Federal agencies collect and analyze data to carry out their missions. A significant portion of these activities requires geospatial data collection in the field. Models for computer-assisted survey information collection are still largely based on the client-server paradigm with symbolic data representation. Little attention has been given to digital geospatial information resources, or emerging mobile computing environments. This paper discusses an infrastructure designs for delivering geospatial data users in a mobile field computing environment. Mobile field computing environments vary widely, and generally offer extremely limited computing resources, visual display, and bandwidth relative to the usual resources required for distributed geospatial data. Key to handling heterogeneity in the field is an infrastructure design that provides flexibility in the location of computing tasks and returns information in forms appropriate for the field computing environment. A view agent based infrastructure has been developed with several components. Wrappers are used for encapsulating not only the data sources, but the mobile field environment as well, localizing the details associated with heterogeneity in data sources and field environments. Within the boundaries of the wrappers, mediators and object-oriented views implemented as mobile agents work in a relatively homogeneous environment to generate query results. Mediators receive a request from the user application via the field wrapper, and generate a sequence of mobile view agents to search for, retrieve, and process data. The internal infrastructure environment is populated with computation servers to provide a location for processing, especially for combining data from multiple locations. Each computation server has a local object-oriented data warehouse equipped with a set of data warehouse tools for working with geospatial data. Since the prospect of query reuse is likely for a field worker, we store the final and intermediate results in the data warehouse, allowing the warehouse to act as an active cache. Even when field computing capacity is ample, the warehouse is used to process data so that network traffic can be minimized.