Modern GIS(Geographic Information System) application programs and simulation systems have to handle large datasets for rendering. Currently three dimensional rendering hardware are facing a memory bus bandwidth bottleneck problem at the graphics pipeline. One general solution for this problem is to compress the static three dimensional geometry in a preprocessing phase. We present a new mesh compression/decompression algorithm for this application. Our compression algorithm breaks down a triangle mesh into a set of triangle strips and vertex chains. After decomposition, we encode vertex connectivity with entropy encoding. Our algorithm provides a 32% improvement in the compression ratio over existing "Generalized Triangle Mesh(GTM)" compression and supports parallel decompression. We also proposed a parallelogram prediction method for vertex coordinate compression.