Context. The spatial model of territorial system in natural emergency conditions dedicated to decision support tasks solving and represented as the model of complex dynamic system is described in the paper. The literature data analysis has shown that applying the established approaches to the class of complex dynamic systems under consideration doesn’t provide the required calculation speed andefficiency of decision support system. This determines the timeliness of the further searching of nontraditional models and methods of decision support, which provide the fulfillment of calculation speed and efficiency requirements applicable to such systems. Objective. The goal of research is decreasing the damage from the natural emergency by means of improving the quality and timeless of forecasting the territorial system dynamics in the natural emergency conditions Method. The concept of territorial system in natural emergency conditions is represented in the form of overlaying static and dynamic topological spaces induced by indiscernibility relation, each of which allows representing geographical and attributive information about nature conditions, value objects demanding protection against natural emergency, as well as about natural emergency dynamics. The model of natural emergency dynamics is based on the discretization of the space by the grid of square cells of equal area and represented in the form of fuzzy dynamic topological space in the set of cells. Results. The web-oriented decision support system is created on the base of developed concept and model. The experiments have been conducted, which have shown that the proposed natural emergency model can provide reasonable characteristics in terms of accuracy and speed providing that the space is discretized with the size of cell being from 8 m to 18 m. Conclusions. The concept of territorial system representation in the natural emergency conditions is first developed in the form of overlaying static and dynamic topological spaces. The formal model of the natural emergency dynamics in the form of moving the bounding region of fuzzy-rough set, which has allowed reducing computational complexity and provided adaptation to the conditions of incomplete and inaccurate information is first created. The software for realization of suggested concept and method is developed. The software has allowed performing the practical task of decision support in natural emergency conditions.