The net tropical forest carbon emission still remains uncertain, challenging the REDD+ monitoring system. This uncertainty is extensively related to the uncertainties in the estimations of the forest cover loss, the extent and intensity of forest degradation events, and the reforestation and regrowth rate. Complex spatiotemporal analysis techniques are needed to improve both spatially and temporally the estimation precision of the forest cover changes on tropical areas. Many remote sensing techniques have been developed to detect Land Cover Changes (LCC), including recent object based approaches exploiting the advantage of extracting changes in their spatial context. However, these previous works have been applied in a limited temporal scale, mostly bi-temporal or reduced multi-temporal image comparison (only few observation dates). In contrast, emerging LCC analysis approaches exploit the full temporal information of land cover dynamics provided by remote sensing dense time series, though they have been mostly applied in a pixel-based approach. Only little research has been done combining both approaches. This MSc. thesis explored the potential of spatiotemporal segmentation and time series analysis. First, LCC was estimated from Landsat time series by the Iterative Reweighted Multivariate Alteration Detection algorithm (iMAD) and spatiotemporal segments were delineated extracting LCC objects. Second, a time series analysis by the mean of the “Break For Additive Seasonal and Trend” (BFAST) algorithm over the LCC segments was performed for characterizing the temporal pattern of the segments. Finally, the information provided by both analysis was integrated for classifying the segments. This spatial and temporal information integration enhanced the capabilities of characterization of the spatial and temporal dynamics of tropical forest disturbances, contributing to increase the understanding of the immediate and long term implications of human activities over the tropical forest. Data abundance was found to be of major importance. Forest and non-forest related land cover changes within the study area have been detected producing satisfactory results. Especially the relations of segments and temporal profile’s need to be further analyzed by making use of multi-temporal change detection products and well framed multivariate analysis.