Abstract: This paper aims to present a methodology to link analysis process to the development of land cover change model. We developed a method to partitioning the study area, located at Central-North region of Rondônia, in land units, characterized by history of occupation using a TM/Landsat image series, from 1985 to 2000 and a geographic database to store spatial and non-spatial information. Land Unities, called UOP, express different land cover patterns linked to different deforestation process in time and space. The UOP mapping allowed us to analyze the study area throughout deforestation indicators and to capture the diversity of occupation processes and some factors that affects deforestation distribution. This work allowed us to define some computational modeling requirement to support Rondônia space-time complexity as, multiple resolutions, multiple times events, multiple actors and behaviors.