Land use and land cover evolution was contributing to the emergence of geographical information systems. Today, most applications benefit from sophisticated technologies for gathering and analysing data, which is of great value for urban and regional planning. This paper presents a typology and conceptual frameworks for the specification of several levels of paradigms for analysing land use evolution. It distinguishes various degrees of abstraction ranging from description to understanding of phenomena, with modelling, hypotheses validation, forecasting and explanation as intermediate levels. Conceptual diagrams are developed. It relates scientific approaches and levels of knowledge in order to better focus on research objectives, choose appropriate analytical procedures and assess the potential for generalizing results with due consideration of closely related domains of knowledge. An example of spatial diffusion model illustrates how this framework could enhance linkage with classical theoretical models. Its strengths and weaknesses are discussed, with linkage to the typology and appropriateness for urban and regional planning.