We present an analysis of the filtering capabilities of wavelet and fourier transforms as they apply to terrain model generation. Additionally, we highlight the benefits of wavelet filtering in automatically producing approximations that progress from user-specified coarse to fine resolutions in a terrain model. In our previous work, we used wavelet filtered data in conjunction with a hierarchical triangulation technique to create a set of increasing resolution models. We now describe an approach that no longer is built from right angle triangles but generates triangular irregular networks (TINS) with automated control of triangle properties such as "sliveriness". This method uses a tunable filtering scheme to retain the dominant terrain features at each level and maintains hierarchy. The hierarchical tree could then be used in a real-time application to create an eyepoint specific multiresolution terrain representation. This automated terrain model generation procedure is implemented with real elevation data and the results presented are the relationship between the number of triangles created and the accuracy of the triangular models along with 2D and 3D plots of the resulting triangulations.