Developers of expert systems for meteorology must first spend a considerable effort building a meteorologically-oriented programming substrate. Meteorology expert systems deal with many kinds of data and with a variety of complex atmospheric entities and processes. As a result, the scope of effort required to develop the substrate can be quite large, and can hinder or prevent the development of many very promising expert systems applications. Our technical approach to these problems is to build a portable object-oriented software environment for meteorology expert systems and to implement an object base containing persistent objects representing all of the available synoptic climatological entities and data. At least two major problems and opportunities are addressed by this effort. One of these is greatly facilitating the development, integration, and delivery of expert systems for meteorology applications, by providing an appropriate programming and runtime environment for these problems. The second is providing a framework in which researchers can use artificial intelligence and related modern software technology, together with numerical and other software technology, to implement the considerable skill embodied in synoptic climatology. This will be particularly helpful in data-sparse and novel situations in military and commercial applications.
