A dense 3-d terrain model obtained using reconstruction methods from aerial images is represented in a probabilistic volumetric framework. The choice of probabilistic representation is to represent inherent ambiguity in reconstruction of surface from images. Such probabilistic representation handles the ambiguity very well but leads to expensive dense volumetric storage. The area coverage required for building 3-d models varies from half a square kilometer to thousands of kilometers. Extensive computational resources are required for rendering and building such large models. Existing methods for rendering 3-d models typically cater to mesh models only and also lack strategies to dynamically update the models due to memory intensive operations conventionally better suited for CPUs. This paper proposes a novel OpenCL implementation catering to both GPUs and CPUs for real-time visualizing as well as updating, and dynamic restructuring of dense volumetric models for 3-d terrain. The major contributions of this paper are hybrid representation of grid and octrees, bit-based representation of octrees, randomization of data to enable parallelization of an otherwise serial strategy for subdividing octrees, real-time rendering of dense volumetric data and segmentation algorithm for minimizing global memory access in GPUs. The ideas and implementations proposed could potentially be used in different applications.