We discuss here the issues faced by earth scientists in analyzing and visualizing large datasets over a GRID-like setup from a client-server perspective. We approach this problem by using a remote, web-based visualization and data analysis framework, called WEB-IS, and by employing second-generation wavelets as a means for reducing the amount of data transferred and for extracting coherent features in complex geophysical flows and surface faulting patterns. As an example, we describe how onboard processors on satellites can function as a server for beaming down extracted information to the client computer on the ground, thus exemplifying WEB-IS as a viable middleware on a GRID network for geosciences.