,which means obtaining different detailed representation of geographic phenomena based on the same spatial database or different data versions,is one of the key technologies in GIS applications.It plays an important role in such applications as seamless data navigation,progressive web transfer and self-adaptable visualizations.This paper aims at the objective by small size of data volume to realize high granularity multi-representation,presenting an object oriented data model,namely the lifespan model over scale space,to divide the spatial data into different representation scenes over scale space.The scale transformation is represented as Tij: < fi,si1,si2,gij >,where fi is the transformation function,si1,si2 the scale range controlled by two key scale points si1 and si2 and gij the base representation status associated with the key scale point of si1 or si2.Then the representation at any scale point can be represented as: Rx=fi(gij,x),x∈si1,si2.The study summaries that there are four transformations,namely map generalization,LOD accumulation,morphing and equal transformation.