Visualization and analysis of large data sets remains a significant challenge to the visualization community. Various data reduction techniques to deal with large data sets have been proposed over past years. As one of the more popular solutions, multi-resolution (MR) data model is used to reduce the size of original data sets into different resolution levels. However, one side effect of using such a technique is that the integrity of original data is adversely affected leading to the introduction of errors. In this paper, we address this issue by introducing two new uncertainty visualization techniques, which represent both MR approximations of original data and the uncertainty information associated with each approximation to enable users to preserve the integrity of the original data. Our techniques are applied to an example of volume data with regular grids from medical domain to demonstrate their effect and usability.