The general objective of this research was the development of knowledgebased computational techniques to support the interpretation of multitemporal remote sensing data, focusing on the investigation of the explicit representation of temporal knowledge and its integration to other types of knowledge; and also on the processing and acquisition of temporal knowledge. Two interrelated, specific objectives were pursued: (i) the development of a novel multitemporal classification method based on the concept of fuzzy Markov chain (FMC) that provides for the automatic estimation of its temporal related parameters and for the exploration of temporal knowledge in the classification process; and (ii) the design and implementation of an open-source, knowledge-based framework for multitemporal interpretation of remote sensing data. In order to validate the new multitemporal classification method, experiments were carried out aiming at the interpretation of a sequence of three LANDSAT images from the central region of Brazil, using both a stochastic and an analytical technique to estimate the class transition possibilities that compose the FMC model. While the monotemporal classifiers used in the experiments attained an average class accuracy of approximately 55%, the multitemporal scheme reached accuracies between 65% and 94%. Similar results in terms of overall accuracy were also observed. Furthermore, when compared to two alternative multitemporal classification approaches, the devised method consistently showed better results. In order to validate the proposed multitemporal framework, the FCM-based method was implemented using its temporal functionalities, and a number of experiments in which different variants of the FCM-based method were structured through the framework were successfully carried out.